Base_JSON/prefixI/json/I08/I08-1013.json

{
    "paper_id": "I08-1013",
    "header": {
        "generated_with": "S2ORC 1.0.0",
        "date_generated": "2023-01-19T07:41:49.177141Z"
    },
    "title": "An Effective Compositional Model for Lexical Alignment",
    "authors": [
        {
            "first": "B\u00e9atrice",
            "middle": [],
            "last": "Daille",
            "suffix": "",
            "affiliation": {
                "laboratory": "",
                "institution": "LINA -FRE CNRS",
                "location": {
                    "addrLine": "2729 2, rue de la Houssini\u00e8re, BP 92208",
                    "postCode": "F-44322",
                    "settlement": "Nantes cedex 03 beatrice.daille"
                }
            },
            "email": ""
        },
        {
            "first": "Emmanuel",
            "middle": [],
            "last": "Morin",
            "suffix": "",
            "affiliation": {
                "laboratory": "",
                "institution": "LINA -FRE CNRS",
                "location": {
                    "addrLine": "2729 2, rue de la Houssini\u00e8re, BP 92208",
                    "postCode": "F-44322",
                    "settlement": "Nantes cedex 03 beatrice.daille"
                }
            },
            "email": "emmanuel.morin\u00a1@univ-nantes.fr"
        }
    ],
    "year": "",
    "venue": null,
    "identifiers": {},
    "abstract": "The automatic compilation of bilingual dictionaries from comparable corpora has been successful for single-word terms (SWTs), but remains disappointing for multi-word terms (MWTs). One of the main problems is the insufficient coverage of the bilingual dictionary. Using the compositional translation method improved the results, but still shows some limits for MWTs of different syntactic structures. In this paper, we propose to bridge the gap between syntactic structures through morphological links. The results show a significant improvement in the compositional translation of MWTs that demonstrate the efficiency of the morphologically based-method for lexical alignment.",
    "pdf_parse": {
        "paper_id": "I08-1013",
        "_pdf_hash": "",
        "abstract": [
            {
                "text": "The automatic compilation of bilingual dictionaries from comparable corpora has been successful for single-word terms (SWTs), but remains disappointing for multi-word terms (MWTs). One of the main problems is the insufficient coverage of the bilingual dictionary. Using the compositional translation method improved the results, but still shows some limits for MWTs of different syntactic structures. In this paper, we propose to bridge the gap between syntactic structures through morphological links. The results show a significant improvement in the compositional translation of MWTs that demonstrate the efficiency of the morphologically based-method for lexical alignment.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Abstract",
                "sec_num": null
            }
        ],
        "body_text": [
            {
                "text": "Current research in the automatic compilation of bilingual dictionaries from corpora uses of comparable corpora. Comparable corpora gather texts sharing common features (domain, topic, genre, discourse) without having a source text-target text relationship. They are considered by human translators more trustworthy than parallel corpora (Bowker and Pearson, 2002) . Moreover, they are available for any written languages and not only for pairs of languages involving English. The compilation of specialized dictionaries should take into account multiword terms (MWTs) that are more precise and specific to a particular scientific domain than singleword terms (SWTs). The standard approach is based on lexical context analysis and relies on the simple observation that a SWT or a MWT and its translation tend to appear in the same lexical contexts. Correct results are obtained for SWTs with an accuracy of about 80% for the top 10-20 proposed candidates using large comparable corpora (Fung, 1998; Rapp, 1999; Chiao and Zweigenbaum, 2002) or 60% using small comparable corpora . In comparison, the results obtained for MWTs are disappointing. For instance, (Morin et al., 2007) have achieved 30% and 42% precision for the top 10 and top 20 candidates in a 0.84 million-word French-Japanese corpus. These results could be explained by the low frequency of MWTs compared to SWTs, by the lack of parallelism between the source and the target MWT extraction systems, and by the low performance of the alignment program. For SWTs, the process is in two steps: looking in a dictionary, and if no direct translation is available, starting the contextual analysis. Looking in the dictionary gives low results for MWTs: 1% compared to 30% for French and 20% for Japanese SWTs (Morin and Daille, 2006) . To extend the coverage of the bilingual dictionary, an intermediate step is added between looking in the dictionary and the contextual analysis that will propose several translation candidates to compare with the target MWTs. These candidate translations are obtained thanks to a compositional translation method (Melamed, 1997; Grefenstette, 1999) . This method reveals some limits when MWTs in the source and the target languages do not share the same syntactic patterns.",
                "cite_spans": [
                    {
                        "start": 338,
                        "end": 364,
                        "text": "(Bowker and Pearson, 2002)",
                        "ref_id": "BIBREF1"
                    },
                    {
                        "start": 986,
                        "end": 998,
                        "text": "(Fung, 1998;",
                        "ref_id": "BIBREF12"
                    },
                    {
                        "start": 999,
                        "end": 1010,
                        "text": "Rapp, 1999;",
                        "ref_id": "BIBREF21"
                    },
                    {
                        "start": 1011,
                        "end": 1039,
                        "text": "Chiao and Zweigenbaum, 2002)",
                        "ref_id": "BIBREF3"
                    },
                    {
                        "start": 1158,
                        "end": 1178,
                        "text": "(Morin et al., 2007)",
                        "ref_id": "BIBREF19"
                    },
                    {
                        "start": 1768,
                        "end": 1792,
                        "text": "(Morin and Daille, 2006)",
                        "ref_id": "BIBREF18"
                    },
                    {
                        "start": 2108,
                        "end": 2123,
                        "text": "(Melamed, 1997;",
                        "ref_id": "BIBREF15"
                    },
                    {
                        "start": 2124,
                        "end": 2143,
                        "text": "Grefenstette, 1999)",
                        "ref_id": "BIBREF13"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Introduction",
                "sec_num": "1"
            },
            {
                "text": "In this paper, we put forward an extended compo-sitional method that bridges the gap between MWTs of different syntactic structures through morphological links. We experiment within this method of French-Japanese lexical alignment, using multilingual terminology mining chain made up of two terminology extraction systems; one in each language, and an alignment program. The term extraction systems are publicly available and both extract MWTs. The alignment program makes use of the direct context-vector approach (Fung, 1998; Rapp, 1999) . The results show an improvement of 33% in the translation of MWTs that demonstrate the efficiency of the morphologically based-method for lexical alignment.",
                "cite_spans": [
                    {
                        "start": 515,
                        "end": 527,
                        "text": "(Fung, 1998;",
                        "ref_id": "BIBREF12"
                    },
                    {
                        "start": 528,
                        "end": 539,
                        "text": "Rapp, 1999)",
                        "ref_id": "BIBREF21"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Introduction",
                "sec_num": "1"
            },
            {
                "text": "Taking a comparable corpora as input, the multilingual terminology mining chain outputs a list of single-and multi-word candidate terms along with their candidate translations (see Figure 1 ). This chain performs a contextual analysis that adapts the direct context-vector approach (Rapp, 1995; Fung and McKeown, 1997) for SWTs to MWTs. It consists of the following five steps:",
                "cite_spans": [
                    {
                        "start": 282,
                        "end": 294,
                        "text": "(Rapp, 1995;",
                        "ref_id": "BIBREF20"
                    },
                    {
                        "start": 295,
                        "end": 318,
                        "text": "Fung and McKeown, 1997)",
                        "ref_id": "BIBREF11"
                    }
                ],
                "ref_spans": [
                    {
                        "start": 181,
                        "end": 189,
                        "text": "Figure 1",
                        "ref_id": "FIGREF0"
                    }
                ],
                "eq_spans": [],
                "section": "Multilingual terminology mining chain",
                "sec_num": "2"
            },
            {
                "text": "1. For each language, the documents are cleaned, tokenized, tagged and lemmatized. For French, Brill's POS tagger 1 and the FLEM lemmatiser 2 are used, and for Japanese, ChaSen 3 . We then extract the MWTs and their variations using the ACABIT terminology extraction system available for French 4 (Daille, 2003), English and Japanese 5 (Takeuchi et al., 2004) . (From now on, we will refer to lexical units as words, SWTs or MWTs).",
                "cite_spans": [
                    {
                        "start": 336,
                        "end": 359,
                        "text": "(Takeuchi et al., 2004)",
                        "ref_id": "BIBREF24"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Multilingual terminology mining chain",
                "sec_num": "2"
            },
            {
                "text": "2. We collect all the lexical units in the context of each lexical unit \u00a2 and count their occurrence frequency in a window of \u00a3 words around \u00a2 . For each lexical unit \u00a2 of the source and the target languages, we obtain a context vector 1 http://www.atilf.fr/winbrill/ 2 http://www.univ-nancy2.fr/pers/namer/ 3 http://chasen-legacy.sourceforge.jp/ 4 http://www.sciences.univ-nantes.fr/ info/perso/permanents/daille/ and release for Mandriva Linux.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Multilingual terminology mining chain",
                "sec_num": "2"
            },
            {
                "text": "5 http://cl.cs.okayama-u.ac.jp/rsc/ jacabit/ \u00a4 \u00a6 \u00a5 which gathers the set of co-occurrence units \u00a7 associated with the number of times that \u00a7 and \u00a2 occur together\u00a6\u00a9 \u00a9 \u00a5 . In order to identify specific words in the lexical context and to reduce word-frequency effects, we normalize context vectors using an association score such as Mutual Information (Fano, 1961) or Log-likelihood (Dunning, 1993) .",
                "cite_spans": [
                    {
                        "start": 350,
                        "end": 362,
                        "text": "(Fano, 1961)",
                        "ref_id": "BIBREF9"
                    },
                    {
                        "start": 381,
                        "end": 396,
                        "text": "(Dunning, 1993)",
                        "ref_id": "BIBREF7"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Multilingual terminology mining chain",
                "sec_num": "2"
            },
            {
                "text": "3. Using a bilingual dictionary, we translate the lexical units of the source context vector. If the bilingual dictionary provides several translations for a lexical unit, we consider all of them but weigh the different translations by their frequency in the target language.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Multilingual terminology mining chain",
                "sec_num": "2"
            },
            {
                "text": "4. For a lexical unit to be translated, we compute the similarity between the translated context vector and all target vectors through vector distance measures such as Cosine (Salton and Lesk, 1968) or Jaccard (Tanimoto, 1958) .",
                "cite_spans": [
                    {
                        "start": 175,
                        "end": 198,
                        "text": "(Salton and Lesk, 1968)",
                        "ref_id": "BIBREF23"
                    },
                    {
                        "start": 210,
                        "end": 226,
                        "text": "(Tanimoto, 1958)",
                        "ref_id": "BIBREF25"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Multilingual terminology mining chain",
                "sec_num": "2"
            },
            {
                "text": "The candidate translations of a lexical unit are the target lexical units closest to the translated context vector according to vector distance.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "5.",
                "sec_num": null
            },
            {
                "text": "In this approach, the translation of the lexical units of the context vectors (step 3 of the previous approach), which depends on the coverage of the bilingual dictionary vis-\u00e0-vis the corpus, is the most important step: the greater the number of elements translated in the context vector, the more discriminating the context vector in selecting translations in the target language. Since the lexical units refer to SWTs and MWTs, the dictionary must contain many entries which occur in the corpus. For SWTs, combining a general bilingual dictionary with a specialized bilingual dictionary or a multilingual thesaurus to translate context vectors ensures that much of their elements will be translated (Chiao and Zweigenbaum, 2002; . For a MWT to be translated, steps 3 to 5 could be avoided thanks to a compositional method that will propose several translation candidates to directly compare with the target MWTs identified in step 1. Moreover, the compositional method is useful in step 3 to compensate for the bilingual dictionary when the multi-word units of the context vector are not directly translated. ",
                "cite_spans": [
                    {
                        "start": 702,
                        "end": 731,
                        "text": "(Chiao and Zweigenbaum, 2002;",
                        "ref_id": "BIBREF3"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "5.",
                "sec_num": null
            },
            {
                "text": "In order to increase the coverage of the dictionary for MWTs that could not be directly translated, we generated possible translations by using a default compositional method (Melamed, 1997; Grefenstette, 1999) . For each element of the MWT found in the bilingual dictionary, we generated all the translated combinations identified by the terminology extraction system. For example, for the French MWT fatigue chronique (chronic fatigue), there are four Japanese translations for fatigue (fatigue) -, , , \" ! -and two translations for chronique (chronic) -",
                "cite_spans": [
                    {
                        "start": 175,
                        "end": 190,
                        "text": "(Melamed, 1997;",
                        "ref_id": "BIBREF15"
                    },
                    {
                        "start": 191,
                        "end": 210,
                        "text": "Grefenstette, 1999)",
                        "ref_id": "BIBREF13"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Default compositional method",
                "sec_num": "3"
            },
            {
                "text": "EQUATION",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [
                    {
                        "start": 0,
                        "end": 8,
                        "text": "EQUATION",
                        "ref_id": "EQREF",
                        "raw_str": "# % $ ' & '",
                        "eq_num": "( , ) 1 0"
                    }
                ],
                "section": "Default compositional method",
                "sec_num": "3"
            },
            {
                "text": ". Next, we generated all possible combinations of the translated elements (see Table 1 6 ) and selected those which refer to an existing MWT in the target language. In the above example, only one term for each element was identified by the Japanese extraction system: ) 2 0 % 2 . In this approach, when it is not possible to translate all parts of an MWT, or when the translated combinations are not identified by the extraction system, the MWT is 6 The French word order is reversed to take into account the different constraints between French and Japanese. not taken into account in the translation step. This approach also differs from that used by (Robitaille et al., 2006) for French-Japanese translation. They first decompose the French MWT into combinations of shorter multi-word unit elements. This approach makes the direct translation of a subpart of the MWT possible if it is present in the bilingual dictionary. For MWTs of length \u00a3 , (Robitaille et al., 2006 ) produce all the combinations of shorter multi-word unit elements of a length less than or equal to \u00a3 . For example, the French MWT syndrome de fatigue chronique (chronic fatigue disorder) yields the following four combinations: i) 6 s yndrome de fatigue chronique7 , ii) 6 s yndrome de fatigue7 8 6 c hronique7 , iii) 6 s yndrome7 9 6 f atigue chronique7 and iv) 6 s yndrome7 6 f atigue7 @ 6 c hronique7 . We limit ourselves to the combination of type iv) above since 90% of the French candidate terms provided by the term extraction process after clustering are only composed of two content words.",
                "cite_spans": [
                    {
                        "start": 448,
                        "end": 449,
                        "text": "6",
                        "ref_id": null
                    },
                    {
                        "start": 653,
                        "end": 678,
                        "text": "(Robitaille et al., 2006)",
                        "ref_id": "BIBREF22"
                    },
                    {
                        "start": 948,
                        "end": 972,
                        "text": "(Robitaille et al., 2006",
                        "ref_id": "BIBREF22"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Default compositional method",
                "sec_num": "3"
            },
            {
                "text": "chronique fatigue # 3 $ & 4 ( ) 0 # 3 $ & 4 ( 3 ) 0 3 # 3 $ & 4 ( ) 0 # 3 $ & 4 ( 5 ! ) 0 5 !",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Default compositional method",
                "sec_num": "3"
            },
            {
                "text": "The compositional translation presents problems which have been reported by (Baldwin and Tanaka, 2004; Brown et al., 1993 The pattern switching problem involves the Adjective/Noun and the Noun/Verb part-of-speech switches. The Adjective/Noun switch commonly involves a relational adjective (ADJR). According to grammatical tradition, there are two main categories among adjectives: epithetic adjectives such as important (significant) and relational adjectives such as sanguin (blood). The former cannot have an agentive interpretation in contrast to the latter: the adjective sanguin (blood) within the MWT acidit\u00e9 sanguine (blood acidity) is an argument to the predicative noun acidit\u00e9 (acidity) and this is not the case for the adjective important (significant) within the noun phrase acidit\u00e9 importante (significant acidity). Such adjectives hold a naming function (Levi, 1978) and are particularly frequent in scientific fields (Daille, 2001) . Relational adjectives are either denominal adjectives, morphologically derived from a noun thanks to a suffix, or adjectives having a noun usage such as math\u00e9matique (mathematical/mathematics). For the former, there are appropriate adjective-forming suffixes for French that lead to relational adjectives such as -ique, -aire, -al. For a noun, it is not possible to guess the adjectiveforming suffix that will be employed as well as the alternation of the noun stem that could occur. Relational adjectives part of a MWT are often translated by a noun whatever the target language is. From French to Japanese, the examples are numerous: prescription m\u00e9dicamenteuse (b c 3 d -medicinal prescription), surveillance glyc\u00e9mique (B 4 e 3 f g -glycemic monitoring), fibre alimentaire (h",
                "cite_spans": [
                    {
                        "start": 76,
                        "end": 102,
                        "text": "(Baldwin and Tanaka, 2004;",
                        "ref_id": "BIBREF0"
                    },
                    {
                        "start": 103,
                        "end": 121,
                        "text": "Brown et al., 1993",
                        "ref_id": "BIBREF2"
                    },
                    {
                        "start": 869,
                        "end": 881,
                        "text": "(Levi, 1978)",
                        "ref_id": "BIBREF14"
                    },
                    {
                        "start": 933,
                        "end": 947,
                        "text": "(Daille, 2001)",
                        "ref_id": "BIBREF4"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Pattern switching",
                "sec_num": "4"
            },
            {
                "text": "% i p 3 q -dietary fibre), produit laitier (r X s I t -dairy product), fonction r\u00e9nale (u 3 v 2 w 1 x -kidney func- tion).",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Pattern switching",
                "sec_num": "4"
            },
            {
                "text": "The problem of fertility could only be solved thanks to a contextual analysis in contrast to the foreign name problem that could be solved by an heuristic. We decided to concentrate on the MWT pattern switching problem.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Pattern switching",
                "sec_num": "4"
            },
            {
                "text": "When it is not possible to directly translate a MWT -i.e. i) before performing the steps 3 to 5 of the contextual analysis for a multi-word term to be translated or ii) during step 3 for the translation of multi-word units of the context vector -, we first try to translate the MWT using the default compositional method. If the default compositional method fails, we use a morphologically-based compositional method. For each MWT of N ADJ structure, we generate candidate MWTs of N Prep N structure thanks to the rewriting rule:",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Morphologically-based compositional method",
                "sec_num": "5"
            },
            {
                "text": "y y 9 d f e g i h j l k y m o n g i h j p q k y m n r 6 t s v u x w z y { k s | u 7 g i h j p q k y p m o n r 6 t s v y l } { u { k s v } 7 g i h j p q k y p m o n r 6 t s v y k 7",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Morphologically-based compositional method",
                "sec_num": "5"
            },
            {
                "text": "(1) g i h j p l k y p m o n gathers a relational adjective p such as glyc\u00e9m-ique and the noun y m from which the adjective has been derived such as glyc\u00e9m-ie thanks to the stripping-recoding rule",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Morphologically-based compositional method",
                "sec_num": "5"
            },
            {
                "text": "6 t s | u w z y { k s v u 7",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Morphologically-based compositional method",
                "sec_num": "5"
            },
            {
                "text": ". We generate all possible forms of y m as matching strippingrecoding rules and keep those that belong to the biligual dictionary such as glyc\u00e9m-ie. Thus, we have created a morphological link between the MWT contr\u00f4le glyc\u00e9mique (glycemic control) of N ADJ structure and multi-word unit (MWU) of N Prep N structure contr\u00f4le de la glyc\u00e9mie (lit. control of glycemia). Since it has not been possible to translate all the parts of the MWT contr\u00f4le glyc\u00e9mique, because glyc\u00e9mique was not found in the dictionary, we use the morpholocally-linked MWU contr\u00f4le de la glyc\u00e9mie of which all the parts are translated. The morpholocally-linked MWU could be seen as a canonical lexical form in the translation process that possibly does not exist in the source language. For instance, if index glyc\u00e9mique (glycemic index) is a French MWT, the MWU index de la glyc\u00e9mie (lit. index of the glycemia) does not appear in the French corpus.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Morphologically-based compositional method",
                "sec_num": "5"
            },
            {
                "text": "The stripping-recoding rules could be manually encoded, mined from a monolingual corpus using a learning method such as (Mikheev, 1997) , or supplied by a source terminology extraction system that handles morphological variations. For such a system, a MWT is a canonical form which merges several synonymic variations. For instance, the French MWT exc\u00e8s pond\u00e9ral (overweight) is the canonical form of the following variants: exc\u00e8s pond\u00e9ral (overweight) of N ADJ structure, exc\u00e8s de poids (overweight) of N PREP N structure. It is this last method that we used for our experiment.",
                "cite_spans": [
                    {
                        "start": 120,
                        "end": 135,
                        "text": "(Mikheev, 1997)",
                        "ref_id": "BIBREF17"
                    }
                ],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Morphologically-based compositional method",
                "sec_num": "5"
            },
            {
                "text": "In this section, we will outline the different linguistic resources used for our experiments. We then evaluate the performance of the default and morphologically-based compositional methods.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Evaluation",
                "sec_num": "6"
            },
            {
                "text": "In order to obtain comparable corpora, we selected the French and Japanese documents from the Web. The documents were taken from the medical domain, within the sub-domain of 'diabetes' and 'nutrition'. Document harvesting was carried out by a domain-based search, then by manual selection. A search for documents sharing the same domain can be achieved using keywords reflecting the specialized domain: for French alimentation, diab\u00e8te and ob\u00e9sit\u00e9 (food, diabetes, and obesity); for Japanese, e 4 and 2 (diabetes, and overweight). Then the documents were manually selected by native speakers of each language who are not domain specialists. These documents (248 for French and 538 for Japanese) were converted into plain text from HTML or PDF, yielding 1.5 million-word corpus (0.7 million-word for French and 0.8 million-word for Japanese).",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Linguistic resources",
                "sec_num": "6.1"
            },
            {
                "text": "The French-Japanese bilingual dictionary used in the translation phase was composed of four dictionaries freely available on the Web (6 dico 17 7 ,",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Linguistic resources",
                "sec_num": "6.1"
            },
            {
                "text": "d ico 27 8 , 6 d ico 37 9 , and 6 d ico 47 10 ), and the French-Japanese Scientific Dictionary 1989 ",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "6",
                "sec_num": null
            },
            {
                "text": "We needed to distinguish between relational and epithetic adjectives appearing among the French N ADJ candidates to demonstrate the relevance of the morphological links. To build two French N ADJ reference lists, we proceeded as follows:",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "French N ADJ reference lists",
                "sec_num": "6.2"
            },
            {
                "text": "1. From the list of MWT candidates, we selected those sharing a N ADJ structure.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "French N ADJ reference lists",
                "sec_num": "6.2"
            },
            {
                "text": "2. We kept only the candidate terms which occur more than 2 twice in the French corpus. As a result of filtering, 1,999 candidate terms were extracted.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "French N ADJ reference lists",
                "sec_num": "6.2"
            },
            {
                "text": "3. We manually selected linguistically wellformed candidate terms. Here, 360 candidate terms were removed that included: misspelled terms, English terms, or subparts of longer terms.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "French N ADJ reference lists",
                "sec_num": "6.2"
            },
            {
                "text": "4. We took out the terms that are directly translated by the bilingual dictionary and found in the comparable corpora. We identified 61 terms of which 30 use a relational adjective such as vaisseau sanguin (blood vessel -B f ), produit laitier (dairy productr s % t ) and insuffisance cardiaque (heart failure -X 3 ).",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "French N ADJ reference lists",
                "sec_num": "6.2"
            },
            {
                "text": "Finally, we created two French reference lists: 6 N ADJE7 composed of 749 terms where ADJE is a epithetic adjective; 6 N ADJR7 composed of 829 terms where ADJR is a relational adjective.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "French N ADJ reference lists",
                "sec_num": "6.2"
            },
            {
                "text": "We first evaluated the quality of the default compositional method for the two French reference lists. Table 2 shows the results obtained. The first three columns indicate the number of French and Japanese terms found in the comparable corpora, and the number of correct French-Japanese translations.",
                "cite_spans": [],
                "ref_spans": [
                    {
                        "start": 103,
                        "end": 110,
                        "text": "Table 2",
                        "ref_id": null
                    }
                ],
                "eq_spans": [],
                "section": "Default compositional method",
                "sec_num": "6.3"
            },
            {
                "text": "The results of this experiment show that only a small quantity of terms were translated by the default compositional method. Here, the terms belonging to 6 N ADJE7 were more easily translated (10% with a precision of 69%) than the terms belonging to 6 N ADJR7 (1%). We were unable to generate any translations for 56 (12%) and 227 (27%) terms respectively from the 6 N ADJE7 and 6 N ADJR7 lists. This was because one or several content words of the MWT candidates were not present in the bilingual dictionary. The best translations of candidates belonging to the 6 N ADJE7 list are those where the adjective refers to a quantity such as faible (low), moyen (medium), or haut (high). Since our French-Japanese dictionary contained a small quantity of medical terms, the identified translations of the candidates belonging to the 6 N ADJR7 list refers to generic relational adjectives such as poids normal (standard weight -2 3 ),\u00e9tude nationale (national study -2 3 ), or activit\u00e9 physique (physical activity -3 3 ). We noticed that some generated MWUs do not exist in French such as poids (de) norme (standard weight), only the N ADJR form exists. # French # Japanese # correct terms terms translations 6 N ADJE7 76 98 68 6 N ADJR7 8 8 5 Table 2 : Production of the default compositional method",
                "cite_spans": [],
                "ref_spans": [
                    {
                        "start": 1238,
                        "end": 1245,
                        "text": "Table 2",
                        "ref_id": null
                    }
                ],
                "eq_spans": [],
                "section": "Default compositional method",
                "sec_num": "6.3"
            },
            {
                "text": "We will now turn to the evaluation of the morphologically-based compositional method is are dedicated to the translation of the 6 N ADJR7 list (see Table 4 ).",
                "cite_spans": [],
                "ref_spans": [
                    {
                        "start": 148,
                        "end": 155,
                        "text": "Table 4",
                        "ref_id": null
                    }
                ],
                "eq_spans": [],
                "section": "Morphologically-based compositional method",
                "sec_num": "6.4"
            },
            {
                "text": "By comparison with the previous method, the results of this experiment show that a significant quantity of terms have now been translated. Since the compositional method can yield several Japanese translations for one French term, we associated 170 Japanese terms to 128 French terms with a high level of precision: 88.2%. Here, we were unable to generate any translations for 136 (16%) terms in comparison with the 227 terms (27%) for the default compositional method.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Morphologically-based compositional method",
                "sec_num": "6.4"
            },
            {
                "text": "# French # Japanese # correct terms terms translations 6 N ADJR7 128 170 150 Table 4 : Production of the morphologically-based compositional method",
                "cite_spans": [],
                "ref_spans": [
                    {
                        "start": 77,
                        "end": 84,
                        "text": "Table 4",
                        "ref_id": null
                    }
                ],
                "eq_spans": [],
                "section": "Morphologically-based compositional method",
                "sec_num": "6.4"
            },
            {
                "text": "In Table 3 , each French suffix is associated with the number of identified translations. The most productive suffixes are -ique such as glyc\u00e9mie/glyc\u00e9mique (glycemia/glycemic), -al such as rein/r\u00e9nal (kidney/renal), -el such as",
                "cite_spans": [],
                "ref_spans": [
                    {
                        "start": 3,
                        "end": 10,
                        "text": "Table 3",
                        "ref_id": null
                    }
                ],
                "eq_spans": [],
                "section": "Morphologically-based compositional method",
                "sec_num": "6.4"
            },
            {
                "text": "http://kanji.free.fr/ 8 http://quebec-japon.com/lexique/index. php?a=index&d=259 http://dico.fj.free.fr/index.php 10 http://quebec-japon.com/lexique/index. php?a=index&d=3",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "",
                "sec_num": null
            }
        ],
        "back_matter": [
            {
                "text": "Finally from 859 terms relative to N ADJR structure, we translated 30 terms (5.1%) with the dictionary, 5 terms (0.6%) by the default compositional method, and 150 terms (17.5%) by the morphologically-based compositional method. It was difficult to find more translations for several reasons: i) some specialized adjectives or nouns were not included in our resources, ii) some terms were not taken into account by the Japanese extraction system, and iii) some terms were not included in the Japanese corpus.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "annex",
                "sec_num": null
            },
            {
                "text": "This study investigated the compilation of bilingual terminologies from comparable corpora and showed how to push back the limits of the methods used in alignment programs to translate both single and multi-word terms. We proposed an extended compositional method that bridges the gap between MWTs of different syntactic structures through morphological links. We experimented with the method on MWTs of N ADJ structure involving a relational adjective. By the use of a list of stripping-recoding rules conjugated with a terminology extraction system, the method was more efficient than the default compositional method. The evaluation proposed at the end of the paper shows that 170 French-Japanese MWTs were extracted with a high precision (88.2%). This increases the coverage of the French-Japanese terminology of MWTs that can be obtained by the bilingual dictionary or the default compositional method. We are aware that the efficiency of this method relies on the completeness of the morphological ressources, dictionaries and stripping-recoding rules. Such resources need to be up todate for new domains and corpus.In this study, we have observed that MWTs are of a different nature in each language: French patterns cover nominal phrases while Japanese patterns focus on morphologically-built compounds. A Japanese nominal phrase is not considered as a term: thus, the Japanese extraction system does not identify\u00b9 \u00ba\u00b0 \u00bb \u00ac I (caloric intake) as a candidate MWT but5\u00b9 \u00ba\u00b0\u00ac \u00bc , unlike the French extraction system which does the contrary (apport calorique -caloric intake). Since our morphologically-based compositional method associated\u00b9 \u00ba\u00b0\u00ac to apport calorique, we could yield the nominal phras\u0229 \u00b9 \u00ba\u00b0 \u00bd \u00bb \u00ac and improve lexical alignment.",
                "cite_spans": [],
                "ref_spans": [],
                "eq_spans": [],
                "section": "Conclusion and future work",
                "sec_num": "7"
            }
        ],
        "bib_entries": {
            "BIBREF0": {
                "ref_id": "b0",
                "title": "Translation by Machine of Complex Nominals: Getting it Right",
                "authors": [
                    {
                        "first": "Timothy",
                        "middle": [],
                        "last": "Baldwin",
                        "suffix": ""
                    },
                    {
                        "first": "Takaaki",
                        "middle": [],
                        "last": "Tanaka",
                        "suffix": ""
                    }
                ],
                "year": 2004,
                "venue": "Proceedings of the ACL 2004 Workshop on Multiword Expressions: Integrating Processing",
                "volume": "",
                "issue": "",
                "pages": "24--31",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Timothy Baldwin and Takaaki Tanaka. 2004. Trans- lation by Machine of Complex Nominals: Getting it Right. In Proceedings of the ACL 2004 Workshop on Multiword Expressions: Integrating Processing, pages 24-31, Barcelona, Spain.",
                "links": null
            },
            "BIBREF1": {
                "ref_id": "b1",
                "title": "Working with Specialized Language: A Practical Guide to Using Corpora",
                "authors": [
                    {
                        "first": "Lynne",
                        "middle": [],
                        "last": "Bowker",
                        "suffix": ""
                    },
                    {
                        "first": "Jennifer",
                        "middle": [],
                        "last": "Pearson",
                        "suffix": ""
                    }
                ],
                "year": 2002,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Lynne Bowker and Jennifer Pearson. 2002. Working with Specialized Language: A Practical Guide to Us- ing Corpora. Routeledge, London/New York.",
                "links": null
            },
            "BIBREF2": {
                "ref_id": "b2",
                "title": "The Mathematics of Statistical Machine Translation: Parameter Estimation",
                "authors": [
                    {
                        "first": "F",
                        "middle": [],
                        "last": "Peter",
                        "suffix": ""
                    },
                    {
                        "first": "Stephen",
                        "middle": [
                            "A Della"
                        ],
                        "last": "Brown",
                        "suffix": ""
                    },
                    {
                        "first": "Vincent",
                        "middle": [
                            "J"
                        ],
                        "last": "Pietra",
                        "suffix": ""
                    },
                    {
                        "first": "Robert",
                        "middle": [
                            "L"
                        ],
                        "last": "Della Pietra",
                        "suffix": ""
                    },
                    {
                        "first": "",
                        "middle": [],
                        "last": "Mercer",
                        "suffix": ""
                    }
                ],
                "year": 1993,
                "venue": "Computational Linguistics",
                "volume": "19",
                "issue": "2",
                "pages": "263--311",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Peter F. Brown, Stephen A. Della Pietra, Vincent J. Della Pietra, and Robert L. Mercer. 1993. The Mathe- matics of Statistical Machine Translation: Parameter Estimation. Computational Linguistics, 19(2):263- 311.",
                "links": null
            },
            "BIBREF3": {
                "ref_id": "b3",
                "title": "Looking for candidate translational equivalents in specialized, comparable corpora",
                "authors": [
                    {
                        "first": "Yun-Chuang",
                        "middle": [],
                        "last": "Chiao",
                        "suffix": ""
                    },
                    {
                        "first": "Pierre",
                        "middle": [],
                        "last": "Zweigenbaum",
                        "suffix": ""
                    }
                ],
                "year": 2002,
                "venue": "Proceedings of the 19th International Conference on Computational Linguistics (COLING'02)",
                "volume": "",
                "issue": "",
                "pages": "1208--1212",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Yun-Chuang Chiao and Pierre Zweigenbaum. 2002. Looking for candidate translational equivalents in spe- cialized, comparable corpora. In Proceedings of the 19th International Conference on Computational Lin- guistics (COLING'02), pages 1208-1212, Tapei, Tai- wan.",
                "links": null
            },
            "BIBREF4": {
                "ref_id": "b4",
                "title": "Qualitative terminology extraction",
                "authors": [
                    {
                        "first": "",
                        "middle": [],
                        "last": "B\u00e9atrice Daille",
                        "suffix": ""
                    }
                ],
                "year": 2001,
                "venue": "Didier Bourigault, Christian Jacquemin, and Marie-Claude L'Homme",
                "volume": "2",
                "issue": "",
                "pages": "149--166",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "B\u00e9atrice Daille. 2001. Qualitative terminology extrac- tion. In Didier Bourigault, Christian Jacquemin, and Marie-Claude L'Homme, editors, Recent Advances in Computational Terminology, volume 2 of Natu- ral Language Processing, pages 149-166. John Ben- jamins.",
                "links": null
            },
            "BIBREF5": {
                "ref_id": "b5",
                "title": "Terminology Mining",
                "authors": [
                    {
                        "first": "",
                        "middle": [],
                        "last": "B\u00e9atrice Daille",
                        "suffix": ""
                    }
                ],
                "year": 2003,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "29--44",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "B\u00e9atrice Daille. 2003. Terminology Mining. In Maria Teresa Pazienza, editor, Information Extraction in the Web Era, pages 29-44. Springer.",
                "links": null
            },
            "BIBREF6": {
                "ref_id": "b6",
                "title": "Une nouvelle approche\u00e0 l'extraction de lexiques bilingues\u00e0 partir de corpus comparables. Lexicometrica, Alignement lexical dans les corpus multilingues",
                "authors": [
                    {
                        "first": "Herv\u00e9",
                        "middle": [],
                        "last": "D\u00e9jean",
                        "suffix": ""
                    },
                    {
                        "first": "",
                        "middle": [],
                        "last": "Gaussier",
                        "suffix": ""
                    }
                ],
                "year": 2002,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "1--22",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Herv\u00e9 D\u00e9jean and\u00c9ric Gaussier. 2002. Une nouvelle ap- proche\u00e0 l'extraction de lexiques bilingues\u00e0 partir de corpus comparables. Lexicometrica, Alignement lexi- cal dans les corpus multilingues, pages 1-22.",
                "links": null
            },
            "BIBREF7": {
                "ref_id": "b7",
                "title": "Accurate Methods for the Statistics of Surprise and Coincidence",
                "authors": [
                    {
                        "first": "Ted",
                        "middle": [],
                        "last": "Dunning",
                        "suffix": ""
                    }
                ],
                "year": 1993,
                "venue": "Computational Linguistics",
                "volume": "19",
                "issue": "1",
                "pages": "61--74",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Ted Dunning. 1993. Accurate Methods for the Statistics of Surprise and Coincidence. Computational Linguis- tics, 19(1):61-74.",
                "links": null
            },
            "BIBREF8": {
                "ref_id": "b8",
                "title": "An approach based on multilingual thesauri and model combination for bilingual lexicon extraction",
                "authors": [
                    {
                        "first": "Fatia",
                        "middle": [],
                        "last": "Herv\u00e9 D\u00e9jean",
                        "suffix": ""
                    },
                    {
                        "first": "",
                        "middle": [],
                        "last": "Sadat",
                        "suffix": ""
                    },
                    {
                        "first": "",
                        "middle": [],
                        "last": "Gaussier",
                        "suffix": ""
                    }
                ],
                "year": 2002,
                "venue": "Proceedings of the 19th International Conference on Computational Linguistics (COLING'02)",
                "volume": "",
                "issue": "",
                "pages": "218--224",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Herv\u00e9 D\u00e9jean, Fatia Sadat, and\u00c9ric Gaussier. 2002. An approach based on multilingual thesauri and model combination for bilingual lexicon extraction. In Pro- ceedings of the 19th International Conference on Computational Linguistics (COLING'02), pages 218- 224, Tapei, Taiwan.",
                "links": null
            },
            "BIBREF9": {
                "ref_id": "b9",
                "title": "Transmission of Information: A statistical Theory of Communications",
                "authors": [
                    {
                        "first": "Robert",
                        "middle": [
                            "M"
                        ],
                        "last": "Fano",
                        "suffix": ""
                    }
                ],
                "year": 1961,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Robert M. Fano. 1961. Transmission of Information: A statistical Theory of Communications. MIT Press, Cambridge, MA, USA.",
                "links": null
            },
            "BIBREF10": {
                "ref_id": "b10",
                "title": "Scientific Dictionary",
                "authors": [
                    {
                        "first": "",
                        "middle": [],
                        "last": "French-Japanese",
                        "suffix": ""
                    }
                ],
                "year": 1989,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "French-Japanese Scientific Dictionary. 1989. Hakusu- isha. 4th edition.",
                "links": null
            },
            "BIBREF11": {
                "ref_id": "b11",
                "title": "Finding Terminology Translations from Non-parallel Corpora",
                "authors": [
                    {
                        "first": "Pascale",
                        "middle": [],
                        "last": "Fung",
                        "suffix": ""
                    },
                    {
                        "first": "Kathleen",
                        "middle": [],
                        "last": "Mckeown",
                        "suffix": ""
                    }
                ],
                "year": 1997,
                "venue": "Proceedings of the 5th Annual Workshop on Very Large Corpora (VLC'97)",
                "volume": "",
                "issue": "",
                "pages": "192--202",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Pascale Fung and Kathleen McKeown. 1997. Find- ing Terminology Translations from Non-parallel Cor- pora. In Proceedings of the 5th Annual Workshop on Very Large Corpora (VLC'97), pages 192-202, Hong Kong, China.",
                "links": null
            },
            "BIBREF12": {
                "ref_id": "b12",
                "title": "A Statistical View on Bilingual Lexicon Extraction: From Parallel Corpora to Nonparallel Corpora",
                "authors": [
                    {
                        "first": "Pascale",
                        "middle": [],
                        "last": "Fung",
                        "suffix": ""
                    }
                ],
                "year": 1998,
                "venue": "Proceedings of the 3rd Conference of the Association for Machine Translation in the Americas (AMTA'98)",
                "volume": "",
                "issue": "",
                "pages": "1--16",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Pascale Fung. 1998. A Statistical View on Bilingual Lexicon Extraction: From Parallel Corpora to Non- parallel Corpora. In David Farwell, Laurie Gerber, and Eduard Hovy, editors, Proceedings of the 3rd Con- ference of the Association for Machine Translation in the Americas (AMTA'98), pages 1-16, Langhorne, PA, USA.",
                "links": null
            },
            "BIBREF13": {
                "ref_id": "b13",
                "title": "The Word Wide Web as a Resource for Example-Based Machine Translation Tasks",
                "authors": [
                    {
                        "first": "Gregory",
                        "middle": [],
                        "last": "Grefenstette",
                        "suffix": ""
                    }
                ],
                "year": 1999,
                "venue": "ASLIB'99 Translating and the Computer",
                "volume": "21",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Gregory Grefenstette. 1999. The Word Wide Web as a Resource for Example-Based Machine Translation Tasks. In ASLIB'99 Translating and the Computer 21, London, UK.",
                "links": null
            },
            "BIBREF14": {
                "ref_id": "b14",
                "title": "The syntax and the semantics of complex nominals",
                "authors": [
                    {
                        "first": "Judith",
                        "middle": [],
                        "last": "Levi",
                        "suffix": ""
                    }
                ],
                "year": 1978,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Judith Levi. 1978. The syntax and the semantics of com- plex nominals. Academic Press, London.",
                "links": null
            },
            "BIBREF15": {
                "ref_id": "b15",
                "title": "A Word-to-Word Model of Translational Equivalence",
                "authors": [
                    {
                        "first": "I",
                        "middle": [],
                        "last": "",
                        "suffix": ""
                    },
                    {
                        "first": "Dan",
                        "middle": [],
                        "last": "Melamed",
                        "suffix": ""
                    }
                ],
                "year": 1997,
                "venue": "Proceedings of the 35th",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "I. Dan Melamed. 1997. A Word-to-Word Model of Translational Equivalence. In Philip R. Cohen and Wolfgang Wahlster, editors, Proceedings of the 35th",
                "links": null
            },
            "BIBREF16": {
                "ref_id": "b16",
                "title": "Annual Meeting of the Association for Computational Linguistics and 8th Conference of the European Chapter of the Association for Computational Linguistics",
                "authors": [],
                "year": null,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "490--497",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Annual Meeting of the Association for Computational Linguistics and 8th Conference of the European Chap- ter of the Association for Computational Linguistics, pages 490-497, Madrid, Spain.",
                "links": null
            },
            "BIBREF17": {
                "ref_id": "b17",
                "title": "Automatic rule induction for unknown-word guessing",
                "authors": [
                    {
                        "first": "Andrei",
                        "middle": [],
                        "last": "Mikheev",
                        "suffix": ""
                    }
                ],
                "year": 1997,
                "venue": "Computational Linguistics",
                "volume": "23",
                "issue": "3",
                "pages": "405--423",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Andrei Mikheev. 1997. Automatic rule induction for unknown-word guessing. Computational Linguistics, 23(3):405-423.",
                "links": null
            },
            "BIBREF18": {
                "ref_id": "b18",
                "title": "Comparabilit\u00e9 de corpus et fouille terminologique multilingue",
                "authors": [
                    {
                        "first": "Emmanuel",
                        "middle": [],
                        "last": "Morin",
                        "suffix": ""
                    },
                    {
                        "first": "B\u00e9atrice",
                        "middle": [],
                        "last": "Daille",
                        "suffix": ""
                    }
                ],
                "year": 2006,
                "venue": "Traitement Automatique des Langues (TAL)",
                "volume": "47",
                "issue": "2",
                "pages": "113--136",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Emmanuel Morin and B\u00e9atrice Daille. 2006. Com- parabilit\u00e9 de corpus et fouille terminologique multi- lingue. Traitement Automatique des Langues (TAL), 47(2):113-136.",
                "links": null
            },
            "BIBREF19": {
                "ref_id": "b19",
                "title": "Bilingual Terminology Mining -Using Brain, not brawn comparable corpora",
                "authors": [
                    {
                        "first": "Emmanuel",
                        "middle": [],
                        "last": "Morin",
                        "suffix": ""
                    },
                    {
                        "first": "B\u00e9atrice",
                        "middle": [],
                        "last": "Daille",
                        "suffix": ""
                    },
                    {
                        "first": "Koichi",
                        "middle": [],
                        "last": "Takeuchi",
                        "suffix": ""
                    },
                    {
                        "first": "Kyo",
                        "middle": [],
                        "last": "Kageura",
                        "suffix": ""
                    }
                ],
                "year": 2007,
                "venue": "Proceedings of the 45th Annual Meeting of the Association for Computational Linguistics (ACL'07)",
                "volume": "",
                "issue": "",
                "pages": "664--671",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Emmanuel Morin, B\u00e9atrice Daille, Koichi Takeuchi, and Kyo Kageura. 2007. Bilingual Terminology Mining - Using Brain, not brawn comparable corpora. In Pro- ceedings of the 45th Annual Meeting of the Association for Computational Linguistics (ACL'07), pages 664- 671, Prague, Czech Republic.",
                "links": null
            },
            "BIBREF20": {
                "ref_id": "b20",
                "title": "Identify Word Translations in Non-Parallel Texts",
                "authors": [
                    {
                        "first": "Reinhard",
                        "middle": [],
                        "last": "Rapp",
                        "suffix": ""
                    }
                ],
                "year": 1995,
                "venue": "Proceedings of the 35th Annual Meeting of the Association for Computational Linguistics (ACL'95)",
                "volume": "",
                "issue": "",
                "pages": "320--322",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Reinhard Rapp. 1995. Identify Word Translations in Non-Parallel Texts. In Proceedings of the 35th Annual Meeting of the Association for Computational Linguis- tics (ACL'95), pages 320-322, Boston, MA, USA.",
                "links": null
            },
            "BIBREF21": {
                "ref_id": "b21",
                "title": "Automatic Identification of Word Translations from Unrelated English and German Corpora",
                "authors": [
                    {
                        "first": "Reinhard",
                        "middle": [],
                        "last": "Rapp",
                        "suffix": ""
                    }
                ],
                "year": 1999,
                "venue": "Proceedings of the 37th Annual Meeting of the Association for Computational Linguistics (ACL'99)",
                "volume": "",
                "issue": "",
                "pages": "519--526",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Reinhard Rapp. 1999. Automatic Identification of Word Translations from Unrelated English and German Cor- pora. In Proceedings of the 37th Annual Meeting of the Association for Computational Linguistics (ACL'99), pages 519-526, College Park, MD, USA.",
                "links": null
            },
            "BIBREF22": {
                "ref_id": "b22",
                "title": "Compiling French-Japanese Terminologies from the Web",
                "authors": [
                    {
                        "first": "Xavier",
                        "middle": [],
                        "last": "Robitaille",
                        "suffix": ""
                    },
                    {
                        "first": "Xavier",
                        "middle": [],
                        "last": "Sasaki",
                        "suffix": ""
                    },
                    {
                        "first": "Masatsugu",
                        "middle": [],
                        "last": "Tonoike",
                        "suffix": ""
                    },
                    {
                        "first": "Satoshi",
                        "middle": [],
                        "last": "Sato",
                        "suffix": ""
                    },
                    {
                        "first": "Satoshi",
                        "middle": [],
                        "last": "Utsuro",
                        "suffix": ""
                    }
                ],
                "year": 2006,
                "venue": "Proceedings of the 11th Conference of the European Chapter of the Association for Computational Linguistics (EACL'06)",
                "volume": "",
                "issue": "",
                "pages": "225--232",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Xavier Robitaille, Xavier Sasaki, Masatsugu Tonoike, Satoshi Sato, and Satoshi Utsuro. 2006. Compil- ing French-Japanese Terminologies from the Web. In Proceedings of the 11th Conference of the European Chapter of the Association for Computational Linguis- tics (EACL'06), pages 225-232, Trento, Italy.",
                "links": null
            },
            "BIBREF23": {
                "ref_id": "b23",
                "title": "Journal of the Association for Computational Machinery",
                "authors": [
                    {
                        "first": "Gerard",
                        "middle": [],
                        "last": "Salton",
                        "suffix": ""
                    },
                    {
                        "first": "Michael",
                        "middle": [
                            "E"
                        ],
                        "last": "Lesk",
                        "suffix": ""
                    }
                ],
                "year": 1968,
                "venue": "",
                "volume": "15",
                "issue": "",
                "pages": "8--36",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Gerard Salton and Michael E. Lesk. 1968. Computer evaluation of indexing and text processing. Jour- nal of the Association for Computational Machinery, 15(1):8-36.",
                "links": null
            },
            "BIBREF24": {
                "ref_id": "b24",
                "title": "Construction of grammar based term extraction model for japanese",
                "authors": [
                    {
                        "first": "Koichi",
                        "middle": [],
                        "last": "Takeuchi",
                        "suffix": ""
                    },
                    {
                        "first": "Kyo",
                        "middle": [],
                        "last": "Kageura",
                        "suffix": ""
                    },
                    {
                        "first": "B\u00e9atrice",
                        "middle": [],
                        "last": "Daille",
                        "suffix": ""
                    },
                    {
                        "first": "Laurent",
                        "middle": [],
                        "last": "Romary",
                        "suffix": ""
                    }
                ],
                "year": 2004,
                "venue": "Proceedings of the COLING 2004, 3rd International Workshop on Computational Terminology (COMPUTERM'04)",
                "volume": "",
                "issue": "",
                "pages": "91--94",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "Koichi Takeuchi, Kyo Kageura, B\u00e9atrice Daille, and Lau- rent Romary. 2004. Construction of grammar based term extraction model for japanese. In Sophia Anana- diou and Pierre Zweigenbaum, editors, Proceedings of the COLING 2004, 3rd International Workshop on Computational Terminology (COMPUTERM'04), pages 91-94, Geneva, Switzerland.",
                "links": null
            },
            "BIBREF25": {
                "ref_id": "b25",
                "title": "An elementary mathematical theory of classification",
                "authors": [
                    {
                        "first": "T",
                        "middle": [
                            "T"
                        ],
                        "last": "Tanimoto",
                        "suffix": ""
                    }
                ],
                "year": 1958,
                "venue": "",
                "volume": "",
                "issue": "",
                "pages": "",
                "other_ids": {},
                "num": null,
                "urls": [],
                "raw_text": "T. T. Tanimoto. 1958. An elementary mathematical the- ory of classification. Technical report, IBM Research.",
                "links": null
            }
        },
        "ref_entries": {
            "FIGREF0": {
                "num": null,
                "text": "Architecture of the multilingual terminology mining chain",
                "type_str": "figure",
                "uris": null
            },
            "FIGREF1": {
                "num": null,
                "text": "the dictionaries yields a single resource with 173,156 entries (114,461 single words and 58,695 multi words) and an average of 2.1 translations per entry.",
                "type_str": "figure",
                "uris": null
            },
            "TABREF0": {
                "type_str": "table",
                "text": "Illustration of the compositional method (the underlined Japanese MWT actually exists)",
                "html": null,
                "num": null,
                "content": "<table/>"
            }
        }
    }
}