Hugging Face's logo

Datasets:
WINGNUS
/
ACL-OCL

Tasks:
Token Classification
Languages:
English
Size:
10K<n<100K
Tags:
research papers
acl
License:
Dataset card Files Community
ACL-OCL / Base_JSON /prefixI /json /isa /2020.isa-1.5.json
Benjamin Aw
Add updated pkl file v3
6fa4bc9
raw
history blame contribute delete
No virus
166 kB
{
"paper_id": "2020",
"header": {
"generated_with": "S2ORC 1.0.0",
"date_generated": "2023-01-19T07:12:15.567373Z"
},
"title": "Annotation-based Semantics",
"authors": [
{
"first": "Kiyong",
"middle": [],
"last": "Lee",
"suffix": "",
"affiliation": {
"laboratory": "",
"institution": "Korea University Seoul",
"location": {
"country": "Korea"
}
},
"email": "ikiyong@gmail.com"
}
],
"year": "",
"venue": null,
"identifiers": {},
"abstract": "This paper proposes a semantics ABS for the model-theoretic interpretation of annotation structures. It provides a language ABSr that represents semantic forms in a (possibly \u03bb-free) type-theoretic first-order logic. For semantic compositionality, the representation language introduces two operators \u2295 and with some subtypes for the conjunctive or distributive composition of semantic forms. ABS also introduces a small set of logical predicates to represent semantic forms in a simplified format. The use of ABSr is illustrated with some annotation structures that conform to ISO 24617 standards on semantic annotation such as ISO-TimeML and ISO-Space.",
"pdf_parse": {
"paper_id": "2020",
"_pdf_hash": "",
"abstract": [
{
"text": "This paper proposes a semantics ABS for the model-theoretic interpretation of annotation structures. It provides a language ABSr that represents semantic forms in a (possibly \u03bb-free) type-theoretic first-order logic. For semantic compositionality, the representation language introduces two operators \u2295 and with some subtypes for the conjunctive or distributive composition of semantic forms. ABS also introduces a small set of logical predicates to represent semantic forms in a simplified format. The use of ABSr is illustrated with some annotation structures that conform to ISO 24617 standards on semantic annotation such as ISO-TimeML and ISO-Space.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Abstract",
"sec_num": null
}
],
"body_text": [
{
"text": "This paper has two aims: [i] to formulate a semantics, called Annotation-based Semantics (ABS), for the modeltheoretic interpretation of annotation structures and [ii] to recommend it as a semantics for ISO 24617 standards on semantic annotation frameworks such as ISO-TimeML (ISO, 2020) or ISO-Space (ISO, 2020) . As a semantics for these annotation frameworks, ABS has two roles. One role is to validate the abstract syntax that formally defines each annotation framework in set theoretic terms (Bunt, 2010) . The other is to interpret the annotation structures that are generated by, or conform to, a relevant annotation framework (see (Lee, 2018) and (Pustejovsky et al., 2019) ). ABS is a structurally simple semantics, consisting of [i] a representation language ABSr and [ii] a finite set of logical predicates that are used in ABSr, but are defined as part of a model structure like meaning postulates or word meanings as introduced by Carnap (1947 Carnap ( 1956 and Montague (1974) , as shown in Figure 1 , and further developed by Dowty (1979) and Pustejovsky (1995) . The rest of the paper develops as follows: Section 2 provides some motivations for ABS . Section 3 describes the basic design of ABS. Section 4 defines the type-theoretic first-order predicate logic-based representation language ABSr . Section 5 breifly outlines some characteristics of an interpretation model structure for ABS . Section 6 shows how the composition rules of ABSr apply to the annotation structures that conform to some of the ISO 24617 standards on semantic annotation. Section 7 introduces some related works and discuses the convertibility of semantic forms of ABS to DRSs or \u03bb-formulas. Section 8 makes some concluding remarks.",
"cite_spans": [
{
"start": 301,
"end": 312,
"text": "(ISO, 2020)",
"ref_id": null
},
{
"start": 497,
"end": 509,
"text": "(Bunt, 2010)",
"ref_id": "BIBREF5"
},
{
"start": 639,
"end": 650,
"text": "(Lee, 2018)",
"ref_id": "BIBREF36"
},
{
"start": 655,
"end": 681,
"text": "(Pustejovsky et al., 2019)",
"ref_id": "BIBREF47"
},
{
"start": 944,
"end": 956,
"text": "Carnap (1947",
"ref_id": "BIBREF10"
},
{
"start": 957,
"end": 970,
"text": "Carnap ( 1956",
"ref_id": "BIBREF10"
},
{
"start": 975,
"end": 990,
"text": "Montague (1974)",
"ref_id": "BIBREF41"
},
{
"start": 1041,
"end": 1053,
"text": "Dowty (1979)",
"ref_id": "BIBREF17"
},
{
"start": 1058,
"end": 1076,
"text": "Pustejovsky (1995)",
"ref_id": "BIBREF48"
}
],
"ref_spans": [
{
"start": 1005,
"end": 1013,
"text": "Figure 1",
"ref_id": "FIGREF0"
}
],
"eq_spans": [],
"section": "Introduction",
"sec_num": "1."
},
{
"text": "The main motivation of ABS is to lighten the burden of automatically generating intermediary interpretations, called semantic forms or logical forms, of semantic annotation structures for both human and machine learning or understanding. For this purpose, ABS and its representation language ABSr introduce two minor operational modifications into the two well-established and model-theoretically interpretable representation languages, the type-theoretic \u03bbcalculus, used for Montague Semantics (MS) (Montague, 1974) , and Kamp and Reyle (1993) 's Discourse Representation Theory (DRT). The representation language ABSr of ABS is designed to to be free from \u03bb-operations, especially involving higher-order variables, by replacing the operation of substitution through the \u03bb-conversion with an equation solving approach (see Lee (1983) ), or to convert its semantic forms into visually more readable Discourse Representation Structures (DRSs) preferably without introducing embedded or stacked structures into them. From a theoretical point of view, neither ABS nor ABSr is totally different from Bunt (2020b) or his earlier efforts to develop an annotation-based semantics with the interpretation function I to convert or annotation structures, defined in abstract (set-theoretic) terms, to DRSs based on Kamp and Reyle (1993) 's Discourse Representation Theory (DRT). From a practical point of view, ABS is characterized by dividing the task of interpreting annotation structures between the representation of simpler or abbreviated semantic forms and their interpretations enriched with lexical meaning in the form of meaning postulates that constrain the set of possible interpretation model structures.",
"cite_spans": [
{
"start": 500,
"end": 516,
"text": "(Montague, 1974)",
"ref_id": "BIBREF41"
},
{
"start": 523,
"end": 544,
"text": "Kamp and Reyle (1993)",
"ref_id": "BIBREF28"
},
{
"start": 824,
"end": 834,
"text": "Lee (1983)",
"ref_id": "BIBREF32"
},
{
"start": 1096,
"end": 1108,
"text": "Bunt (2020b)",
"ref_id": "BIBREF8"
},
{
"start": 1305,
"end": 1326,
"text": "Kamp and Reyle (1993)",
"ref_id": "BIBREF28"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Motivation for ABS",
"sec_num": "2."
},
{
"text": "Based on a type-theoretic first-order predicate logic (FOL), ABSr is augmented with [i] a small set of operators and [ii] a set of logical predicates. As is developed in Section 3, for any a that refers to the abstract specification of an annotation structure or its substructures, either an entity or a link structure, preferably through its ID, the operator \u03c3 maps a to a semantic form \u03c3(a), represented in a first-order logic, while the two non-Boolean operators \u2295 and , with their finer-grained subtypes of merging, each relate \u03c3(a) to another semantic form, constrained by their semantic type. Without much depending on the particular syntactic analysis of each input, these operators combine, in a compositional manner, the pieces of information conveyed by each annotation structure or its substructures into a model-theoretically interpretable logical form, called semantic form, in FOL. Besides the Boolean connectives in FOL, these non-Boolean operators are needed to combine semantic forms that are not of type t (sentential type) as bridges that connect annotation structures to logical forms: for instance, to combine \u03c3(F ido) of individual entity type e with \u03c3([runs(e) \u2227 agent(e, x)]) of type e \u2192 (v \u2192 t) without using \u03bb-operations in an overt way.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Motivation for ABS",
"sec_num": "2."
},
{
"text": "As is elaborated in Section 3, ABS also introduces a small set of logical predicates into its representation language ABSr and treats them as meaning postulates that constrain a model structure (see Montague (1974) and Dowty (1979) ). There are at least two reasons for the introduction of a small set of logical predicates. One reason is representational simplicity: it can, for instance, represent the semantic form of the past tense of a verb in English as past(e), where past is a predicate to be defined as part of an interpretation model and e is a variable of type v for eventualities, instead of introducing one of its definitions, which is the most common one [\u03c4 (e) \u2286 t \u2227 t \u227a n] into the semantic form. This semantic form requires the introduction of a real-time function \u03c4 from events to times, two temporal relations, those of inclusion \u2286 and precedence \u227a, and the notion of the present time n. Furthermore, it is a straightforward process to translate an entity structure like event(e1, ran, pred:run, tense:past) into a semantic form [run(e 1 ) \u2227 past(e 1 )]. Another reason is representational flexibility. ABS can first choose an appropriate definition or meaning from a set of possible definitions given in a model structure and then decide on an appropriate model M and an assignment g that together satisfy a semantic form like [run(e 1 ) \u2227 past(e 1 )]. This would be the case particularly if the past tense needs to be interpreted in a deitic or situational sense, as discussed by Partee (1973) and Quirk et al. (1985) . ABS upholds the principle of minimalism and partiality in its representation. It does not aim nor claim to treat the total interpretation of natural language expressions. Being based on a restricted set of markables in data, either textual or audio-visual, and their annotation, the task of annotation and that of its semantics such as ABS are bound to be restrictive: the semantics can be either simple or complex depending on what needs to be annotated. The granularity or complexity of semantic forms only depends on that of the input annotation structures and their substructures. The granularity of perceiving and constructing these structures, especially involving spatio-temporal information, is controlled or modulated through common-sense logic by the need of their applications, as is discussed by Miller and Shanahan (1999) and Gordon and Hobbs (2017) ).",
"cite_spans": [
{
"start": 199,
"end": 214,
"text": "Montague (1974)",
"ref_id": "BIBREF41"
},
{
"start": 219,
"end": 231,
"text": "Dowty (1979)",
"ref_id": "BIBREF17"
},
{
"start": 1501,
"end": 1514,
"text": "Partee (1973)",
"ref_id": "BIBREF43"
},
{
"start": 1519,
"end": 1538,
"text": "Quirk et al. (1985)",
"ref_id": "BIBREF50"
},
{
"start": 2349,
"end": 2375,
"text": "Miller and Shanahan (1999)",
"ref_id": "BIBREF39"
},
{
"start": 2380,
"end": 2403,
"text": "Gordon and Hobbs (2017)",
"ref_id": "BIBREF18"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Motivation for ABS",
"sec_num": "2."
},
{
"text": "The main characteristics of ABS are the following. First, ABS is based on annotation work, making use of the semantic annotation of coumminicative linguistic data for their semantic interpretation. Without relying on a pre-defined syntax, it manipulates minimally what is encoded in annotation structures and their substructures and converts these structures to logical forms that can be interpreted modeltheoretically. ABS is, for instance, designed to support spatio-temporal annotation by validating the abstract syntax of ISO-Space (ISO, 2020), as proposed and outlined by Lee (2016) , Lee (2018) , and Lee et al. (2018) as well as ISO-TimeML (ISO, 2012) and Pustejovsky et al. (2010) . Second, ABS only provides partial information on a restricted set of markables for semantic annotation. Unlike ordinary semantics like Montague Semantics (Montague, 1974) or even Minimal Recursion Semantics (Copestake et al., 2005) , ABS is not a general semantics that attempts to treat all aspects of language in an abstract way. Third, ABS leaves much of the information unspecified. It allows, for instance, some variables to occur unbound in well-formed semantic forms, as in the interval temporal logic of Pratt-Hartmann (2007) , while their scoping is left unspecified till the last stage of composing semantic forms or being interpreted (model-theoretically), unless the scope is specified as part of annotation. As a result, the semantic type of semantic forms is partially non-deterministic: it can be interpreted either as of type t potentially denoting a proposition or a truth-value or of a functional type \u03b1 \u2192 t, where \u03b1 is a well-defined type, denoting a set of individual objects or of higher-order objects. Fourth, ABS introduces a small set of predicates such as past and perfective for the specification of tense and aspect. It can also introduce the predicates holds and occurs, as defined in Allen (1984) and others, for the event-type dependent temporal anchoring into semantic forms. All these predicates that occur in semantic forms are defined as part of an interpretation model or leaving room for various uses of grammatical concepts or their contextually dependent interpretations. Being based on annotations, ABS must deal with complex issues in semantic annotation such as quantification, for instance, as raised by Bunt (2020a) and Bunt (2020b) or the meaning of determiners that include numerals as in \"two donkeys\" in language in general. It may also have to deal with the structure and substructures of eventualities, especially dealing with dynamic motions, as discussed in Mani and Pustejovsky (2012) . The complexity or granularity of ABS thus totally depends on that of annotation structures or the type of annotations. In addition, ABS upholds a couple of well-established basic assumptions as its theoretical basis:",
"cite_spans": [
{
"start": 577,
"end": 587,
"text": "Lee (2016)",
"ref_id": "BIBREF35"
},
{
"start": 590,
"end": 600,
"text": "Lee (2018)",
"ref_id": "BIBREF36"
},
{
"start": 607,
"end": 624,
"text": "Lee et al. (2018)",
"ref_id": "BIBREF31"
},
{
"start": 663,
"end": 688,
"text": "Pustejovsky et al. (2010)",
"ref_id": "BIBREF46"
},
{
"start": 845,
"end": 861,
"text": "(Montague, 1974)",
"ref_id": "BIBREF41"
},
{
"start": 898,
"end": 922,
"text": "(Copestake et al., 2005)",
"ref_id": "BIBREF11"
},
{
"start": 1203,
"end": 1224,
"text": "Pratt-Hartmann (2007)",
"ref_id": "BIBREF44"
},
{
"start": 1904,
"end": 1916,
"text": "Allen (1984)",
"ref_id": "BIBREF1"
},
{
"start": 2337,
"end": 2349,
"text": "Bunt (2020a)",
"ref_id": "BIBREF7"
},
{
"start": 2354,
"end": 2366,
"text": "Bunt (2020b)",
"ref_id": "BIBREF8"
},
{
"start": 2600,
"end": 2627,
"text": "Mani and Pustejovsky (2012)",
"ref_id": "BIBREF38"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Basic Assumptions",
"sec_num": "3.1."
},
{
"text": "1. Semantics is constrained by a type theory (Montague semantics: Montague (1974) and Dowty et al. (1981)) 2. Events are viewed as individuals (Neo-Davidsonian semantics: Davidson (1979) , , Parsons (1990) , Pustejovsky (1995)) 3. Variables are linked to discourse referents (Discourse representation theory: Kamp and Reyle (1993)) 3.2. Metamodel Figure 1 shows the general design of ABS , which consists of (1) a representation language ABS and (2) an interpretation model M with logical predicates defined. ABS is an annotation-based semantics, meaning that its representation language ABSr translates each a of the abstract specification of entity or link structures that constitute annotation structures to a well-defined semantic form \u03c3(a). ABS then interprets each semantic form \u03c3(a) with respect to a model M , a list D of definitions of logical predicates, and an assignment g of values to variables, [[\u03c3(a) ]] M,D,g . Each \u03c3(a) in ABSr is an expression of first-order logic, but each of the logical predicates that my occur in \u03c3(a) may be defined in terms of higher-order logic as part of the model structure. ",
"cite_spans": [
{
"start": 45,
"end": 81,
"text": "(Montague semantics: Montague (1974)",
"ref_id": null
},
{
"start": 86,
"end": 106,
"text": "Dowty et al. (1981))",
"ref_id": null
},
{
"start": 171,
"end": 186,
"text": "Davidson (1979)",
"ref_id": "BIBREF12"
},
{
"start": 191,
"end": 205,
"text": "Parsons (1990)",
"ref_id": "BIBREF42"
},
{
"start": 208,
"end": 227,
"text": "Pustejovsky (1995))",
"ref_id": "BIBREF48"
},
{
"start": 309,
"end": 331,
"text": "Kamp and Reyle (1993))",
"ref_id": "BIBREF28"
},
{
"start": 909,
"end": 915,
"text": "[[\u03c3(a)",
"ref_id": null
}
],
"ref_spans": [
{
"start": 347,
"end": 355,
"text": "Figure 1",
"ref_id": "FIGREF0"
}
],
"eq_spans": [],
"section": "Basic Assumptions",
"sec_num": "3.1."
},
{
"text": "ABS adopts the system of semantic types which Kracht (2002) and Pustejovsky et al. (2019) have developed. They extend the list of basic types from Montague (1974) 's basic set of types {e, t} to an enlarged list, as specified in (1).",
"cite_spans": [
{
"start": 46,
"end": 59,
"text": "Kracht (2002)",
"ref_id": "BIBREF30"
},
{
"start": 64,
"end": 89,
"text": "Pustejovsky et al. (2019)",
"ref_id": "BIBREF47"
},
{
"start": 147,
"end": 162,
"text": "Montague (1974)",
"ref_id": "BIBREF41"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Basic and Functional Types",
"sec_num": "4.1."
},
{
"text": "(1) Extended List of Types:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Basic and Functional Types",
"sec_num": "4.1."
},
{
"text": "[i] Basic Types: a. t, the type of truth-values b. e, the type of individual entities c. v, the type of eventualities d. i, the type of time points e. p, the type of spatial points f. m, the type of measures g. int, the type of intervals h. vec, the type of vectors 1 [ii] Functional Types: h. If \u03b1 and \u03b2 are any types, then \u03b1 \u2192 \u03b2 is a type.",
"cite_spans": [
{
"start": 266,
"end": 267,
"text": "1",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Basic and Functional Types",
"sec_num": "4.1."
},
{
"text": "Type constructors such as \u2192 are introduced to define functional types: e.g., e \u2192 t, v \u2192 t, i \u2192 t, p \u2192 t or e \u2192 (e \u2192 t). Eventuality descriptions such as run or love are of type v \u2192 t, which is abbreviated to E (see Pustejovsky (1995) ), while the same symbol E is also used as as a symbol for a variable ranging over a set of eventualities or instances of an eventuality. The functional type p \u2192 t, denoting a set of spatial points, is often represented by a type r of regions 2 I may call these functional types E and r pseudo-basic types, for they are seldom analyzed as functional types. As introduced by Pustejovsky et al. (2019), path types are defined on the basis of the type of intervals int, which is defined [0, 1] \u2282 R, where R is a set of reals. A path \u03c0 will be that function int \u2192 p, which indexes locations on the path to values from the interval [0,1] (see Pustejovsky et al. (2019) ). A vector path \u03c0 v can also be defined as int \u2192 vec. An event path \u03c0 v will be defined as v \u2192 \u03c0 v as the function from eventualities to the vector path. Kracht (2002) and Pustejovsky et al. (2019) also introduce the group operator \u2022 to form group types, for example, p \u2022 for the group of spatial points. Link (1998) introduces two symbols * and and prefixes them to a predicate P to generate the group predicate * P and the plural predicate P , both based on the predicate P . Corresponding to each of the IDs of annotation structures or its substructures, entity or link structures, and of each of the types as defined in (1), there is a list of variables. Some of them are listed below: Table 1 : IDs, variables, and types",
"cite_spans": [
{
"start": 215,
"end": 233,
"text": "Pustejovsky (1995)",
"ref_id": "BIBREF48"
},
{
"start": 872,
"end": 897,
"text": "Pustejovsky et al. (2019)",
"ref_id": "BIBREF47"
},
{
"start": 1053,
"end": 1066,
"text": "Kracht (2002)",
"ref_id": "BIBREF30"
},
{
"start": 1071,
"end": 1096,
"text": "Pustejovsky et al. (2019)",
"ref_id": "BIBREF47"
},
{
"start": 1204,
"end": 1215,
"text": "Link (1998)",
"ref_id": "BIBREF37"
}
],
"ref_spans": [
{
"start": 1589,
"end": 1596,
"text": "Table 1",
"ref_id": null
}
],
"eq_spans": [],
"section": "Basic and Functional Types",
"sec_num": "4.1."
},
{
"text": "Categories 3 Ids Types Variables annotation a 1,... t a 1 , ... entity x 1,... e x, x 1 , ... v s, e, e 1 , ... event e 1,... E, e \u2192 t E, ... timex3 t 1,... I, i \u2192 t t, t 1 , ... place pl 1,... r, p \u2192 t l, l 0 , ... path p 1,... \u03c0 v , int \u2192 p p, p 1 , ... event-path ep 1,... \u03c0 v \u2192 \u03c0 v measure me 1,... m m, m 1 , ... link l 1,... t",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Basic and Functional Types",
"sec_num": "4.1."
},
{
"text": "The list of variables is just a conventionally used list. To be precise, for each entity structure E that confirms to a recognized annotation scheme such as ISO-TimeML or ISO-Space, a variable is defined as a pair <var:\u03c4 >, where var is a variable and \u03c4 is a type. Conventionally, any lowercase Latin characters such as x, y, etc. or e and s are used as variable for any one of the basic types provided that its type is specified: for example, x:<var, p \u2192 t> to use x as a variable ranging over regions of type r, or p \u2192 t. Uppercase Latin characters or special characters like E are used for functional types: E is a variable for eventuality descriptions such as what is denoted by a verb like \"run\". Note that run(e) is of type t, while the eventuality description run is of type v \u2192 t and its argument e is a variable of eventuality type v. 4 ",
"cite_spans": [
{
"start": 844,
"end": 845,
"text": "4",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Basic and Functional Types",
"sec_num": "4.1."
},
{
"text": "The part of ABSr that introduces the merge operators and their use is defined by Syntax absR . This syntax specifies what constitutes ABSr and how its constituents are formed. Some preliminary remarks are made before specifying the syntax of ABSr .",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Syntax Proper",
"sec_num": "4.2."
},
{
"text": "Just like any language, the representation language ABSr is a language that consists of a non-empty set of strings of character symbols. Each of such character strings in",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Preliminary Remarks",
"sec_num": "4.2.1."
},
{
"text": "ABSr is called a semantic form because it serves as an intermediary form for the model-theoretic interpretation of annotation structures. Further to clarify what ABSr is, I make some technical remarks.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Preliminary Remarks",
"sec_num": "4.2.1."
},
{
"text": "Remark 1: Mapping \u03c3 For any a that refers to the abstract specification of each of the entity or link structures which together constitute an annotation structure, independent of how these structures are represented, \u03c3 maps a to a semantic form in ABSr . \u03c3(a) is read as \"the semantic form of a\" in ABSr and is a well-formed form (wff) of ABSr . \u03c3(a) is considered independent of the format that represents it, but has to check the abstract syntax that validates the abstract specification a. Hence, a must be the same as the interpretation function I that is introduced in Bunt (2020b) and Bunt (2020a) . Remark 3: Typing ABSr is a type-based language. Hence, every well-formed (semantic) form A and any c of its constituents such as variables in ABSr is assigned a type. The type \u03c4 of A or c is represented as a pair: e.g., <A:\u03c4 >, <c:\u03c4 >, <var:\u03c4 >, or as a subscript to A or one of its constituents:",
"cite_spans": [
{
"start": 591,
"end": 603,
"text": "Bunt (2020a)",
"ref_id": "BIBREF7"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Preliminary Remarks",
"sec_num": "4.2.1."
},
{
"text": "A \u03c4 , c \u03c4 or x e .",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Preliminary Remarks",
"sec_num": "4.2.1."
},
{
"text": "Like the syntax of an ordinary language, Syntax absR consists of a vocabulary and a set of formation rules, as specified in (2).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Formulation of Syntactic Rules",
"sec_num": "4.2.2."
},
{
"text": "(2) Syntax absR = <V ,R> such that a. V is a vocabulary that includes binary merge operators {\u2295, } over the set of semantic forms in ABSr and their subtypes, and b. R is a set of composition rules for merging, as formulated in (7).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Formulation of Syntactic Rules",
"sec_num": "4.2.2."
},
{
"text": "There are two sorts of well-formed semantic forms (swff) in ABSr: basic and composed, each defined by a rule in R, a list of rules, in (4.2.3) and (7).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Formulation of Syntactic Rules",
"sec_num": "4.2.2."
},
{
"text": "Atomic semantic forms are defined by Rule A.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Atomic Semantic Forms",
"sec_num": "4.2.3."
},
{
"text": "(3) Rule A for Atomic semantic forms:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Atomic Semantic Forms",
"sec_num": "4.2.3."
},
{
"text": "For any abstract specification aEc of an entity structure E of category c, 5 and a type \u03c4 associated with cat, \u03c3(a Ec ) \u03c4 is a well-formed form of type \u03c4 in ABSr .",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Atomic Semantic Forms",
"sec_num": "4.2.3."
},
{
"text": "Remark 4: a Ec in \u03c3(a Ec ) \u03c4 is replaced by the ID of Ec.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Atomic Semantic Forms",
"sec_num": "4.2.3."
},
{
"text": "Following DRT (Kamp and Reyle, 1993) , the new occurrences of variables in a semantic form are registered.",
"cite_spans": [
{
"start": 14,
"end": 36,
"text": "(Kamp and Reyle, 1993)",
"ref_id": "BIBREF28"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Atomic Semantic Forms",
"sec_num": "4.2.3."
},
{
"text": "(4) Rule A.1 for Variable Registry: Any variable that is newly introduced to \u03c3(a Ec ) is listed in the preamble: i.e., \u03a3 var:type \u03c3(a Ec ). Note: These variables may not be registered if they can be recognized contextually.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Atomic Semantic Forms",
"sec_num": "4.2.3."
},
{
"text": "The variables in the preamble \u03a3 var:type are treated as discourse referents, to which each occurrence of the variables in \u03c3(a Ec ) is bound.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Atomic Semantic Forms",
"sec_num": "4.2.3."
},
{
"text": "Consider an example, annotated as in (5):",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Atomic Semantic Forms",
"sec_num": "4.2.3."
},
{
"text": "(5) a. Fido ran w2 away w3 .",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Atomic Semantic Forms",
"sec_num": "4.2.3."
},
{
"text": "b. Annotation(id=a5) event(e1, w2-3, pred:run, tense:past)",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Atomic Semantic Forms",
"sec_num": "4.2.3."
},
{
"text": "c. Semantic form: \u03c3(e1 e ) \u03b1 := {e 1 :e}[run(e 1 ) t \u2227 past(e 1 ) t ] \u03b1 where \":=\" is a meta-symbol standing for \"is\".",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Atomic Semantic Forms",
"sec_num": "4.2.3."
},
{
"text": "Some notes are needed here. (1) For now temporally, the type of \u03c3(e1) is left unspecified: it is only marked with \u03b1, whereas the type of e 1 in the registry is specified as the individual type e. (2) The ID \"e1\" in \u03c3(e1) does not refer to the entity structure of category event, but its abstract specification that conforms to the abstract syntax of the relevant annotation scheme. 3The representation of \u03a3 var:type \u03c3(a Ec ) is exactly the same as DRS except that \u03c3(e1) in ABSr is typed, as in Bos et al. 2017's Groningen Meaning Bank (GMB). The semantic form in (5) can be converted to a type-based DRS except that the type of the entire DRS is not specified.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Atomic Semantic Forms",
"sec_num": "4.2.3."
},
{
"text": "e 1 :e run(e 1 ) t past(e 1 ) t",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Atomic Semantic Forms",
"sec_num": "4.2.3."
},
{
"text": "The current version of ABSr introduces two merge operators, \u2295 and , and their subtypes each marked with a different superscript to represent the merging of (1) two semantic forms or (2) a pair of semantic forms with a functorlike semantic form. The second type of merging is motivated by the treatment of tripartite link structures of the form <\u03b7, E, \u03c1>, where \u03c1 is a type of relation between an entity \u03b7 and a set E of entities, in ABSr . These operators are non-Boolean connectives. They are needed to be able to merge semantic forms of type other than the truth-type t. More operators may need to be introduced to treat finer-grained compositions, especially involving the semantics of determiners that include generalized quantifiers, plurals, and the merging of scopes. As suggested by Bunt (personal communication), different symbols will be introduced to represent various subtypes of composition. 6 For the formulation of composition rules, it is assumed that these rules hold for any well-formed semantic forms A \u03b1 , B \u03b2 , and C \u03b3 , each of which is typed as \u03b1, \u03b2, and \u03b3, respectively. For these semantic forms, there are two major types of composition, conjunctive (\u2295) and distributive ( ), and then their subtypes:",
"cite_spans": [
{
"start": 905,
"end": 906,
"text": "6",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Composed Semantic Forms",
"sec_num": "4.2.4."
},
{
"text": "(7) Types of composition:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Composed Semantic Forms",
"sec_num": "4.2.4."
},
{
"text": "Conjunctive composition (\u2295): ",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Composed Semantic Forms",
"sec_num": "4.2.4."
},
{
"text": "a. [A t \u2295 bo C t ] \u03b1 := [A t \u2227 C t ] t b. [{A t , B t } \u03b1 \u2295 bo C t ] := [[A t \u2227 B t ] t \u2227 C t ]",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Composed Semantic Forms",
"sec_num": "4.2.4."
},
{
"text": "Rule 1 bo applies to most of the annotation structures in ISO-TimeML (ISO, 2012), ISO-Space (ISO, 2020), and ISO standard on semantic role annotation (ISO, 2014). For illustration, consider 9 (10) a. Semantic forms of the entity structures:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Composed Semantic Forms",
"sec_num": "4.2.4."
},
{
"text": "\u03c3(x1) t := {x 1 :e}[dog(x 1 ) \u2227 named(x 1 , F ido)] \u03c3(e1) t := {e 1 :v}[bark(e 1 ) \u2227 presProg(e 1 )]",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Composed Semantic Forms",
"sec_num": "4.2.4."
},
{
"text": "b. Semantic form of Semantic role link: \u03c3(srlink) t := {x 1 :e, e 1 :v}",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Composed Semantic Forms",
"sec_num": "4.2.4."
},
{
"text": "[{\u03c3(x1) t , \u03c3(e 1 ) t } \u2295 bo agent(e 1 , x 1 ) t ] := {x 1 :e, e 1 :v} [[\u03c3(x1) t \u2227 \u03c3(e 1 ) t ] \u2227 agent(e 1 , x 1 ) t ] := {x 1 :e, e 1 :v} [[dog(x 1 ) \u2227 named(x 1 , F ido)] \u2227 [bark(e 1 ) \u2227 presProg(e 1 )] \u2227 agent(e 1 , x 1 )]",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Composed Semantic Forms",
"sec_num": "4.2.4."
},
{
"text": "c. Semantic form of annotation structure: \u03c3(a 9 ) := {x:e, e:v}\u03c3(srlink)",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Composed Semantic Forms",
"sec_num": "4.2.4."
},
{
"text": "by Variable renaming and binding := {x:e, e:v}[bark(e) \u2227 presProg(e)] \u2227 agent(e, x)]",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Composed Semantic Forms",
"sec_num": "4.2.4."
},
{
"text": "Rule 1 fa Functional conjunctive composition reflects the functional application of a functor applying to its argument(s) in Montague Semantics (Montague, 1974) or (Dowty et al., 1981) . Rule 1 fa is formulated in (11):",
"cite_spans": [
{
"start": 144,
"end": 160,
"text": "(Montague, 1974)",
"ref_id": "BIBREF41"
},
{
"start": 164,
"end": 184,
"text": "(Dowty et al., 1981)",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Composed Semantic Forms",
"sec_num": "4.2.4."
},
{
"text": "(11) Rule 1 fa Functional conjunctive composition: a. [A \u03b1 \u2295 fa C \u03b1\u2192t) ] := [A t \u2227 C t ] or b. [{A \u03b1 , B \u03b2 }] \u2295 fa C \u03b2\u2192(\u03b1\u2192t) ] := [[A t \u2227 B t ] \u2227 C t ]",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Composed Semantic Forms",
"sec_num": "4.2.4."
},
{
"text": "Example (9) can be analyzed in terms of a functor-argument analysis by assigning a functional type \u03b1 \u2192 t, where \u03b1 is a type, to the type of each of the annotation structures.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Composed Semantic Forms",
"sec_num": "4.2.4."
},
{
"text": "(12) a. Semantic forms of the entity structures:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Composed Semantic Forms",
"sec_num": "4.2.4."
},
{
"text": "\u03c3(x1) e\u2192t := {x 1 :e}[dog(x 1 ) \u2227 named(x 1 , F ido)] \u03c3(e1) v\u2192t := {e 1 :v}[bark(e 1 )\u2227 presProg(e 1 )]",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Composed Semantic Forms",
"sec_num": "4.2.4."
},
{
"text": "b. Semantic form of Semantic role link: The functional composition with the operator \u2295 fa is equivalent to the functional application in \u03bb-calculus, as shown by (13):",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Composed Semantic Forms",
"sec_num": "4.2.4."
},
{
"text": "\u03c3(srlink) := {x 1 :e, e 1 :v} [{\u03c3(x1) e\u2192t , \u03c3(e 1 ) v\u2192t } \u2295 f a agent(e 1 , x 1 ) (v\u2192t)\u2192((e\u2192t)\u2192t) ] := {x 1 :e, e 1 :v} [[\u03c3(x1) t \u2227 \u03c3(e 1 ) t ] \u2227 agent(e 1 , x 1 ) t ] := {x 1 :e, e 1 :v} [[dog(x 1 ) \u2227 named(x 1 , F ido)] t \u2227 [bark(e 1 ) \u2227 presProg(e 1 )] t \u2227 agent(e 1 , x 1 ) t ]",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Composed Semantic Forms",
"sec_num": "4.2.4."
},
{
"text": "(13) a. Arguments:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Composed Semantic Forms",
"sec_num": "4.2.4."
},
{
"text": "\u03c3(x1) e\u2192t := \u03bbx 1 [dog(x 1 ) \u2227 named(x 1 , F ido)] \u03c3(e1) v\u2192t := \u03bbe 1 [bark(e 1 )\u2227 presProg(e 1 )]",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Composed Semantic Forms",
"sec_num": "4.2.4."
},
{
"text": "b. Funtor for Semantic role link applying to the two arguments in (a):",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Composed Semantic Forms",
"sec_num": "4.2.4."
},
{
"text": "\u03c3(srlink) := [\u03bbQ[\u03bbP [P (x 1 ) \u2227 Q(e 1 ) \u2227 agent(e, x)](\u03c3(e 1 ))](\u03c3(x 1 ))]",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Composed Semantic Forms",
"sec_num": "4.2.4."
},
{
"text": "By applying four \u03bb-conversions to (13b), we obtain the same result as (12c). One noticeable problem with the functional application in \u03bb-calculus is the placing of the arguments in the right order when the functor applies to them. Unlike the equation solving approach proposed here, Kamp and Reyle (1993) represents names like Fido as F ido(x) of type t in DRSs. This is acceptable but fails to apply the substitution of identicals. Note also that the equation solving approach can be extended to basic types other than entity type e.",
"cite_spans": [
{
"start": 283,
"end": 304,
"text": "Kamp and Reyle (1993)",
"ref_id": "BIBREF28"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Composed Semantic Forms",
"sec_num": "4.2.4."
},
{
"text": "[{A \u03b1 , B \u03b2 } C \u03b2\u2192(\u03b1\u2192t) ] := [A t \u2192 c B t ] t ,",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Rule 2 Distributive Composition ( ):",
"sec_num": null
},
{
"text": "where \u2192c refers to an implication the type of which needs to be specified for each case and A and B are minimal modifications of A and B.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Rule 2 Distributive Composition ( ):",
"sec_num": null
},
{
"text": "The conjunctive operator \u2295 and its subtypes generate truthfunctional conjunctions. In contrast, the distributive operator possibly with its subtypes generates non-conjunctive relations of implication the type or meaning of which needs further analysis. ",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Rule 2 Distributive Composition ( ):",
"sec_num": null
},
{
"text": ":= {e 1 , t 1 }[{\u03c3(e1) v\u2192t , \u03c3(t1) i\u2192t } \u2295 f a occurs(e 1 , t 1 ) (i\u2192t)((v\u2192t)\u2192t) ] := {e 1 , t 1 }[[\u03c3(e1) t \u2227 \u03c3(t1) t ] \u2227 \u03c3(tlink) t ] :={e 1 , t 1 }[[die(e 1 ) \u2227 past(e 1 )] \u2227 year(t 1 ,2019) \u2227 occurs(e 1 , t 1 )]",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Additional Illustrations",
"sec_num": "4.3."
},
{
"text": "c. Semantic form of annotation structure: \u03c3(a 17 ) := {e, t}\u03c3(tlink) := {e, t}[die(e) \u2227 past(e) \u2227 year(t,2019) \u2227occurs(e, t)]",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Additional Illustrations",
"sec_num": "4.3."
},
{
"text": "Rule 1 eq Equation solving (\u2295 eq ) applies to the annotation structures that contain names or other basic types. Consider an example taken from Pustejovsky et al. 2019 The treatment of a spatial relation given in (19d,e) fails to indicate which location stands for x and which for y. In fact, one of the difficulties with \u03bb-operation is where to place its arguments. Example (19) can be treated more explicitly with Rule 1 eq equation solving. With the rule of substitution of identicals, we then obtain the same result in(G, S), as given in (19e).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Additional Illustrations",
"sec_num": "4.3."
},
{
"text": "Rule 2 Distributive composition with the operator applies to subordination or quantification constructions. Consider example (21), called equi-NP construction. 8",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Additional Illustrations",
"sec_num": "4.3."
},
{
"text": "(21) a. John x1,w1 wants e1,w2 to teach e2,w4 on Monday.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Additional Illustrations",
"sec_num": "4.3."
},
{
"text": "b. Annotation (id = a 21 ): Entity structures: entity(x1, w1, form:John) event(e1, w2, pred:want, theme(e1,e2)) event(e2, w4, pred:teach, agent(e2,x1)) Subordination link structure: slink(e1, e2, modal) 9 Pustejovsky et al. (2005) annotated the subordination relation between two events, want(e 1 ) and teach(e 2 ) as being modal. Montague Semantics, in contrast, treats it as a relation between the intensional predicate want and the property of teaching. However, the intensionality of the predicate want in the main clause requires Rule 2 i with an operator i , a subtype of disjunctive composition for intensional cases like \u03c3(a 21 ).",
"cite_spans": [
{
"start": 205,
"end": 230,
"text": "Pustejovsky et al. (2005)",
"ref_id": "BIBREF45"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Additional Illustrations",
"sec_num": "4.3."
},
{
"text": "(22) a. Semantic forms of the entity structures: The semantic form \u03c3(a 21 ) shows that the predicate want has the event e 2 as its theme and that the agent of the predicate go in the subordinated complement is John. The non-Boolean connective \u2192 int connects the semantic forms of the two components of the subordination construction (21) involving the intensional predicate want. The connective \u2192 i needs to be defined as part of a model structure with a tentative definition as in (23):",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Additional Illustrations",
"sec_num": "4.3."
},
{
"text": "\u03c3(x1) t := {x 1 }[x 1 =John] \u03c3(e1) E ,where E=(v",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Additional Illustrations",
"sec_num": "4.3."
},
{
"text": "(23) Definition of \u2192 int (tentative)",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Additional Illustrations",
"sec_num": "4.3."
},
{
"text": "Given a model M for a modal logic with a set W of possible worlds W that includes the actual world w0 and an intentional world wi accessible from w0, and two semantic forms, \u03c6 and \u03c8, of type t,",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Additional Illustrations",
"sec_num": "4.3."
},
{
"text": "[[\u03c6 \u2192 i \u03c8]] M,w0 =1 iff [[\u03c8]] M,wi =1 provided [[\u03c6]] M,w0 =1.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Additional Illustrations",
"sec_num": "4.3."
},
{
"text": "This means that the eventuality of \"teaching (on Monday)\" is or becomes realized in the mind (intended world) of the experiencer John only.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Additional Illustrations",
"sec_num": "4.3."
},
{
"text": "8 Annotation a21 is simplified to focus on the subordination link (slink).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Additional Illustrations",
"sec_num": "4.3."
},
{
"text": "9 This example is taken from Pustejovsky et al. 2005, p. 553.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Additional Illustrations",
"sec_num": "4.3."
},
{
"text": "Semantic forms are subject to a model-theoretic interpretation. Each well-formed semantic form \u03c3(a) of an annotation structure a is interpreted with respect to a model M and an assignment g of values to variables. [[\u03c3(a) ]] M,g is then understood as the interpretation or denotation of \u03c3(a).",
"cite_spans": [
{
"start": 214,
"end": 220,
"text": "[[\u03c3(a)",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "General",
"sec_num": "5.1."
},
{
"text": "The structure of each model M depends on the kind of semantic annotation. For the interpretation of temporal annotation, for instance, a set of times T and a set of temporal relations such as the precedence relation \u227a over T become a part of its model structure. Furthermore, the construction of such a model is constrained by some possible uses or definitions of logical predicates, called meaning postulates, as is discussed in 5.2.1.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "General",
"sec_num": "5.1."
},
{
"text": "There may be some unbound occurrences of variables in well-formed semantic forms of ABSr . By Rule A.1 for Variable Registry, these variables may be either bound to the discourse referents registered before the semantic form of each of the substructures of an annotation structure or bound existentially when their scope is explicitly specified. Or else they can be interpreted with the assignment g as if they were bound existentially. as in Kamp and Reyle (1993, page 521) . Then its definition is given in (25) as part of an interpretation model structure.",
"cite_spans": [
{
"start": 443,
"end": 474,
"text": "Kamp and Reyle (1993, page 521)",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Interpretation of unbound occurrences of variables",
"sec_num": "5.2."
},
{
"text": "(25) Truth Definition of Predicate past:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Meaning Postulates as Constraints",
"sec_num": "5.2.1."
},
{
"text": "Given an event e, a runtime function \u03c4 from events to times, a time t, and the present time n, as specified in a model structure M , past(e) is true with respect to a model M if and only if \u03c4 (e) \u2286 t and t \u227a n.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Meaning Postulates as Constraints",
"sec_num": "5.2.1."
},
{
"text": "The predicate past may be defined differently to accommodate its deitic or situational use (see Partee (1973) or Quirk et al. (1985) ). Semantic form (26c) is then interpreted by the definition of presPerfect given as part of a model structure. Otherwise, its representation gets complicated similar to DRS, for instance. Here is an example from Cann et al. (2009) . 27 The interpretation of \u03c3(e 1 ) in (28c), for instance, requires the truth-conditional definition of presPerfect(e) that reflects those notions of the perfective aspect encoded in DRS (27b) above. Furthermore, the proposed way of treating tense, aspect, and other complex predicates allows different interpretations or uses of them. Those predicates that constitute part of the representation language of semantic forms in ABSr , however, require truth-definitions or meaning postulates that constrain and define a set of admissible model structures (see Carnap (1947 Carnap ( 1956 Montague (1974; Dowty (1979) ).",
"cite_spans": [
{
"start": 96,
"end": 109,
"text": "Partee (1973)",
"ref_id": "BIBREF43"
},
{
"start": 113,
"end": 132,
"text": "Quirk et al. (1985)",
"ref_id": "BIBREF50"
},
{
"start": 346,
"end": 364,
"text": "Cann et al. (2009)",
"ref_id": "BIBREF9"
},
{
"start": 923,
"end": 935,
"text": "Carnap (1947",
"ref_id": "BIBREF10"
},
{
"start": 936,
"end": 949,
"text": "Carnap ( 1956",
"ref_id": "BIBREF10"
},
{
"start": 950,
"end": 965,
"text": "Montague (1974;",
"ref_id": "BIBREF41"
},
{
"start": 966,
"end": 978,
"text": "Dowty (1979)",
"ref_id": "BIBREF17"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Meaning Postulates as Constraints",
"sec_num": "5.2.1."
},
{
"text": "6. Applications",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Meaning Postulates as Constraints",
"sec_num": "5.2.1."
},
{
"text": "ISO-Space (ISO, 2020) introduces the movement link (movelink) to annotate motions involving paths. The predicate traverses associated with motions is one of the logical predicates that need to be defined in the model structure of ABS . It can also be illustrated how the semantic forms involving motions and paths can be derived through Rule 1 bo Boolean conjunctive composition, as is demonstrated in (29).",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Boolean Conjunctive Composition",
"sec_num": "6.1."
},
{
"text": "(29) a. Marakbles: Mia x1,w1 arrived m1,w2 \u2205 ep1 in Boston pl1,w4 yesterday.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Boolean Conjunctive Composition",
"sec_num": "6.1."
},
{
"text": "b. Annotation (id=a 29 ): Entity structures: entity(x1,w1, type:person, form:nam) motion(m1,w2, pred:arrive, type: transition, tense:past) eventPath(ep1,\u2205, start:unspecified, end:pl1, trigger(m1,ep1)) place(pl1,w4, type:city, form:nam) Movement link structure: movelink(figure:x1, ground:ep1, relType:traverses)",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Boolean Conjunctive Composition",
"sec_num": "6.1."
},
{
"text": "Each markable is identified with an ID associated with its category and anchored to a word. Motions, as denoted by verbs like arrive, trigger a path, called event-path. This path is marked with a null category or non-consuming tag \u2205 because it is not associated with any non-null string of words.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Boolean Conjunctive Composition",
"sec_num": "6.1."
},
{
"text": "(30) a. Semantic forms of entity structures:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Boolean Conjunctive Composition",
"sec_num": "6.1."
},
{
"text": "\u03c3(x1) t := [person(x 1 ) \u2227 named(x 1 , M ia)] \u03c3(m1) t := [arrive(m 1 ) \u2227 past(m 1 )] \u03c3(ep1) t := [start(\u03c0, \u03b3(l 0 )) \u2227 end(\u03c0, l 1 ) \u2227 triggers(m 1 , \u03c0)] \u03c3(pl1) t := [named(l 1 , Boston) \u2227 city(l 1 )]",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Boolean Conjunctive Composition",
"sec_num": "6.1."
},
{
"text": "b. Semantic form of the movement link structure:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Boolean Conjunctive Composition",
"sec_num": "6.1."
},
{
"text": "\u03c3(movelink) := [{\u03c3(x1) t , \u03c3(ep1) t } \u2295 bo traverses(x, \u03c0) t ] := [[[person(x 1 ) \u2227 named(x 1 , M ia)] \u2227 [start(\u03c0, \u03b3(l 0 )) \u2227 end(\u03c0, l 1 ) \u2227 triggers(m 1 , \u03c0)] \u2227 [named(l 1 , Boston) \u2227 city(l 1 )]] \u2227 traverses(x, \u03c0)]",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Boolean Conjunctive Composition",
"sec_num": "6.1."
},
{
"text": "c. Annotation structure:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Boolean Conjunctive Composition",
"sec_num": "6.1."
},
{
"text": "\u03c3(a 29 ) := {x 1 , \u03c0 1 , l 0 , l 1 , m 1 }\u03c3(movelink) =: {x, \u03c0, l 0 , l 1 , m} [[[person(x) \u2227 named(x, M ia)] \u2227 [start(\u03c0, \u03b3(l 0 )) \u2227 end(\u03c0, l 1 ) \u2227 triggers(m, \u03c0)] \u2227 [named(l 1 , Boston) \u2227 city(l 1 )]] \u2227 traverses(x, \u03c0)]",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Boolean Conjunctive Composition",
"sec_num": "6.1."
},
{
"text": "All of the semantic forms that are derived through various links have been shown to undergo Rule 1 bo Boolean conjunctive composition only. This was illustrated with srlink for semantic role labeling, tlink for temporal anchoring, qslink for the location of regions, and movelink for the annotation of motions involving their movers and event-paths.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Boolean Conjunctive Composition",
"sec_num": "6.1."
},
{
"text": "Besides its subtype int for intensional subordinate constructions, the distributive composition can have other subtypes. Here I introduce Rule 2 imp with the operator imp for the case of implication. The word if in English triggers a conditional sentence which is often interpreted as a d. \u03c3(a 32b ) := {e 1 , e 2 , e 3 .t 1 , \u03b3(t2)}\u03c3(slink) [[rain(e 1 ) \u2227 date(t 1 ,2019-02-04) \u2227 occurs(e 1 , t 1 )] t \u2192 [[beCanceled(e 3 ) \u2227 theme(e 3 , e 2 ) \u2227 future(e 3 )] \u2227 \u03b3(t 2 ) \u2227 occurs(e 3 , \u03b3(t 2 )) t ]]",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Distributive Composition for Conditionals",
"sec_num": "6.2."
},
{
"text": "With respect to the operator imp , the semantic form of the antecedent, \u03c3(tl1), is understood to be the restrictor R and that of the consequent, \u03c3(tl2), is the nuclear scope N , while the relation of implication between them is represented by the operator \u2192.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Distributive Composition for Conditionals",
"sec_num": "6.2."
},
{
"text": "There have been several theoretical works showing how annotation structures can be interpreted and a variety of largescale computational efforts to implement them for computational applications. Some of them are annotation-based semantics in one way or another. Hobbs and Pustejovsky (2003) develop a semantics for TimeML (Pustejovsky et al., 2005) , based on the OWLtime ontology. They provide a fine-grained way of annotating and interpreting various temporal relations. ABS is designed to accommodate the OWL-time ontology in defining its logical predicates related to temporal annotation. Katz (2007) introduces a denotational semantics that directly interprets TimeML annotation structures represented in XML. The model structure proposed in Katz (2007) becomes part of the temporal model structure for ABS . Bunt (2007) and Bunt (2011) introduce a semantics for semantic annotation. This eventually develops into a semantics based on the abstract syntax of a semantic annotation scheme. Bunt (2020a) and Bunt (2020b) have developed QuantML, a markup language for quantification, that can apply to the annotation and interpretation of a full-range of features related to quantification such as the definiteness, involvement or collectivity (distributivity) of entities or scope ambiguity involving quantifiers and eventualities. Lee (2008) and Lee (2011) follow the OWL-time ontology and a compositional approach to work on temporal annotations with an extensive use of \u03bb-operations. It shows some degree of complexity in the use of \u03bb-operations when they are recursively embedded, for it requires to raise the order of variables as the embedding gets deeper. One of the reasons for introducing ABSr is to avoid recursive embedding and substitutions (see Hausser (2015) ). For now, ABSr has Rule 1 sub Substitutive conjunctive composition, but this should be deleted eventually except for the illustration of rudimentary annotations involving names and other basic types. Database Semantics (DBS) (Hausser, 2006) provides a theoretical foundation for the understanding of language analysis and generation without recursions and substitutions, but with the associative linear processing of language. This has motivated the design of ABS to some extent. Then there are other types of semantics that present different ways of representing meaning in language. Banarescu et al. (2013) introduce AMR (the Abstract Meaning Representation) to represent the semantic roles mainly based on PropBank in a logical format, PENNMAN format, or directed graph structure. He (2018) also introduces a way of annotating semantic roles, which is called Shallow Semantics, without relying on pre-defined syntactic structures but introducing syntax-independent span-based neural models or labelled span-graph networks (LSGNs). Based on syntax-free annotations, ABSr is also syntaxindependent. Its current representation format is strictly linear but needs to move onto a graphic mode for visual purposes. The composition rules of ABSr are constrained by type matching and also syntax-independent unlike Moens and Steedman (1988) 's categorial grammar or Kamp and Reyle (1993) 's DRSs. Dobnik et al. (2012) and Dobnik and Cooper (2017) introduce a type theory with records to constrain semantic representations and their manipulations in language processing. Their type system, especially related to spatial perception, will properly orient the spatiotemporal annotation of ISO-Space and meaning representation through ABS . The earlier work of Pustejovsky (2001) on type construction also lays a basis for the type theory of ABS for a finer-grained treatment of entities and eventualities.",
"cite_spans": [
{
"start": 262,
"end": 290,
"text": "Hobbs and Pustejovsky (2003)",
"ref_id": "BIBREF22"
},
{
"start": 322,
"end": 348,
"text": "(Pustejovsky et al., 2005)",
"ref_id": "BIBREF45"
},
{
"start": 593,
"end": 604,
"text": "Katz (2007)",
"ref_id": "BIBREF29"
},
{
"start": 747,
"end": 758,
"text": "Katz (2007)",
"ref_id": "BIBREF29"
},
{
"start": 814,
"end": 825,
"text": "Bunt (2007)",
"ref_id": "BIBREF4"
},
{
"start": 830,
"end": 841,
"text": "Bunt (2011)",
"ref_id": "BIBREF6"
},
{
"start": 993,
"end": 1005,
"text": "Bunt (2020a)",
"ref_id": "BIBREF7"
},
{
"start": 1010,
"end": 1022,
"text": "Bunt (2020b)",
"ref_id": "BIBREF8"
},
{
"start": 1334,
"end": 1344,
"text": "Lee (2008)",
"ref_id": "BIBREF33"
},
{
"start": 1349,
"end": 1359,
"text": "Lee (2011)",
"ref_id": "BIBREF34"
},
{
"start": 1760,
"end": 1774,
"text": "Hausser (2015)",
"ref_id": "BIBREF20"
},
{
"start": 2002,
"end": 2017,
"text": "(Hausser, 2006)",
"ref_id": "BIBREF19"
},
{
"start": 2362,
"end": 2385,
"text": "Banarescu et al. (2013)",
"ref_id": "BIBREF2"
},
{
"start": 3087,
"end": 3112,
"text": "Moens and Steedman (1988)",
"ref_id": "BIBREF40"
},
{
"start": 3138,
"end": 3159,
"text": "Kamp and Reyle (1993)",
"ref_id": "BIBREF28"
},
{
"start": 3169,
"end": 3189,
"text": "Dobnik et al. (2012)",
"ref_id": "BIBREF15"
},
{
"start": 3194,
"end": 3218,
"text": "Dobnik and Cooper (2017)",
"ref_id": "BIBREF14"
},
{
"start": 3528,
"end": 3546,
"text": "Pustejovsky (2001)",
"ref_id": "BIBREF49"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Related Work",
"sec_num": "7.1."
},
{
"text": "For the computational applications of semantic annotations, the Gronigen Meaning Bank (GMB) (Bos et al., 2017) is very much related to the basic motivation of ABS in efforts to modify the classical version of DRT by making its syntax based on a (Montagovian) type systems consisting of two types, e and t, and by translating DRSs into a firstorder logic only, for instance, while deleting so-called duplex conditions in DRSs. The basic design of the Parellel Meaning Bank (PMB) also adopts DRT as its formalism for meaning representation while adopting Combinatory Categorial Grammar as its syntax. Since it applies to multilingual annotation, ABS can make use of it when the ISO standards on semantic annotation are extended to multilingual annotations, especially for the purposes of multilingual translations. Nevertheless, the theoretical framework of ABS and its representation language is conservative in practice, being essentially based on the \u03bb-calculus and the graphic representation of Kamp and Reyle (1993) 's DRT. This will be shown in the ensuing Subsection 7.2.",
"cite_spans": [
{
"start": 92,
"end": 110,
"text": "(Bos et al., 2017)",
"ref_id": "BIBREF3"
},
{
"start": 997,
"end": 1018,
"text": "Kamp and Reyle (1993)",
"ref_id": "BIBREF28"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Related Work",
"sec_num": "7.1."
},
{
"text": "The composition of semantic forms is constrained by their semantic types. These types simply reflect those in Montague semantics (Montague, 1974) and (Dowty et al., 1981) and also the extended type theory by Kracht (2002) and Pustejovsky et al. (2019) , thus making all these semantic forms isomorphic to those \u03bb-constructions in \u03bb-calculus. If such a typing of the semantic forms of annotation structures is ignored or if each of the semantic forms is treated as being of type t, then these semantic forms can easily be converted to DRSs (Kamp and Reyle, 1993) . There is an option to choose a type-theoretic semantics or not. ABS allows both but prefers to choose a type-theoretic semantics to constrain its representation language ABSr , while enriching its interpretation model structure, as shown in Figure 2 . 11 11 Although Figure 2 indicates that DRT/DRSs are not based on If a type theory is adopted, then the logical predicates can be defined in terms of type-theoretic higher-order logic.",
"cite_spans": [
{
"start": 129,
"end": 145,
"text": "(Montague, 1974)",
"ref_id": "BIBREF41"
},
{
"start": 150,
"end": 170,
"text": "(Dowty et al., 1981)",
"ref_id": null
},
{
"start": 208,
"end": 221,
"text": "Kracht (2002)",
"ref_id": "BIBREF30"
},
{
"start": 226,
"end": 251,
"text": "Pustejovsky et al. (2019)",
"ref_id": "BIBREF47"
},
{
"start": 539,
"end": 561,
"text": "(Kamp and Reyle, 1993)",
"ref_id": "BIBREF28"
},
{
"start": 816,
"end": 821,
"text": "11 11",
"ref_id": null
}
],
"ref_spans": [
{
"start": 805,
"end": 813,
"text": "Figure 2",
"ref_id": "FIGREF1"
},
{
"start": 831,
"end": 839,
"text": "Figure 2",
"ref_id": "FIGREF1"
}
],
"eq_spans": [],
"section": "Convertibility",
"sec_num": "7.2."
},
{
"text": "In ABS , the choice of a theory depends on the treatment of unbound variables and unspecified types. ABS treats logical forms with occurrences of unbound variables as wellformed semantic forms. Individual (or predicate) variables may occur unbound in well-formed semantic forms, as in the interval temporal logic of Pratt-Hartmann (2007) . 12 Here is an example with a markable \"visited\" e1 :",
"cite_spans": [
{
"start": 316,
"end": 337,
"text": "Pratt-Hartmann (2007)",
"ref_id": "BIBREF44"
},
{
"start": 340,
"end": 342,
"text": "12",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Convertibility",
"sec_num": "7.2."
},
{
"text": "(34) a. Data:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Convertibility",
"sec_num": "7.2."
},
{
"text": "Mia x1 visited e1 Berlin, New York, [last year] t1 .",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Convertibility",
"sec_num": "7.2."
},
{
"text": "b. Annotation (id=a 5.unbound ): Entity structures: event(e1, m1, pred:visit, tense:past) timex3(t1, m2, type:gYear, value:2019)",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Convertibility",
"sec_num": "7.2."
},
{
"text": "Link structure: tlink(e1, t1, isIncluded) c. Semantic Forms:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Convertibility",
"sec_num": "7.2."
},
{
"text": "\u03c3(e1) \u03b1 := {e 1 }[visit(e 1 ) \u2227 past(e 1 )] \u03c3(t1) \u03b2 := {t 1 }[gYear=(t 1 , 2019)] \u03c3(tlink) \u03b3 := {e 1 , t 1 }[{\u03c3(e1), \u03c3(t1)} occurs(e 1 , t 1 )]",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Convertibility",
"sec_num": "7.2."
},
{
"text": "Each of the semantic forms in (34c) contains some variables which are registered in its preamble. In ABSr , these variables can be bound in two different ways, either by the existential quantifier or by the \u03bb-operator. The assignment of a type to each semantic form depends on which way these (registered) variables are bound. The type of each semantic form is:",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Convertibility",
"sec_num": "7.2."
},
{
"text": "\u2022 Case 1: either of type t (truth-value carrying) as if the unbound variables were bound by the existential quantifier \u2203: i.e., \u2203{e}[visit(e) \u2227 past(e)] (type t)",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Convertibility",
"sec_num": "7.2."
},
{
"text": "\u2022 Case 2: or of some functional type (predicate) as if the unbound variables were bound by the \u03bb-operator: i.e., \u03bbe[visit(e) \u2227 past(e)] (type v \u2192 t) a type theory, the DRT formalism adopted by Bos et al. (2017) is based on a type theory.",
"cite_spans": [
{
"start": 193,
"end": 210,
"text": "Bos et al. (2017)",
"ref_id": "BIBREF3"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Convertibility",
"sec_num": "7.2."
},
{
"text": "12 ABS has no predicate variables.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Convertibility",
"sec_num": "7.2."
},
{
"text": "Depending on which case is chosen, the semantic form of a link like \u03c3(tlink) in (34c) undergoes a different rule of composition.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Convertibility",
"sec_num": "7.2."
},
{
"text": "Case 1 allows the conversion of semantic forms in ABS to DRSs. As shown in (35), Case 1 Boolean conjunctive composition (\u2295 bo ) can easily be converted to an equivalent DRS.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Convertibility",
"sec_num": "7.2."
},
{
"text": "(36) Case 1 in DRS: e t visit(e) past(e) gYear (t,2019) occurs(e,t)",
"cite_spans": [
{
"start": 47,
"end": 55,
"text": "(t,2019)",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Convertibility",
"sec_num": "7.2."
},
{
"text": "Although the application of Rule 1 bo Boolean conjunctive composition is type-constrained, there is no such a constraint on the derivation of DRSs.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Convertibility",
"sec_num": "7.2."
},
{
"text": "Case 2 allows the conversion of semantic forms in ABSr to well-formed forms in \u03bb-calculus as in Montague Semantics (Montague, 1974) . For the illustration of Case 2, consider example (34), as was just given:",
"cite_spans": [
{
"start": 115,
"end": 131,
"text": "(Montague, 1974)",
"ref_id": "BIBREF41"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Convertibility",
"sec_num": "7.2."
},
{
"text": "(37) Case 2: Rule 2 Functional conjunctive composition (\u2295 f a ): a. \u03c3(tlink) t := [{\u03c3(e 1 ) E , \u03c3(t 1 ) I } \u2295 f a occurs(e 1 , t 1 ) I\u2192(E\u2192t) ] := [[visit(e 1 ) \u2227 past(e 1 )] \u2227 gYear(t 1 ,2019) \u2227 occurs(e 1 , t 1 )] b. \u03c3(a 34 ) = \u03c3(tlink) t",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Convertibility",
"sec_num": "7.2."
},
{
"text": "The semantic form \u03c3(tlink) in (37) is treated of a functional type, I \u2192 (E \u2192 t), where I is i \u2192 t and E is v \u2192 t. Then the semantic forms \u03c3(e1) and \u03c3(t1) are treated as arguments of \u03c3(tlink) such that they are of types E (set of eventuality descriptions) and I (set of time points), respectively. In the process of the Boolean conjunctive composition, the unbound occurrences of the variables are anchored to the discourse referents e and t, as in DRS, or existentially quantified, while adjusting their variable names accordingly. As for the case of the functional conjunctive composition, the whole process is understood as if all the semantic forms were subject to a series of \u03bb-conversions as in (38):",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Convertibility",
"sec_num": "7.2."
},
{
"text": "(38) \u03bb-operations: a. \u03c3(e 1 ) v\u2192t := \u03bbe 1 [visit(e 1 ) \u2227 past(e 1 )] b. \u03c3(t 1 ) i\u2192t := \u03bbt 1 [gYear(t 1 ,2019)] c. \u03c3(tlink) t := \u03bbT \u03bbE\u2203{e, t}[E(e) \u2227 T (t) \u2227 occurs(e, t)] (\u03c3(e 1 ))(\u03c3(t 1 )) := \u2203{e, t}[\u03c3(e1)(e) \u2227 \u03c3(t1)(t)] := \u2203{e, t} [[visit(e) \u2227 past(e)] \u2227 gYear(t,2019) \u2227 occurs(e, t)]",
"cite_spans": [
{
"start": 232,
"end": 242,
"text": "[[visit(e)",
"ref_id": null
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Convertibility",
"sec_num": "7.2."
},
{
"text": "It should again be stated that the derivation of semantic forms in ABSr does not undergo such \u03bb-operations. The application of Rule 2 Functional conjunctive composition is only implicitly understood to undergo such operations. Unlike semantic forms that involve \u03bb-operations, the application of the \u2295 f a in ABSr does not introduce predicate variables of a higher-order, but individual variables of the first order only. This keeps ABSr to remain at the level of first-order.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Convertibility",
"sec_num": "7.2."
},
{
"text": "As in other parts of ISO 24617 standards on semantic annotation, this paper has a gap in dealing with the semantics of entities and determiners that include generalized quantifiers. Specifically, this paper fails to fully accommodate the new developments on quantification that have been made by Bunt (2020a) and Bunt (2020b) . ABS aims to lighten the burden and possible complexity of generating semantic annotation structures. It would be an ideal situation if semantic annotation structures could have every piece of relevant semantic information encoded into them and be interpreted directly without relying on any intermediate auxiliary representation scheme. But the task of generating such annotation structures and interpreting them directly should easily run into enormous cost and complexity.",
"cite_spans": [
{
"start": 296,
"end": 308,
"text": "Bunt (2020a)",
"ref_id": "BIBREF7"
},
{
"start": 313,
"end": 325,
"text": "Bunt (2020b)",
"ref_id": "BIBREF8"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Concluding Remarks",
"sec_num": "8."
},
{
"text": "ABS is an annotation-based semantics that converts annotation structures to semantic forms for their (model-theoretic) interpretation. For the representation of these semantic forms, ABS provides a simple representation language, a type-theoretic first-order logic without the overuse of \u03bboperations. This language makes use of a small set of logical predicates, such as referring to semantic roles or event and time structures and types, that are defined as part of an interpretation model. The meta-language that defines these logical predicates may be of a higher-order logic.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "Concluding Remarks",
"sec_num": "8."
},
{
"text": "To follow the principle of semantic compositionality, ABS introduces two types of composition with the conjunctive \u2295 and distributive operators and their subtypes over the semantic forms of annotation structures that consist of entity and link structures. Most, if not all, of the link structures in ISO-TimeML and ISO-Space only require conjunctive composition, while quantificational, plural constructions or some subordinated constructions such as the if-then construction may undergo distributive (selective) composition. There are two major types of conjunctive composition: the Boolean type \u2295 boo and the functional type \u2295 f a . Then the functional type has two subtypes, one by substitution \u2295 sub and the other by equation solving \u2295 eq . Annotation structures that are isomorphic to non-embedded structures in Kamp and Reyle (1993) 's DRSs are considered as undergoing the process of Boolean conjunctive composition. In contrast, those annotation structures that match \u03bb-structures in Montague Semantics (Montague, 1974) undergo the functional conjunctive composition. This distinction is not very significant, for the semantic forms of most",
"cite_spans": [
{
"start": 817,
"end": 838,
"text": "Kamp and Reyle (1993)",
"ref_id": "BIBREF28"
},
{
"start": 1011,
"end": 1027,
"text": "(Montague, 1974)",
"ref_id": "BIBREF41"
}
],
"ref_spans": [],
"eq_spans": [],
"section": "Concluding Remarks",
"sec_num": "8."
},
{
"text": "(g) and (h) are my own additions to the list of basic types. 2 See Mani and Pustejovsky (2012) for the discussion of 3.2.2 regions as primitive objects vs. 3.2.3 regions as sets of points.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "Here, it is a bit confusing to use e as standing for a basic type for individual entities and use it as referring to an eventuality of type v: e.g. [runv\u2192t(ev) \u2227 agent(e,x)] e\u2192(v\u2192t ].",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "In a concrete syntax, this category is often called tag or element.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "Bunt (2020b), for instance, introduces the scope merge operator \u2295 s and the possessive scoped merge operator \u2295 ps .",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "In practice, the semantic treatment of names is much more complicated than treating it merely for its referential use. Kamp and Reyle (1993) treat names like \"John\" as a predicate, thus representing it as John(x) in a DRS.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
},
{
"text": "\u03b3 is a function that assigns a time to a deitic temporal expression or a contextually determinable unspecified time.",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "",
"sec_num": null
}
],
"back_matter": [
{
"text": "truth-functional implication in Propositional Logic. Given two well-formed formulas \u03c6 and \u03c8, the conditional formula [\u03c6 \u2192 \u03c8] is treated as a well-formed formula in Propositional Logic and interpreted truth-functionally as being false only if \u03c6 is true but \u03c8 is false. Although the interpretation of conditionals in ordinary language is more complex than the truth-functional interpretation just given, (31) and (32) illustrate how if-constructions are annotated and how their semantic forms are represented in a tripartite structure.(31) Data:If it rains tomorrow, then the picnic will be canceled.(32) a. Annotation of Antecedent (id=a 32a ): \u2227 theme(e 3 , e 2 ) \u2227 future(e 3 )] \u2227 \u03b3(t 2 ) \u2227 occurs(e 3 , \u03b3(t 2 )) t ]]",
"cite_spans": [],
"ref_spans": [],
"eq_spans": [],
"section": "annex",
"sec_num": null
}
],
"bib_entries": {
"BIBREF0": {
"ref_id": "b0",
"title": "Acknowledgements Thanks to Jae-Woong Choe, Chongwon Park, and James Pustejovsky for their reading the preliminary draft with invaluable comments and to the four anonymous reviewers for their detailed constructive comments. I am very much indebted to Harry Bunt for his laborious work to help improve the final submission for publication. I thank them all, but do not claim that all these reviewers agree with my proposal or that I have fully",
"authors": [],
"year": null,
"venue": "the annotation structures undergo the process of Boolean conjunctive composition only",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "the annotation structures undergo the process of Boolean conjunctive composition only. This is the first version of ABS. It requires to be further tested against a variety of larger data and annotation struc- tures. This should be the case especially for the distributive composition involving complex semantic structures. 9. Acknowledgements Thanks to Jae-Woong Choe, Chongwon Park, and James Pustejovsky for their reading the preliminary draft with in- valuable comments and to the four anonymous reviewers for their detailed constructive comments. I am very much indebted to Harry Bunt for his laborious work to help im- prove the final submission for publication. I thank them all, but do not claim that all these reviewers agree with my pro- posal or that I have fully succeeded in accommodating their comments and suggestions. 10. Bibliographical References",
"links": null
},
"BIBREF1": {
"ref_id": "b1",
"title": "Towards a general theory of action and time",
"authors": [
{
"first": "J",
"middle": [
"F"
],
"last": "Allen",
"suffix": ""
}
],
"year": 1984,
"venue": "Artifical Intelligence",
"volume": "23",
"issue": "",
"pages": "123--54",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Allen, J. F. (1984). Towards a general theory of action and time. Artifical Intelligence, 23:123-54.",
"links": null
},
"BIBREF2": {
"ref_id": "b2",
"title": "Abstract meaning representationf or sembanking",
"authors": [
{
"first": "L",
"middle": [],
"last": "Banarescu",
"suffix": ""
},
{
"first": "C",
"middle": [],
"last": "Bonial",
"suffix": ""
},
{
"first": "S",
"middle": [],
"last": "Cai",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Georgescu",
"suffix": ""
},
{
"first": "K",
"middle": [],
"last": "Griffitt",
"suffix": ""
},
{
"first": "U",
"middle": [],
"last": "Hermjakob",
"suffix": ""
},
{
"first": "K",
"middle": [],
"last": "Knight",
"suffix": ""
},
{
"first": "P",
"middle": [],
"last": "Koehn",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Palmer",
"suffix": ""
},
{
"first": "N",
"middle": [],
"last": "Schenider",
"suffix": ""
}
],
"year": 2013,
"venue": "Proceedings of the 7th Linguistic Annotation Workshop and Interoperability with Discourse",
"volume": "",
"issue": "",
"pages": "178--186",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Banarescu, L., Bonial, C., Cai, S., Georgescu, M., Griffitt, K., Hermjakob, U., Knight, K., Koehn, P., Palmer, M., and Schenider, N. (2013). Abstract meaning representa- tionf or sembanking. In Proceedings of the 7th Linguis- tic Annotation Workshop and Interoperability with Dis- course, pages 178-186, Sofia, Bulgaria, August.",
"links": null
},
"BIBREF3": {
"ref_id": "b3",
"title": "The Groningen Meaning Bank",
"authors": [
{
"first": "J",
"middle": [],
"last": "Bos",
"suffix": ""
},
{
"first": "V",
"middle": [],
"last": "Basile",
"suffix": ""
},
{
"first": "K",
"middle": [],
"last": "Evang",
"suffix": ""
},
{
"first": "N",
"middle": [
"J"
],
"last": "Venhuizen",
"suffix": ""
},
{
"first": "J",
"middle": [],
"last": "Bjerva",
"suffix": ""
}
],
"year": 2017,
"venue": "Handbook of Linguistic Annotation",
"volume": "",
"issue": "",
"pages": "463--496",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Bos, J., Basile, V., Evang, K., Venhuizen, N. J., and Bjerva, J. (2017). The Groningen Meaning Bank. In Nancy Ide et al., editors, Handbook of Linguistic Annotation, pages 463-496. Springer, Berlin.",
"links": null
},
"BIBREF4": {
"ref_id": "b4",
"title": "The Korea Society for Language and Information",
"authors": [
{
"first": "H",
"middle": [],
"last": "Bunt",
"suffix": ""
}
],
"year": 2007,
"venue": "Proceedings of the 21st Pacific Asia Conference on Language, Information and Computation",
"volume": "",
"issue": "",
"pages": "13--28",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Bunt, H. (2007). The semantics of semantic annotations. In Proceedings of the 21st Pacific Asia Conference on Language, Information and Computation, pages 13-28, Seoul, Korea. The Korea Society for Language and In- formation.",
"links": null
},
"BIBREF5": {
"ref_id": "b5",
"title": "A methodology for designing semantic annotation languages exploiting semantic-syntactic isomorphisms",
"authors": [
{
"first": "H",
"middle": [],
"last": "Bunt",
"suffix": ""
}
],
"year": 2010,
"venue": "Proceedings of the Second International Conference on Global Interoperability for Language Resources (ICGL20100)",
"volume": "",
"issue": "",
"pages": "29--46",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Bunt, H. (2010). A methodology for designing semantic annotation languages exploiting semantic-syntactic iso- morphisms. In Alex C. Fang, et al., editors, Proceedings of the Second International Conference on Global Inter- operability for Language Resources (ICGL20100), pages 29-46, City University of Hong Kong, Hong Kong.",
"links": null
},
"BIBREF6": {
"ref_id": "b6",
"title": "Introducing abstract syntax + semantics in semantic annotation, and its consequences for the annotation of time and events",
"authors": [
{
"first": "H",
"middle": [],
"last": "Bunt",
"suffix": ""
}
],
"year": 2011,
"venue": "Recent Trends in Language an Knowledge Processing",
"volume": "",
"issue": "",
"pages": "157--204",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Bunt, H. (2011). Introducing abstract syntax + semantics in semantic annotation, and its consequences for the an- notation of time and events. In Eunryoung Lee et al., editors, Recent Trends in Language an Knowledge Pro- cessing, pages 157-204. Hankookmunhwasa, Seoul.",
"links": null
},
"BIBREF7": {
"ref_id": "b7",
"title": "Annotation of quantification: the current state of ISO 24617-12",
"authors": [
{
"first": "H",
"middle": [],
"last": "Bunt",
"suffix": ""
}
],
"year": 2020,
"venue": "Proceedings of the 16th Joint ISO-ACL/SIGSEM Workshop on Interoperable Semantic Annotation",
"volume": "",
"issue": "",
"pages": "1--13",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Bunt, H. (2020a). Annotation of quantification: the cur- rent state of ISO 24617-12. In Harry Bunt, editor, Pro- ceedings of the 16th Joint ISO-ACL/SIGSEM Workshop on Interoperable Semantic Annotation, pages 1-13, May. A satellite workshop at LREC 2020, May 11-15, 2020, Marseille, France (postponed due to COVID-19).",
"links": null
},
"BIBREF8": {
"ref_id": "b8",
"title": "Semantic Annotation of Quantification in Natural Language",
"authors": [
{
"first": "H",
"middle": [],
"last": "Bunt",
"suffix": ""
}
],
"year": 2002,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Bunt, H., (2020b). Semantic Annotation of Quantification in Natural Language. TiCC/Department of Cognitive Science and Artificial Intelligence, Tilburg University, Tilburg, 2nd edition, February. TiCC TR 2020-2.",
"links": null
},
"BIBREF9": {
"ref_id": "b9",
"title": "Semantics: An Introduction to Meaning in Language",
"authors": [
{
"first": "R",
"middle": [],
"last": "Cann",
"suffix": ""
},
{
"first": "R",
"middle": [],
"last": "Kempson",
"suffix": ""
},
{
"first": "E",
"middle": [],
"last": "Gregoromichelaki",
"suffix": ""
}
],
"year": 2009,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Cann, R., Kempson, R., and Gregoromichelaki, E. (2009). Semantics: An Introduction to Meaning in Language. Cambridge University Press, Cambridge.",
"links": null
},
"BIBREF10": {
"ref_id": "b10",
"title": "Meaning and Necessity: A Study in Semantics and Modal Logic",
"authors": [
{
"first": "R",
"middle": [],
"last": "Carnap",
"suffix": ""
}
],
"year": 1947,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Carnap, R. (1947, 1956). Meaning and Necessity: A Study in Semantics and Modal Logic. The University of Chicago Press, Chicago, 2nd edition.",
"links": null
},
"BIBREF11": {
"ref_id": "b11",
"title": "Minimal recursion semantics: an introduction",
"authors": [
{
"first": "A",
"middle": [],
"last": "Copestake",
"suffix": ""
},
{
"first": "D",
"middle": [],
"last": "Flickinger",
"suffix": ""
},
{
"first": "I",
"middle": [],
"last": "Sag",
"suffix": ""
},
{
"first": "C",
"middle": [],
"last": "Pollard",
"suffix": ""
}
],
"year": 2005,
"venue": "Journal of Research on Language and Computation",
"volume": "",
"issue": "",
"pages": "281--332",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Copestake, A., Flickinger, D., Sag, I., and Pollard, C. (2005). Minimal recursion semantics: an introduction. Journal of Research on Language and Computation, pages 281-332.",
"links": null
},
"BIBREF12": {
"ref_id": "b12",
"title": "The logical form of action sentences",
"authors": [
{
"first": "D",
"middle": [],
"last": "Davidson",
"suffix": ""
}
],
"year": 1979,
"venue": "The Logic of Decision and Action",
"volume": "",
"issue": "",
"pages": "81--120",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Davidson, D. (1979). The logical form of action sentences. In N. Rescher, editor, The Logic of Decision and Action, pages 81-120, Pittsburgh. University of Pittsburgh Press. Reprinted in Davidson (2001).",
"links": null
},
"BIBREF13": {
"ref_id": "b13",
"title": "Essays on Actions and Events",
"authors": [
{
"first": "D",
"middle": [],
"last": "Davidson",
"suffix": ""
}
],
"year": 2001,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Davidson, D. (2001). Essays on Actions and Events. Ox- ford University Press, Oxford, 2nd edition.",
"links": null
},
"BIBREF14": {
"ref_id": "b14",
"title": "Interfacing language, spatial perception and cognition in type theory with records",
"authors": [
{
"first": "S",
"middle": [],
"last": "Dobnik",
"suffix": ""
},
{
"first": "R",
"middle": [],
"last": "Cooper",
"suffix": ""
}
],
"year": 2017,
"venue": "Journal of Language Modelling",
"volume": "5",
"issue": "2",
"pages": "273--301",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Dobnik, S. and Cooper, R. (2017). Interfacing language, spatial perception and cognition in type theory with records. Journal of Language Modelling, 5(2):273-301.",
"links": null
},
"BIBREF15": {
"ref_id": "b15",
"title": "Modelling language, action, and perception in type theory with records",
"authors": [
{
"first": "S",
"middle": [],
"last": "Dobnik",
"suffix": ""
},
{
"first": "R",
"middle": [],
"last": "Cooper",
"suffix": ""
},
{
"first": "S",
"middle": [],
"last": "Larsson",
"suffix": ""
}
],
"year": 2012,
"venue": "Constraint Solving and Language Processing -7th International Workshop on Constraint Solving and Language Processing",
"volume": "2012",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Dobnik, S., Cooper, R., and Larsson, S. (2012). Modelling language, action, and perception in type theory with records. In D. Duchier et al., editors, Constraint Solving and Language Processing -7th International Workshop on Constraint Solving and Language Processing, CSLP 2012, Orelans, France, September. Revised Selected Pa- pers, number 8114 in Publications on Logic, Language and Information (FoLLI), Springer, Berlin, Heidelberg, 2013.",
"links": null
},
"BIBREF17": {
"ref_id": "b17",
"title": "Word Meaning and Montague Grammar: The Semantics of Verbs and Times in Generative Semantics and in Montague's PTQ. D. Reidel",
"authors": [
{
"first": "D",
"middle": [
"R"
],
"last": "Dowty",
"suffix": ""
}
],
"year": 1979,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Dowty, D. R. (1979). Word Meaning and Montague Gram- mar: The Semantics of Verbs and Times in Generative Semantics and in Montague's PTQ. D. Reidel, Dor- drecht.",
"links": null
},
"BIBREF18": {
"ref_id": "b18",
"title": "A Formal Theory of Common Sense Psychology: How People Think People Think",
"authors": [
{
"first": "A",
"middle": [
"S"
],
"last": "Gordon",
"suffix": ""
},
{
"first": "J",
"middle": [
"R"
],
"last": "Hobbs",
"suffix": ""
}
],
"year": 2017,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Gordon, A. S. and Hobbs, J. R. (2017). A Formal Theory of Common Sense Psychology: How People Think People Think. Cambridge University Press, Cambridge.",
"links": null
},
"BIBREF19": {
"ref_id": "b19",
"title": "A Computational Model of Natural Language Communication: Interpretation, Inference, and Production in Database Semantics",
"authors": [
{
"first": "R",
"middle": [],
"last": "Hausser",
"suffix": ""
}
],
"year": 2006,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Hausser, R. (2006). A Computational Model of Nat- ural Language Communication: Interpretation, Infer- ence, and Production in Database Semantics. Springer, Berlin.",
"links": null
},
"BIBREF20": {
"ref_id": "b20",
"title": "From montague grammar to database semantics",
"authors": [
{
"first": "R",
"middle": [],
"last": "Hausser",
"suffix": ""
}
],
"year": 2015,
"venue": "Language and Information",
"volume": "19",
"issue": "2",
"pages": "1--16",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Hausser, R. (2015). From montague grammar to database semantics. Language and Information, 19(2):1-16. available at lagrammar.net.",
"links": null
},
"BIBREF21": {
"ref_id": "b21",
"title": "Annotating and Modeling Shallow Semantics Directly from Text. Dissertation of doctor of philosophy in computer science and engineering",
"authors": [
{
"first": "L",
"middle": [],
"last": "He",
"suffix": ""
}
],
"year": 2018,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "He, L. (2018). Annotating and Modeling Shallow Seman- tics Directly from Text. Dissertation of doctor of philos- ophy in computer science and engineering, University of Washington.",
"links": null
},
"BIBREF22": {
"ref_id": "b22",
"title": "Annotating and reasoning about time and events",
"authors": [
{
"first": "J",
"middle": [],
"last": "Hobbs",
"suffix": ""
},
{
"first": "J",
"middle": [],
"last": "Pustejovsky",
"suffix": ""
}
],
"year": 2003,
"venue": "Proceedings of AAAI Spring Symposium on Logical Formalizations of Common Sense Reasoning",
"volume": "",
"issue": "",
"pages": "301--315",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Hobbs, J. and Pustejovsky, J. (2003). Annotating and rea- soning about time and events. In Proceedings of AAAI Spring Symposium on Logical Formalizations of Com- mon Sense Reasoning, Stanford, CA, March. Reprinted in Mani et al. (eds), 2005, pages 301-315.",
"links": null
},
"BIBREF23": {
"ref_id": "b23",
"title": "ISO 24617-1 Language resource management -Semantic annotation framework -Part 1: Time and events. International Organization for Standardization, Geneva. Working group: ISO/TC 37/SC 4/WG 2 semantic annotation",
"authors": [],
"year": 2012,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "ISO, (2012). ISO 24617-1 Language resource management -Semantic annotation framework -Part 1: Time and events. International Organization for Standardization, Geneva. Working group: ISO/TC 37/SC 4/WG 2 seman- tic annotation.",
"links": null
},
"BIBREF24": {
"ref_id": "b24",
"title": "ISO 24617-4 Language resource management -Semantic annotation framework -Part 4: Semantic roles (SemAF-SR). International Organization for Standardization",
"authors": [],
"year": 2014,
"venue": "ISO",
"volume": "37",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "ISO, (2014). ISO 24617-4 Language resource management -Semantic annotation framework -Part 4: Semantic roles (SemAF-SR). International Organization for Stan- dardization, Geneva. Working group: ISO/TC 37/SC",
"links": null
},
"BIBREF25": {
"ref_id": "b25",
"title": "/WG 2 semantic annotation",
"authors": [],
"year": null,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "/WG 2 semantic annotation.",
"links": null
},
"BIBREF26": {
"ref_id": "b26",
"title": "ISO 24617-7 Language resource management -Semantic annotation framework -Part 7: Spatial information. International Organization for Standardization",
"authors": [],
"year": 2020,
"venue": "ISO",
"volume": "37",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "ISO, (2020). ISO 24617-7 Language resource management -Semantic annotation framework -Part 7: Spatial infor- mation. International Organization for Standardization, Geneva, 2nd edition. Working group: ISO/TC 37/SC",
"links": null
},
"BIBREF27": {
"ref_id": "b27",
"title": "/WG 2 semantic annotation",
"authors": [],
"year": null,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "/WG 2 semantic annotation.",
"links": null
},
"BIBREF28": {
"ref_id": "b28",
"title": "From Discourse to Logic: Introduction to Modeltheoretic Semantics of Natural Language, Formal Logic and Discourse Representation Theory",
"authors": [
{
"first": "H",
"middle": [],
"last": "Kamp",
"suffix": ""
},
{
"first": "U",
"middle": [],
"last": "Reyle",
"suffix": ""
}
],
"year": 1993,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Kamp, H. and Reyle, U. (1993). From Discourse to Logic: Introduction to Modeltheoretic Semantics of Nat- ural Language, Formal Logic and Discourse Represen- tation Theory. Kluwer Academic Publishers, Dordrecht.",
"links": null
},
"BIBREF29": {
"ref_id": "b29",
"title": "Towards a denotational semantics for TimeML",
"authors": [
{
"first": "G",
"middle": [],
"last": "Katz",
"suffix": ""
}
],
"year": 2007,
"venue": "Annotating, Extracting and Reasoning about Time and Events",
"volume": "",
"issue": "",
"pages": "88--106",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Katz, G. (2007). Towards a denotational semantics for TimeML. In Frank Schilder, et al., editors, Annotating, Extracting and Reasoning about Time and Events, pages 88-106, Berlin. Springer.",
"links": null
},
"BIBREF30": {
"ref_id": "b30",
"title": "On the semantics of locatives",
"authors": [
{
"first": "M",
"middle": [],
"last": "Kracht",
"suffix": ""
}
],
"year": 2002,
"venue": "Linguistics and Philosophy",
"volume": "25",
"issue": "",
"pages": "157--232",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Kracht, M. (2002). On the semantics of locatives. Linguis- tics and Philosophy, 25:157-232.",
"links": null
},
"BIBREF31": {
"ref_id": "b31",
"title": "Revising ISO-Space and the role of the movement link",
"authors": [
{
"first": "K",
"middle": [],
"last": "Lee",
"suffix": ""
},
{
"first": "J",
"middle": [],
"last": "Pustejovsky",
"suffix": ""
},
{
"first": "H",
"middle": [],
"last": "Bunt",
"suffix": ""
}
],
"year": 2018,
"venue": "Proceedings of the 14th Joint ACL-ISO Workshop on Interoperable Semantic Annotation (ISA-14): COLING 2018 Workshop",
"volume": "",
"issue": "",
"pages": "35--44",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Lee, K., Pustejovsky, J., and Bunt, H. (2018). Revising ISO-Space and the role of the movement link. In Harry Bunt, editor, Proceedings of the 14th Joint ACL-ISO Workshop on Interoperable Semantic Annotation (ISA- 14): COLING 2018 Workshop, pages 35-44, Santa Fe, New Mexico, U.S.A, August.",
"links": null
},
"BIBREF32": {
"ref_id": "b32",
"title": "Equation solving",
"authors": [
{
"first": "K",
"middle": [],
"last": "Lee",
"suffix": ""
}
],
"year": 1983,
"venue": "Language, Information and Computation",
"volume": "",
"issue": "",
"pages": "14--26",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Lee, K. (1983). Equation solving. In Chungmin Lee et al., editors, Language, Information and Computation, pages 14-26. Taehaksa, Seoul.",
"links": null
},
"BIBREF33": {
"ref_id": "b33",
"title": "Formal semantics for interpreting temporal annotation",
"authors": [
{
"first": "K",
"middle": [],
"last": "Lee",
"suffix": ""
}
],
"year": 2008,
"venue": "Unity and Diversity of Languages",
"volume": "",
"issue": "",
"pages": "97--108",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Lee, K. (2008). Formal semantics for interpreting tem- poral annotation. In Piet van Sterkenburg, editor, Unity and Diversity of Languages, pages 97-108, Amsterdam. John Benjamins Publishing Co. Invited talk at the 18th Congress of Linguists, held in Seoul on July 21-26 2008.",
"links": null
},
"BIBREF34": {
"ref_id": "b34",
"title": "A compositional interval semantics for temporal annotation",
"authors": [
{
"first": "K",
"middle": [],
"last": "Lee",
"suffix": ""
}
],
"year": 2011,
"venue": "Recent Trends in Language an Knowledge Processing",
"volume": "",
"issue": "",
"pages": "122--156",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Lee, K. (2011). A compositional interval semantics for temporal annotation. In Eunryoung Lee et al., editors, Recent Trends in Language an Knowledge Processing, pages 122-156. Hankookmunhwasa, Seoul.",
"links": null
},
"BIBREF35": {
"ref_id": "b35",
"title": "An abstract syntax for ISO-Space with its <moveLink> reformulated",
"authors": [
{
"first": "K",
"middle": [],
"last": "Lee",
"suffix": ""
}
],
"year": 2016,
"venue": "Proceedings of the LREC 2016 Workshop ISA-12 -12th Joint ACL-ISO Workshop on Interoperable Semantic Annotation",
"volume": "",
"issue": "",
"pages": "28--37",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Lee, K. (2016). An abstract syntax for ISO-Space with its <moveLink> reformulated. In Harry Bunt, editor, Proceedings of the LREC 2016 Workshop ISA-12 -12th Joint ACL-ISO Workshop on Interoperable Semantic An- notation, pages 28-37, Portoroz, Slovenia, May.",
"links": null
},
"BIBREF36": {
"ref_id": "b36",
"title": "Revising ISO-Space for the semantic annotation of dynamic spatial information in language. Language and Information",
"authors": [
{
"first": "K",
"middle": [],
"last": "Lee",
"suffix": ""
}
],
"year": 2018,
"venue": "",
"volume": "22",
"issue": "",
"pages": "221--245",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Lee, K. (2018). Revising ISO-Space for the semantic annotation of dynamic spatial information in language. Language and Information, 22.1:221-245.",
"links": null
},
"BIBREF37": {
"ref_id": "b37",
"title": "Algebraic Semantics in Language and Philosophy",
"authors": [
{
"first": "G",
"middle": [],
"last": "Link",
"suffix": ""
}
],
"year": 1998,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Link, G. (1998). Algebraic Semantics in Language and Philosophy. CSLI Publications, Stanford, CA.",
"links": null
},
"BIBREF38": {
"ref_id": "b38",
"title": "Interpreting Motion: Grounded Representations for Spatial Language",
"authors": [
{
"first": "I",
"middle": [],
"last": "Mani",
"suffix": ""
},
{
"first": "J",
"middle": [],
"last": "Pustejovsky",
"suffix": ""
}
],
"year": 2012,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Mani, I. and Pustejovsky, J. (2012). Interpreting Motion: Grounded Representations for Spatial Language. Ox- ford University Press, Oxford.",
"links": null
},
"BIBREF39": {
"ref_id": "b39",
"title": "The event-calculus in classical logic -alternative axiomatizations",
"authors": [
{
"first": "R",
"middle": [],
"last": "Miller",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Shanahan",
"suffix": ""
}
],
"year": 1999,
"venue": "Electronic Transactions on Artificial Intelligence",
"volume": "3",
"issue": "1",
"pages": "77--105",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Miller, R. and Shanahan, M. (1999). The event-calculus in classical logic -alternative axiomatizations. Electronic Transactions on Artificial Intelligence, 3(1):77-105.",
"links": null
},
"BIBREF40": {
"ref_id": "b40",
"title": "Temporal ontology and temporal reference",
"authors": [
{
"first": "M",
"middle": [],
"last": "Moens",
"suffix": ""
},
{
"first": "M",
"middle": [],
"last": "Steedman",
"suffix": ""
}
],
"year": 1988,
"venue": "Computational Linguistics",
"volume": "14",
"issue": "2",
"pages": "15--28",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Moens, M. and Steedman, M. (1988). Temporal ontol- ogy and temporal reference. Computational Linguistics, 14(2):15-28.",
"links": null
},
"BIBREF41": {
"ref_id": "b41",
"title": "Formal Philosophy: Selected Papers of Richard Montague",
"authors": [
{
"first": "R",
"middle": [],
"last": "Montague",
"suffix": ""
}
],
"year": 1974,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Montague, R. (1974). Formal Philosophy: Selected Papers of Richard Montague. Yale University Press, New Haven and London.",
"links": null
},
"BIBREF42": {
"ref_id": "b42",
"title": "Events in the Semantics of English: A Study in Subatomic Semantics",
"authors": [
{
"first": "T",
"middle": [],
"last": "Parsons",
"suffix": ""
}
],
"year": 1990,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Parsons, T. (1990). Events in the Semantics of English: A Study in Subatomic Semantics. The MIT Press, Cam- bridge, MA.",
"links": null
},
"BIBREF43": {
"ref_id": "b43",
"title": "Some structural analogies between tenses and pronouns in English",
"authors": [
{
"first": "B",
"middle": [
"H"
],
"last": "Partee",
"suffix": ""
}
],
"year": 1973,
"venue": "Reprinted in Compositionality in Formal Semantics: Selected Papers by Barbara H. Partee",
"volume": "80",
"issue": "",
"pages": "50--58",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Partee, B. H. (1973). Some structural analogies between tenses and pronouns in English. The Journal of Philos- ophy, 80(18):601-9. Reprinted in Compositionality in Formal Semantics: Selected Papers by Barbara H. Par- tee, Malden, MA: Blackwell. pp. 50-58.",
"links": null
},
"BIBREF44": {
"ref_id": "b44",
"title": "From TimeML to interval temporal logic",
"authors": [
{
"first": "I",
"middle": [],
"last": "Pratt-Hartmann",
"suffix": ""
}
],
"year": 2007,
"venue": "Proceedings of the Seventh International Workshop on Computational Semantics",
"volume": "",
"issue": "",
"pages": "111--180",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Pratt-Hartmann, I. (2007). From TimeML to interval temporal logic. In Harry Bunt, editor, Proceedings of the Seventh International Workshop on Computational Semantics, pages 111-180, Tilburg, the Netherlands. Tilburg University.",
"links": null
},
"BIBREF45": {
"ref_id": "b45",
"title": "The specification language TimeML",
"authors": [
{
"first": "J",
"middle": [],
"last": "Pustejovsky",
"suffix": ""
},
{
"first": "R",
"middle": [],
"last": "Ingria",
"suffix": ""
},
{
"first": "R",
"middle": [],
"last": "Saur\u00ed",
"suffix": ""
},
{
"first": "J",
"middle": [
"C"
],
"last": "Littman",
"suffix": ""
},
{
"first": "J",
"middle": [],
"last": "Gaizauskas",
"suffix": ""
},
{
"first": "R",
"middle": [],
"last": "Setzer",
"suffix": ""
},
{
"first": "A",
"middle": [],
"last": "Katz",
"suffix": ""
},
{
"first": "G",
"middle": [],
"last": "Mani",
"suffix": ""
},
{
"first": "I",
"middle": [],
"last": "",
"suffix": ""
}
],
"year": 2005,
"venue": "The Language of Time",
"volume": "",
"issue": "",
"pages": "545--557",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Pustejovsky, J., Ingria, R., Saur\u00ed, R., o, J. C., Littman, J., Gaizauskas, R., Setzer, A., Katz, G., and Mani, I. (2005). The specification language TimeML. In James Puste- jovsky Inderjeet Mani et al., editors, The Language of Time, pages 545-557. Oxford University Press, Oxford.",
"links": null
},
"BIBREF46": {
"ref_id": "b46",
"title": "ISO-TimeML: An international standard for semantic annotation",
"authors": [
{
"first": "J",
"middle": [],
"last": "Pustejovsky",
"suffix": ""
},
{
"first": "K",
"middle": [],
"last": "Lee",
"suffix": ""
},
{
"first": "H",
"middle": [],
"last": "Bunt",
"suffix": ""
},
{
"first": "L",
"middle": [],
"last": "Romary",
"suffix": ""
}
],
"year": 2010,
"venue": "Proceedings of LREC 2010",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Pustejovsky, J., Lee, K., Bunt, H., and Romary, L. (2010). ISO-TimeML: An international standard for semantic annotation. In Harry Bunt, editor, Proceedings of LREC 2010, Valletta, Malta, May. LREC 2010.",
"links": null
},
"BIBREF47": {
"ref_id": "b47",
"title": "The semantics of ISO-Space",
"authors": [
{
"first": "J",
"middle": [],
"last": "Pustejovsky",
"suffix": ""
},
{
"first": "K",
"middle": [],
"last": "Lee",
"suffix": ""
},
{
"first": "H",
"middle": [],
"last": "Bunt",
"suffix": ""
}
],
"year": 2019,
"venue": "Proceedings of the 15th Joint ACL -ISO Workshop on Interoperable Semantic Annotation (ISA-15)",
"volume": "",
"issue": "",
"pages": "46--53",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Pustejovsky, J., Lee, K., and Bunt, H. (2019). The seman- tics of ISO-Space. In Harry Bunt, editor, Proceedings of the 15th Joint ACL -ISO Workshop on Interoperable Se- mantic Annotation (ISA-15), pages 46-53, Gothenburg, Sweden, May. International Workshop on Computational Semantics (IWCS 2029).",
"links": null
},
"BIBREF48": {
"ref_id": "b48",
"title": "The Generative Lexicon",
"authors": [
{
"first": "J",
"middle": [],
"last": "Pustejovsky",
"suffix": ""
}
],
"year": 1995,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Pustejovsky, J. (1995). The Generative Lexicon. The MIT Press, Cambridge, MA.",
"links": null
},
"BIBREF49": {
"ref_id": "b49",
"title": "Type construction and the logic of concepts",
"authors": [
{
"first": "J",
"middle": [],
"last": "Pustejovsky",
"suffix": ""
}
],
"year": 2001,
"venue": "The Language of Word Meaning",
"volume": "",
"issue": "",
"pages": "91--135",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Pustejovsky, J. (2001). Type construction and the logic of concepts. In Pierrette Bouillon et al., editors, The Lan- guage of Word Meaning, pages 91-135. Cambridge Uni- versity Press, Cambridge, UK.",
"links": null
},
"BIBREF50": {
"ref_id": "b50",
"title": "A Comprehensive Grammar of the English Language",
"authors": [
{
"first": "R",
"middle": [],
"last": "Quirk",
"suffix": ""
},
{
"first": "S",
"middle": [],
"last": "Greenbaum",
"suffix": ""
},
{
"first": "G",
"middle": [],
"last": "Leech",
"suffix": ""
},
{
"first": "J",
"middle": [],
"last": "Svartvik",
"suffix": ""
}
],
"year": 1985,
"venue": "",
"volume": "",
"issue": "",
"pages": "",
"other_ids": {},
"num": null,
"urls": [],
"raw_text": "Quirk, R., Greenbaum, S., Leech, G., and Svartvik, J. (1985). A Comprehensive Grammar of the English Lan- guage. Longman, London and New York, January.",
"links": null
}
},
"ref_entries": {
"FIGREF0": {
"text": "metamodel of ABS",
"uris": null,
"num": null,
"type_str": "figure"
},
"FIGREF1": {
"text": "Model-theoretic Interpretation The symbol [[ ]] is used to represent a (model-theoretic) denotation. Given any semantic form \u03c3(a) in ABSr, its denotation with respect to a model M , an assignment g of values to variables, and a set D of definitions for logical predicates is represented by [[\u03c3(a)]] M,g,D .",
"uris": null,
"num": null,
"type_str": "figure"
},
"FIGREF2": {
"text": "Structures: entity(x1, w1, type:dog, form:nam) event(e1, w2-3, pred:bark, tense:present, aspect:progressive) c. Link Structure: srlink(e1, x1, agent) The annotation of text fragment (9a) consists of a list of entity structures in (b) and a link structure (c) over them. Here, srlink specifies the semantic role of the participant x 1 as an agent participating in the event e 1 of barking, as illustrated in (10).",
"uris": null,
"num": null,
"type_str": "figure"
},
"FIGREF3": {
"text": "t c. Semantic form of annotation structure: \u03c3(a 9 ) := {x:e, e:v}\u03c3(srlink) := {x:e, e:v} [[dog(x) \u2227 named(x, F ido)] \u2227 [bark(e) \u2227 presProg(e)] \u2227 agent(e, x)] by Variable renaming and binding",
"uris": null,
"num": null,
"type_str": "figure"
},
"FIGREF4": {
"text": "\u2192 t), := {e 1 , e 2 }[want(e 1 ) \u2227 theme(e 1 , e 2 )] \u03c3(e2) e\u2192(E\u2192t) := {x 1 , e 2 }[teach(e 2 ) \u2227 agent(e 2 , x 1 )] b. Semantic form of the subordination link structure: \u03c3(slink) t := {x 1 , e 1 , e 2 }[{\u03c3(e 1 ) E , \u03c3(e 2 ) e\u2192(E\u2192t) ] i (\u03c3(e 1 ), \u03c3(e 2 )) (e\u2192(E\u2192t))\u2192(E\u2192t) ] := {x 1 , e 1 , e 2 }[\u03c3(e 1 ) t \u2192 int \u03c3(e 2 ) t ] := {x 1 , e 1 , e 2 }[[want(e 1 ) \u2227 theme(e 1 , e 2 )] \u2192 i ([go(e 2 ) \u2227 agent(e 2 , x 1 )])]c. Semantic form of the whole annotation structure: \u03c3(a 21 ) := \u03c3(slink) t",
"uris": null,
"num": null,
"type_str": "figure"
},
"FIGREF5": {
"text": "ABS makes use of logical predicates as part of the (object) representation language to simplify the representation of semantic forms or make it flexible to accommodate different interpretations. These predicates, marked in boldface, in ABSr are defined possibly in terms of higher-order logic as part of the model structure. The predicate past is, for instance, introduced to represent the tense of an event as in (24): (24) a. [walk(e) \u2227 past(e)] b. instead of [walk(e) \u2227 e \u2286 t \u2227 t \u227a n]",
"uris": null,
"num": null,
"type_str": "figure"
},
"FIGREF6": {
"text": "Aspectual features such as present perfect and progressive are also encoded into annotations just as they are. Consider a case of the present perfect aspect in (26). (26) a. Mia [has visited] e1 Boston. b. Annotation (id=a 26 ): event (e1, w2-3, pred:visit, tense:present, aspect: perfect) c. Semantic Form: \u03c3(e1) := [visit(e 1 ) \u2227 presPerfect(e 1 )]",
"uris": null,
"num": null,
"type_str": "figure"
},
"FIGREF7": {
"text": "a. The plant has died. b. {a, e, t, n, r, s, u} e s, u) ABSr , in contrast, yields the following representation: (28) a. The plant has died. b. Annotation: entity(x1, w2, type:plant) event(e1, w4, pred:die, tense:present, aspect:perfct) srlink(e1,x1, theme) c. Semantic Forms: \u03c3(x 1 ) := plant(x 1 ) \u03c3(e 1 ) := [die(e 1 )\u2227 presPerfect(e 1 )] \u03c3(srlink) := [{\u03c3(x 1 ) t , \u03c3(e 1 ) t } \u2295 bo theme(e 1 , x 1 ) t ] \u03c3( 26 ) := {e, x}[die(e)\u2227 presPerfect(e) \u2227 theme(e, x)]",
"uris": null,
"num": null,
"type_str": "figure"
},
"FIGREF8": {
"text": "Options: Type-theoretic or Not",
"uris": null,
"num": null,
"type_str": "figure"
},
"TABREF1": {
"html": null,
"text": "Rule 1 bo Boolean conjunctive composition (\u2295 bo ) Rule 1 fa Functional conjunctive composition (\u2295 fa )",
"num": null,
"type_str": "table",
"content": "<table><tr><td>Rule 1 sub Substitutive conjunctive composition</td></tr><tr><td>by substitution (\u2295 sub )</td></tr><tr><td>Rule 1 eq Equative conjunctive composition</td></tr><tr><td>by equation solving (\u2295 eq )</td></tr><tr><td>Disjunctive composition ( ):</td></tr><tr><td>Rule 2 Disjunctive composition ( )</td></tr><tr><td>Rule 2 int Intensional disjunctive composition</td></tr><tr><td>( int )</td></tr><tr><td>Rule 2 imp Implicational disjunctive composition</td></tr><tr><td>( imp )</td></tr><tr><td>Rule 1 bo Boolean conjunctive composition (\u2295 bo ) is the</td></tr><tr><td>most common type of composition, as formulated in</td></tr><tr><td>(8) Rule 1</td></tr></table>"
},
"TABREF2": {
"html": null,
"text": "Rules 1 sub and 1 eq , subtypes of conjunctive composition, are needed when one of the inputs to links is treated as of some basic or pseudo basic type. Consider the same example (9) but with a different semantic treatment: 7 (14) a. \u03c3(x1) e := f ido e \u03c3(e1) v\u2192t := {e 1 :v}[bark(e 1 )\u2227 presProg(e 1 )]",
"num": null,
"type_str": "table",
"content": "<table><tr><td>b. \u03c3(srlink3)</td></tr><tr><td>:= {e 1 :v}</td></tr><tr><td>[{\u03c3(x1) e , \u03c3(e1) v\u2192t } \u2295 sub</td></tr><tr><td>agent(e 1 , x 1 ) (v\u2192t)\u2192(e\u2192t) ]</td></tr><tr><td>:= {e 1 :v}</td></tr><tr><td>[\u03c3(e1) t \u2227 agent(e 1 , f ido) t ]</td></tr><tr><td>:= {e 1 :v}</td></tr><tr><td>[[bark(e 1 ) \u2227 presProg(e 1 )] t \u2227</td></tr><tr><td>agent(e 1 , f ido)]</td></tr><tr><td>d. \u03c3(a 9 ) := \u03c3(srlink4)</td></tr><tr><td>Now by the rule of substitution of identicals in FOL, we</td></tr><tr><td>have:</td></tr><tr><td>(16) {e 1 :v}</td></tr><tr><td>[[bark(e</td></tr></table>"
}
}
}
}