,text,id,label
0,"This paper presents an [[ algorithm ]] for << computing optical flow , shape , motion , lighting , and albedo >> from an image sequence of a rigidly-moving Lambertian object under distant illumination .",0,3
1,"This paper presents an << algorithm >> for computing optical flow , shape , motion , lighting , and albedo from an [[ image sequence ]] of a rigidly-moving Lambertian object under distant illumination .",1,3
2,"This paper presents an algorithm for computing optical flow , shape , motion , lighting , and albedo from an << image sequence >> of a [[ rigidly-moving Lambertian object ]] under distant illumination .",2,1
3,"This paper presents an algorithm for computing optical flow , shape , motion , lighting , and albedo from an image sequence of a << rigidly-moving Lambertian object >> under [[ distant illumination ]] .",3,1
4,"The problem is formulated in a manner that subsumes structure from [[ motion ]] , << multi-view stereo >> , and photo-metric stereo as special cases .",4,0
5,"The problem is formulated in a manner that subsumes structure from motion , [[ multi-view stereo ]] , and << photo-metric stereo >> as special cases .",5,0
6,The << algorithm >> utilizes both [[ spatial and temporal intensity variation ]] as cues : the former constrains flow and the latter constrains surface orientation ; combining both cues enables dense reconstruction of both textured and texture-less surfaces .,6,3
7,The algorithm utilizes both spatial and temporal intensity variation as << cues >> : the [[ former ]] constrains flow and the latter constrains surface orientation ; combining both cues enables dense reconstruction of both textured and texture-less surfaces .,7,2
8,The algorithm utilizes both spatial and temporal intensity variation as cues : the [[ former ]] constrains << flow >> and the latter constrains surface orientation ; combining both cues enables dense reconstruction of both textured and texture-less surfaces .,8,3
9,The algorithm utilizes both spatial and temporal intensity variation as cues : the [[ former ]] constrains flow and the << latter >> constrains surface orientation ; combining both cues enables dense reconstruction of both textured and texture-less surfaces .,9,0
10,The algorithm utilizes both spatial and temporal intensity variation as << cues >> : the former constrains flow and the [[ latter ]] constrains surface orientation ; combining both cues enables dense reconstruction of both textured and texture-less surfaces .,10,2
11,The algorithm utilizes both spatial and temporal intensity variation as cues : the former constrains flow and the [[ latter ]] constrains << surface orientation >> ; combining both cues enables dense reconstruction of both textured and texture-less surfaces .,11,3
12,The algorithm utilizes both spatial and temporal intensity variation as cues : the former constrains flow and the latter constrains surface orientation ; combining both [[ cues ]] enables << dense reconstruction of both textured and texture-less surfaces >> .,12,3
13,"The << algorithm >> works by iteratively [[ estimating affine camera parameters , illumination , shape , and albedo ]] in an alternating fashion .",13,3
14,An [[ entity-oriented approach ]] to << restricted-domain parsing >> is proposed .,14,3
15,"Like semantic grammar , [[ this ]] allows easy exploitation of << limited domain semantics >> .",15,3
16,"In addition , [[ it ]] facilitates << fragmentary recognition >> and the use of multiple parsing strategies , and so is particularly useful for robust recognition of extra-grammatical input .",16,3
17,"In addition , [[ it ]] facilitates fragmentary recognition and the use of << multiple parsing strategies >> , and so is particularly useful for robust recognition of extra-grammatical input .",17,3
18,"In addition , it facilitates fragmentary recognition and the use of [[ multiple parsing strategies ]] , and so is particularly useful for robust << recognition of extra-grammatical input >> .",18,3
19,"Representative samples from an entity-oriented language definition are presented , along with a [[ control structure ]] for an << entity-oriented parser >> , some parsing strategies that use the control structure , and worked examples of parses .",19,3
20,"Representative samples from an entity-oriented language definition are presented , along with a control structure for an entity-oriented parser , some << parsing strategies >> that use the [[ control structure ]] , and worked examples of parses .",20,3
21,A << parser >> incorporating the [[ control structure ]] and the parsing strategies is currently under implementation .,21,4
22,This paper summarizes the formalism of Category Cooccurrence Restrictions -LRB- CCRs -RRB- and describes two [[ parsing algorithms ]] that interpret << it >> .,22,3
23,The use of CCRs leads to << syntactic descriptions >> formulated entirely with [[ restrictive statements ]] .,23,1
24,The paper shows how conventional [[ algorithms ]] for the analysis of context free languages can be adapted to the << CCR formalism >> .,24,3
25,The paper shows how conventional << algorithms >> for the analysis of [[ context free languages ]] can be adapted to the CCR formalism .,25,3
26,Special attention is given to the part of the parser that checks the fulfillment of [[ logical well-formedness conditions ]] on << trees >> .,26,1
27,We present a [[ text mining method ]] for finding << synonymous expressions >> based on the distributional hypothesis in a set of coherent corpora .,27,3
28,We present a << text mining method >> for finding synonymous expressions based on the [[ distributional hypothesis ]] in a set of coherent corpora .,28,3
29,This paper proposes a new methodology to improve the [[ accuracy ]] of a << term aggregation system >> using each author 's text as a coherent corpus .,29,6
30,This paper proposes a new << methodology >> to improve the accuracy of a [[ term aggregation system ]] using each author 's text as a coherent corpus .,30,6
31,"Our proposed method improves the [[ accuracy ]] of our << term aggregation system >> , showing that our approach is successful .",31,6
32,"Our proposed << method >> improves the accuracy of our [[ term aggregation system ]] , showing that our approach is successful .",32,6
33,"In this work , we present a [[ technique ]] for << robust estimation >> , which by explicitly incorporating the inherent uncertainty of the estimation procedure , results in a more efficient robust estimation algorithm .",33,3
34,"In this work , we present a [[ technique ]] for robust estimation , which by explicitly incorporating the inherent uncertainty of the estimation procedure , results in a more << efficient robust estimation algorithm >> .",34,3
35,"In this work , we present a << technique >> for robust estimation , which by explicitly incorporating the [[ inherent uncertainty of the estimation procedure ]] , results in a more efficient robust estimation algorithm .",35,3
36,"The combination of these two [[ strategies ]] results in a << robust estimation procedure >> that provides a significant speed-up over existing RANSAC techniques , while requiring no prior information to guide the sampling process .",36,3
37,"The combination of these two strategies results in a << robust estimation procedure >> that provides a significant speed-up over existing [[ RANSAC techniques ]] , while requiring no prior information to guide the sampling process .",37,5
38,"In particular , our [[ algorithm ]] requires , on average , 3-10 times fewer samples than standard << RANSAC >> , which is in close agreement with theoretical predictions .",38,5
39,The efficiency of the << algorithm >> is demonstrated on a selection of [[ geometric estimation problems ]] .,39,6
40,An attempt has been made to use an [[ Augmented Transition Network ]] as a procedural << dialog model >> .,40,2
41,The development of such a model appears to be important in several respects : as a << device >> to represent and to use different [[ dialog schemata ]] proposed in empirical conversation analysis ; as a device to represent and to use models of verbal interaction ; as a device combining knowledge about dialog schemata and about verbal interaction with knowledge about task-oriented and goal-directed dialogs .,41,3
42,The development of such a model appears to be important in several respects : as a device to represent and to use different [[ dialog schemata ]] proposed in empirical << conversation analysis >> ; as a device to represent and to use models of verbal interaction ; as a device combining knowledge about dialog schemata and about verbal interaction with knowledge about task-oriented and goal-directed dialogs .,42,3
43,The development of such a model appears to be important in several respects : as a device to represent and to use different dialog schemata proposed in empirical conversation analysis ; as a << device >> to represent and to use [[ models ]] of verbal interaction ; as a device combining knowledge about dialog schemata and about verbal interaction with knowledge about task-oriented and goal-directed dialogs .,43,3
44,The development of such a model appears to be important in several respects : as a device to represent and to use different dialog schemata proposed in empirical conversation analysis ; as a device to represent and to use [[ models ]] of << verbal interaction >> ; as a device combining knowledge about dialog schemata and about verbal interaction with knowledge about task-oriented and goal-directed dialogs .,44,3
45,The development of such a model appears to be important in several respects : as a device to represent and to use different dialog schemata proposed in empirical conversation analysis ; as a device to represent and to use models of verbal interaction ; as a device combining knowledge about [[ dialog schemata ]] and about << verbal interaction >> with knowledge about task-oriented and goal-directed dialogs .,45,0
46,A standard [[ ATN ]] should be further developed in order to account for the << verbal interactions >> of task-oriented dialogs .,46,3
47,A standard ATN should be further developed in order to account for the [[ verbal interactions ]] of << task-oriented dialogs >> .,47,1
48,We present a practically [[ unsupervised learning method ]] to produce << single-snippet answers >> to definition questions in question answering systems that supplement Web search engines .,48,3
49,We present a practically unsupervised learning method to produce single-snippet answers to definition questions in [[ question answering systems ]] that supplement << Web search engines >> .,49,3
50,"The [[ method ]] exploits << on-line encyclopedias and dictionaries >> to generate automatically an arbitrarily large number of positive and negative definition examples , which are then used to train an svm to separate the two classes .",50,3
51,"The method exploits [[ on-line encyclopedias and dictionaries ]] to generate automatically an arbitrarily large number of << positive and negative definition examples >> , which are then used to train an svm to separate the two classes .",51,3
52,"The method exploits on-line encyclopedias and dictionaries to generate automatically an arbitrarily large number of [[ positive and negative definition examples ]] , which are then used to train an << svm >> to separate the two classes .",52,3
53,"We show experimentally that the proposed method is viable , that [[ it ]] outperforms the << alternative >> of training the system on questions and news articles from trec , and that it helps the search engine handle definition questions significantly better .",53,5
54,"We show experimentally that the proposed method is viable , that it outperforms the alternative of training the << system >> on questions and [[ news articles ]] from trec , and that it helps the search engine handle definition questions significantly better .",54,3
55,"We show experimentally that the proposed method is viable , that it outperforms the alternative of training the system on questions and [[ news articles ]] from << trec >> , and that it helps the search engine handle definition questions significantly better .",55,4
56,"We show experimentally that the proposed method is viable , that it outperforms the alternative of training the system on questions and news articles from trec , and that [[ it ]] helps the << search engine >> handle definition questions significantly better .",56,3
57,We revisit the << classical decision-theoretic problem of weighted expert voting >> from a [[ statistical learning perspective ]] .,57,3
58,"In the case of known expert competence levels , we give [[ sharp error estimates ]] for the << optimal rule >> .",58,3
59,We analyze a [[ reweighted version of the Kikuchi approximation ]] for estimating the << log partition function of a product distribution >> defined over a region graph .,59,3
60,We analyze a reweighted version of the Kikuchi approximation for estimating the [[ log partition function of a product distribution ]] defined over a << region graph >> .,60,1
61,"We establish sufficient conditions for the [[ concavity ]] of our << reweighted objective function >> in terms of weight assignments in the Kikuchi expansion , and show that a reweighted version of the sum product algorithm applied to the Kikuchi region graph will produce global optima of the Kikuchi approximation whenever the algorithm converges .",61,1
62,"We establish sufficient conditions for the concavity of our reweighted objective function in terms of weight assignments in the Kikuchi expansion , and show that a [[ reweighted version of the sum product algorithm ]] applied to the << Kikuchi region graph >> will produce global optima of the Kikuchi approximation whenever the algorithm converges .",62,3
63,"We establish sufficient conditions for the concavity of our reweighted objective function in terms of weight assignments in the Kikuchi expansion , and show that a reweighted version of the sum product algorithm applied to the Kikuchi region graph will produce [[ global optima ]] of the << Kikuchi approximation >> whenever the algorithm converges .",63,1
64,"Finally , we provide an explicit characterization of the polytope of concavity in terms of the [[ cycle structure ]] of the << region graph >> .",64,1
65,We apply a [[ decision tree based approach ]] to << pronoun resolution >> in spoken dialogue .,65,3
66,We apply a decision tree based approach to [[ pronoun resolution ]] in << spoken dialogue >> .,66,3
67,Our [[ system ]] deals with << pronouns >> with NP - and non-NP-antecedents .,67,3
68,Our system deals with << pronouns >> with [[ NP - and non-NP-antecedents ]] .,68,3
69,We present a set of [[ features ]] designed for << pronoun resolution >> in spoken dialogue and determine the most promising features .,69,3
70,We present a set of features designed for [[ pronoun resolution ]] in << spoken dialogue >> and determine the most promising features .,70,3
71,We evaluate the << system >> on twenty [[ Switchboard dialogues ]] and show that it compares well to Byron 's -LRB- 2002 -RRB- manually tuned system .,71,6
72,We evaluate the system on twenty Switchboard dialogues and show that [[ it ]] compares well to << Byron 's -LRB- 2002 -RRB- manually tuned system >> .,72,5
73,"We present a new [[ approach ]] for building an efficient and robust << classifier >> for the two class problem , that localizes objects that may appear in the image under different orien-tations .",73,3
74,"We present a new approach for building an efficient and robust [[ classifier ]] for the two << class problem >> , that localizes objects that may appear in the image under different orien-tations .",74,3
75,"In contrast to other works that address this problem using multiple classifiers , each one specialized for a specific orientation , we propose a simple two-step << approach >> with an [[ estimation stage ]] and a classification stage .",75,4
76,"In contrast to other works that address this problem using multiple classifiers , each one specialized for a specific orientation , we propose a simple two-step approach with an [[ estimation stage ]] and a << classification stage >> .",76,0
77,"In contrast to other works that address this problem using multiple classifiers , each one specialized for a specific orientation , we propose a simple two-step << approach >> with an estimation stage and a [[ classification stage ]] .",77,4
78,The estimator yields an initial set of potential << object poses >> that are then validated by the [[ classifier ]] .,78,3
79,This methodology allows reducing the [[ time complexity ]] of the << algorithm >> while classification results remain high .,79,6
80,"The << classifier >> we use in both stages is based on a [[ boosted combination of Random Ferns ]] over local histograms of oriented gradients -LRB- HOGs -RRB- , which we compute during a pre-processing step .",80,3
81,"The classifier we use in both stages is based on a << boosted combination of Random Ferns >> over [[ local histograms of oriented gradients -LRB- HOGs -RRB- ]] , which we compute during a pre-processing step .",81,1
82,"The classifier we use in both stages is based on a boosted combination of Random Ferns over << local histograms of oriented gradients -LRB- HOGs -RRB- >> , which we compute during a [[ pre-processing step ]] .",82,3
83,Both the use of [[ supervised learning ]] and working on the gradient space makes our << approach >> robust while being efficient at run-time .,83,3
84,Both the use of supervised learning and working on the [[ gradient space ]] makes our << approach >> robust while being efficient at run-time .,84,3
85,"We show these properties by thorough testing on standard databases and on a new << database >> made of [[ motorbikes under planar rotations ]] , and with challenging conditions such as cluttered backgrounds , changing illumination conditions and partial occlusions .",85,1
86,"We show these properties by thorough testing on standard databases and on a new << database >> made of motorbikes under planar rotations , and with challenging [[ conditions ]] such as cluttered backgrounds , changing illumination conditions and partial occlusions .",86,1
87,"We show these properties by thorough testing on standard databases and on a new database made of motorbikes under planar rotations , and with challenging << conditions >> such as [[ cluttered backgrounds ]] , changing illumination conditions and partial occlusions .",87,2
88,"We show these properties by thorough testing on standard databases and on a new database made of motorbikes under planar rotations , and with challenging conditions such as [[ cluttered backgrounds ]] , << changing illumination conditions >> and partial occlusions .",88,0
89,"We show these properties by thorough testing on standard databases and on a new database made of motorbikes under planar rotations , and with challenging << conditions >> such as cluttered backgrounds , [[ changing illumination conditions ]] and partial occlusions .",89,2
90,"We show these properties by thorough testing on standard databases and on a new database made of motorbikes under planar rotations , and with challenging conditions such as cluttered backgrounds , [[ changing illumination conditions ]] and << partial occlusions >> .",90,0
91,"We show these properties by thorough testing on standard databases and on a new database made of motorbikes under planar rotations , and with challenging << conditions >> such as cluttered backgrounds , changing illumination conditions and [[ partial occlusions ]] .",91,2
92,A very simple improved [[ duration model ]] has reduced the error rate by about 10 % in both << triphone and semiphone systems >> .,92,3
93,A very simple improved duration model has reduced the [[ error rate ]] by about 10 % in both << triphone and semiphone systems >> .,93,6
94,"A new << training strategy >> has been tested which , by itself , did not provide useful improvements but suggests that improvements can be obtained by a related [[ rapid adaptation technique ]] .",94,3
95,"Finally , the << recognizer >> has been modified to use [[ bigram back-off language models ]] .",95,3
96,The [[ system ]] was then transferred from the << RM task >> to the ATIS CSR task and a limited number of development tests performed .,96,3
97,The [[ system ]] was then transferred from the RM task to the << ATIS CSR task >> and a limited number of development tests performed .,97,3
98,The system was then transferred from the [[ RM task ]] to the << ATIS CSR task >> and a limited number of development tests performed .,98,0
99,A new [[ approach ]] for << Interactive Machine Translation >> where the author interacts during the creation or the modification of the document is proposed .,99,3
100,This paper presents a new << interactive disambiguation scheme >> based on the [[ paraphrasing ]] of a parser 's multiple output .,100,3
101,We describe a novel [[ approach ]] to << statistical machine translation >> that combines syntactic information in the source language with recent advances in phrasal translation .,101,3
102,We describe a novel << approach >> to statistical machine translation that combines [[ syntactic information ]] in the source language with recent advances in phrasal translation .,102,4
103,We describe a novel approach to statistical machine translation that combines [[ syntactic information ]] in the source language with recent advances in << phrasal translation >> .,103,0
104,We describe a novel << approach >> to statistical machine translation that combines syntactic information in the source language with recent advances in [[ phrasal translation ]] .,104,4
105,"This << method >> requires a [[ source-language dependency parser ]] , target language word segmentation and an unsupervised word alignment component .",105,3
106,"This method requires a [[ source-language dependency parser ]] , << target language word segmentation >> and an unsupervised word alignment component .",106,0
107,"This << method >> requires a source-language dependency parser , [[ target language word segmentation ]] and an unsupervised word alignment component .",107,3
108,"This method requires a source-language dependency parser , [[ target language word segmentation ]] and an << unsupervised word alignment component >> .",108,0
109,"This << method >> requires a source-language dependency parser , target language word segmentation and an [[ unsupervised word alignment component ]] .",109,3
110,We describe an efficient decoder and show that using these [[ tree-based models ]] in combination with conventional << SMT models >> provides a promising approach that incorporates the power of phrasal SMT with the linguistic generality available in a parser .,110,0
111,We describe an efficient decoder and show that using these [[ tree-based models ]] in combination with conventional SMT models provides a promising << approach >> that incorporates the power of phrasal SMT with the linguistic generality available in a parser .,111,3
112,We describe an efficient decoder and show that using these tree-based models in combination with conventional [[ SMT models ]] provides a promising << approach >> that incorporates the power of phrasal SMT with the linguistic generality available in a parser .,112,3
113,We describe an efficient decoder and show that using these tree-based models in combination with conventional SMT models provides a promising approach that incorporates the power of [[ phrasal SMT ]] with the << linguistic generality >> available in a parser .,113,0
114,We describe an efficient decoder and show that using these tree-based models in combination with conventional SMT models provides a promising approach that incorporates the power of [[ phrasal SMT ]] with the linguistic generality available in a << parser >> .,114,3
115,We describe an efficient decoder and show that using these tree-based models in combination with conventional SMT models provides a promising approach that incorporates the power of phrasal SMT with the [[ linguistic generality ]] available in a << parser >> .,115,1
116,"<< Video >> provides not only rich [[ visual cues ]] such as motion and appearance , but also much less explored long-range temporal interactions among objects .",116,1
117,"Video provides not only rich << visual cues >> such as [[ motion ]] and appearance , but also much less explored long-range temporal interactions among objects .",117,2
118,"Video provides not only rich visual cues such as [[ motion ]] and << appearance >> , but also much less explored long-range temporal interactions among objects .",118,0
119,"Video provides not only rich << visual cues >> such as motion and [[ appearance ]] , but also much less explored long-range temporal interactions among objects .",119,2
120,We aim to capture such interactions and to construct a powerful [[ intermediate-level video representation ]] for subsequent << recognition >> .,120,3
121,"First , we develop an efficient << spatio-temporal video segmentation algorithm >> , which naturally incorporates [[ long-range motion cues ]] from the past and future frames in the form of clusters of point tracks with coherent motion .",121,3
122,"First , we develop an efficient spatio-temporal video segmentation algorithm , which naturally incorporates << long-range motion cues >> from the past and future frames in the form of [[ clusters of point tracks ]] with coherent motion .",122,3
123,"Second , we devise a new << track clustering cost function >> that includes [[ occlusion reasoning ]] , in the form of depth ordering constraints , as well as motion similarity along the tracks .",123,4
124,"Second , we devise a new track clustering cost function that includes << occlusion reasoning >> , in the form of [[ depth ordering constraints ]] , as well as motion similarity along the tracks .",124,1
125,"Second , we devise a new << track clustering cost function >> that includes occlusion reasoning , in the form of depth ordering constraints , as well as [[ motion similarity ]] along the tracks .",125,4
126,We evaluate the proposed << approach >> on a challenging set of [[ video sequences of office scenes ]] from feature length movies .,126,6
127,"In this paper , we introduce [[ KAZE features ]] , a novel << multiscale 2D feature detection and description algorithm >> in nonlinear scale spaces .",127,2
128,"In this paper , we introduce KAZE features , a novel << multiscale 2D feature detection and description algorithm >> in [[ nonlinear scale spaces ]] .",128,1
129,"In contrast , we detect and describe << 2D features >> in a [[ nonlinear scale space ]] by means of nonlinear diffusion filtering .",129,1
130,"In contrast , we detect and describe << 2D features >> in a nonlinear scale space by means of [[ nonlinear diffusion filtering ]] .",130,3
131,The << nonlinear scale space >> is built using efficient [[ Additive Operator Splitting -LRB- AOS -RRB- techniques ]] and variable con-ductance diffusion .,131,3
132,The nonlinear scale space is built using efficient [[ Additive Operator Splitting -LRB- AOS -RRB- techniques ]] and << variable con-ductance diffusion >> .,132,0
133,The << nonlinear scale space >> is built using efficient Additive Operator Splitting -LRB- AOS -RRB- techniques and [[ variable con-ductance diffusion ]] .,133,3
134,"Even though our [[ features ]] are somewhat more expensive to compute than << SURF >> due to the construction of the nonlinear scale space , but comparable to SIFT , our results reveal a step forward in performance both in detection and description against previous state-of-the-art methods .",134,5
135,"Even though our [[ features ]] are somewhat more expensive to compute than SURF due to the construction of the nonlinear scale space , but comparable to << SIFT >> , our results reveal a step forward in performance both in detection and description against previous state-of-the-art methods .",135,5
136,"Even though our features are somewhat more expensive to compute than SURF due to the construction of the nonlinear scale space , but comparable to SIFT , our [[ results ]] reveal a step forward in performance both in detection and description against previous << state-of-the-art methods >> .",136,5
137,"Even though our features are somewhat more expensive to compute than SURF due to the construction of the nonlinear scale space , but comparable to SIFT , our << results >> reveal a step forward in performance both in [[ detection ]] and description against previous state-of-the-art methods .",137,6
138,"Even though our features are somewhat more expensive to compute than SURF due to the construction of the nonlinear scale space , but comparable to SIFT , our results reveal a step forward in performance both in [[ detection ]] and << description >> against previous state-of-the-art methods .",138,0
139,"Even though our features are somewhat more expensive to compute than SURF due to the construction of the nonlinear scale space , but comparable to SIFT , our results reveal a step forward in performance both in [[ detection ]] and description against previous << state-of-the-art methods >> .",139,6
140,"Even though our features are somewhat more expensive to compute than SURF due to the construction of the nonlinear scale space , but comparable to SIFT , our << results >> reveal a step forward in performance both in detection and [[ description ]] against previous state-of-the-art methods .",140,6
141,"Even though our features are somewhat more expensive to compute than SURF due to the construction of the nonlinear scale space , but comparable to SIFT , our results reveal a step forward in performance both in detection and [[ description ]] against previous << state-of-the-art methods >> .",141,6
142,[[ Creating summaries ]] on lengthy Semantic Web documents for quick << identification of the corresponding entity >> has been of great contemporary interest .,142,3
143,<< Creating summaries >> on [[ lengthy Semantic Web documents ]] for quick identification of the corresponding entity has been of great contemporary interest .,143,3
144,"Specifically , we highlight the importance of << diversified -LRB- faceted -RRB- summaries >> by combining three dimensions : [[ diversity ]] , uniqueness , and popularity .",144,1
145,"Specifically , we highlight the importance of diversified -LRB- faceted -RRB- summaries by combining three dimensions : [[ diversity ]] , << uniqueness >> , and popularity .",145,0
146,"Specifically , we highlight the importance of << diversified -LRB- faceted -RRB- summaries >> by combining three dimensions : diversity , [[ uniqueness ]] , and popularity .",146,1
147,"Specifically , we highlight the importance of diversified -LRB- faceted -RRB- summaries by combining three dimensions : diversity , [[ uniqueness ]] , and << popularity >> .",147,0
148,"Specifically , we highlight the importance of << diversified -LRB- faceted -RRB- summaries >> by combining three dimensions : diversity , uniqueness , and [[ popularity ]] .",148,1
149,"Our novel << diversity-aware entity summarization approach >> mimics [[ human conceptual clustering techniques ]] to group facts , and picks representative facts from each group to form concise -LRB- i.e. , short -RRB- and comprehensive -LRB- i.e. , improved coverage through diversity -RRB- summaries .",149,3
150,We evaluate our [[ approach ]] against the state-of-the-art techniques and show that our work improves both the quality and the efficiency of << entity summarization >> .,150,3
151,We evaluate our << approach >> against the [[ state-of-the-art techniques ]] and show that our work improves both the quality and the efficiency of entity summarization .,151,5
152,We evaluate our approach against the [[ state-of-the-art techniques ]] and show that our work improves both the quality and the efficiency of << entity summarization >> .,152,3
153,We evaluate our approach against the state-of-the-art techniques and show that our work improves both the [[ quality ]] and the efficiency of << entity summarization >> .,153,6
154,We evaluate our approach against the state-of-the-art techniques and show that our work improves both the quality and the [[ efficiency ]] of << entity summarization >> .,154,6
155,We present a [[ framework ]] for the << fast computation of lexical affinity models >> .,155,3
156,"The << framework >> is composed of a novel [[ algorithm ]] to efficiently compute the co-occurrence distribution between pairs of terms , an independence model , and a parametric affinity model .",156,4
157,"The framework is composed of a novel [[ algorithm ]] to efficiently compute the << co-occurrence distribution >> between pairs of terms , an independence model , and a parametric affinity model .",157,3
158,"The framework is composed of a novel [[ algorithm ]] to efficiently compute the co-occurrence distribution between pairs of terms , an << independence model >> , and a parametric affinity model .",158,0
159,"The << framework >> is composed of a novel algorithm to efficiently compute the co-occurrence distribution between pairs of terms , an [[ independence model ]] , and a parametric affinity model .",159,4
160,"The framework is composed of a novel algorithm to efficiently compute the co-occurrence distribution between pairs of terms , an [[ independence model ]] , and a << parametric affinity model >> .",160,0
161,"The << framework >> is composed of a novel algorithm to efficiently compute the co-occurrence distribution between pairs of terms , an independence model , and a [[ parametric affinity model ]] .",161,4
162,"In comparison with previous models , which either use arbitrary windows to compute similarity between words or use [[ lexical affinity ]] to create << sequential models >> , in this paper we focus on models intended to capture the co-occurrence patterns of any pair of words or phrases at any distance in the corpus .",162,3
163,"In comparison with previous << models >> , which either use arbitrary windows to compute similarity between words or use lexical affinity to create sequential models , in this paper we focus on [[ models ]] intended to capture the co-occurrence patterns of any pair of words or phrases at any distance in the corpus .",163,5
164,"In comparison with previous models , which either use arbitrary windows to compute similarity between words or use lexical affinity to create sequential models , in this paper we focus on [[ models ]] intended to capture the << co-occurrence patterns >> of any pair of words or phrases at any distance in the corpus .",164,3
165,"We apply [[ it ]] in combination with a terabyte corpus to answer << natural language tests >> , achieving encouraging results .",165,3
166,"We apply << it >> in combination with a [[ terabyte corpus ]] to answer natural language tests , achieving encouraging results .",166,6
167,This paper introduces a [[ system ]] for << categorizing unknown words >> .,167,3
168,The << system >> is based on a [[ multi-component architecture ]] where each component is responsible for identifying one class of unknown words .,168,3
169,The system is based on a << multi-component architecture >> where each [[ component ]] is responsible for identifying one class of unknown words .,169,4
170,The system is based on a multi-component architecture where each [[ component ]] is responsible for identifying one class of << unknown words >> .,170,3
171,The focus of this paper is the [[ components ]] that identify << names >> and spelling errors .,171,3
172,The focus of this paper is the [[ components ]] that identify names and << spelling errors >> .,172,3
173,The focus of this paper is the components that identify [[ names ]] and << spelling errors >> .,173,0
174,Each << component >> uses a [[ decision tree architecture ]] to combine multiple types of evidence about the unknown word .,174,3
175,The << system >> is evaluated using data from [[ live closed captions ]] - a genre replete with a wide variety of unknown words .,175,6
176,"At MIT Lincoln Laboratory , we have been developing a << Korean-to-English machine translation system >> [[ CCLINC -LRB- Common Coalition Language System at Lincoln Laboratory -RRB- ]] .",176,2
177,"The << CCLINC Korean-to-English translation system >> consists of two [[ core modules ]] , language understanding and generation modules mediated by a language neutral meaning representation called a semantic frame .",177,4
178,"The CCLINC Korean-to-English translation system consists of two core modules , << language understanding and generation modules >> mediated by a [[ language neutral meaning representation ]] called a semantic frame .",178,3
179,"The CCLINC Korean-to-English translation system consists of two core modules , language understanding and generation modules mediated by a << language neutral meaning representation >> called a [[ semantic frame ]] .",179,2
180,"The key features of the system include : -LRB- i -RRB- Robust efficient parsing of [[ Korean ]] -LRB- a << verb final language >> with overt case markers , relatively free word order , and frequent omissions of arguments -RRB- .",180,2
181,"The key features of the system include : -LRB- i -RRB- Robust efficient parsing of Korean -LRB- a << verb final language >> with [[ overt case markers ]] , relatively free word order , and frequent omissions of arguments -RRB- .",181,1
182,-LRB- ii -RRB- High quality << translation >> via [[ word sense disambiguation ]] and accurate word order generation of the target language .,182,3
183,-LRB- ii -RRB- High quality translation via [[ word sense disambiguation ]] and accurate << word order generation >> of the target language .,183,0
184,-LRB- ii -RRB- High quality << translation >> via word sense disambiguation and accurate [[ word order generation ]] of the target language .,184,3
185,"Having been trained on [[ Korean newspaper articles ]] on missiles and chemical biological warfare , the << system >> produces the translation output sufficient for content understanding of the original document .",185,3
186,"Having been trained on << Korean newspaper articles >> on [[ missiles and chemical biological warfare ]] , the system produces the translation output sufficient for content understanding of the original document .",186,1
187,"The [[ JAVELIN system ]] integrates a flexible , planning-based architecture with a variety of language processing modules to provide an << open-domain question answering capability >> on free text .",187,3
188,"The << JAVELIN system >> integrates a flexible , [[ planning-based architecture ]] with a variety of language processing modules to provide an open-domain question answering capability on free text .",188,4
189,"The << JAVELIN system >> integrates a flexible , planning-based architecture with a variety of [[ language processing modules ]] to provide an open-domain question answering capability on free text .",189,4
190,"The JAVELIN system integrates a flexible , << planning-based architecture >> with a variety of [[ language processing modules ]] to provide an open-domain question answering capability on free text .",190,0
191,We present the first application of the [[ head-driven statistical parsing model ]] of Collins -LRB- 1999 -RRB- as a << simultaneous language model >> and parser for large-vocabulary speech recognition .,191,3
192,We present the first application of the [[ head-driven statistical parsing model ]] of Collins -LRB- 1999 -RRB- as a simultaneous language model and << parser >> for large-vocabulary speech recognition .,192,3
193,We present the first application of the head-driven statistical parsing model of Collins -LRB- 1999 -RRB- as a [[ simultaneous language model ]] and << parser >> for large-vocabulary speech recognition .,193,0
194,We present the first application of the head-driven statistical parsing model of Collins -LRB- 1999 -RRB- as a [[ simultaneous language model ]] and parser for << large-vocabulary speech recognition >> .,194,3
195,We present the first application of the head-driven statistical parsing model of Collins -LRB- 1999 -RRB- as a simultaneous language model and [[ parser ]] for << large-vocabulary speech recognition >> .,195,3
196,"The [[ model ]] is adapted to an << online left to right chart-parser >> for word lattices , integrating acoustic , n-gram , and parser probabilities .",196,3
197,"The model is adapted to an [[ online left to right chart-parser ]] for << word lattices >> , integrating acoustic , n-gram , and parser probabilities .",197,3
198,"The model is adapted to an << online left to right chart-parser >> for word lattices , integrating [[ acoustic , n-gram , and parser probabilities ]] .",198,4
199,"The << parser >> uses [[ structural and lexical dependencies ]] not considered by n-gram models , conditioning recognition on more linguistically-grounded relationships .",199,3
200,Experiments on the [[ Wall Street Journal treebank ]] and << lattice corpora >> show word error rates competitive with the standard n-gram language model while extracting additional structural information useful for speech understanding .,200,0
201,Experiments on the [[ Wall Street Journal treebank ]] and lattice corpora show word error rates competitive with the standard << n-gram language model >> while extracting additional structural information useful for speech understanding .,201,6
202,Experiments on the Wall Street Journal treebank and [[ lattice corpora ]] show word error rates competitive with the standard << n-gram language model >> while extracting additional structural information useful for speech understanding .,202,6
203,Experiments on the Wall Street Journal treebank and lattice corpora show [[ word error rates ]] competitive with the standard << n-gram language model >> while extracting additional structural information useful for speech understanding .,203,6
204,Experiments on the Wall Street Journal treebank and lattice corpora show word error rates competitive with the standard n-gram language model while extracting additional [[ structural information ]] useful for << speech understanding >> .,204,3
205,[[ Image composition -LRB- or mosaicing -RRB- ]] has attracted a growing attention in recent years as one of the main elements in << video analysis and representation >> .,205,4
206,In this paper we deal with the problem of [[ global alignment ]] and << super-resolution >> .,206,0
207,We also propose to evaluate the quality of the resulting << mosaic >> by measuring the [[ amount of blurring ]] .,207,6
208,<< Global registration >> is achieved by combining a [[ graph-based technique ]] -- that exploits the topological structure of the sequence induced by the spatial overlap -- with a bundle adjustment which uses only the homographies computed in the previous steps .,208,3
209,Global registration is achieved by combining a [[ graph-based technique ]] -- that exploits the << topological structure >> of the sequence induced by the spatial overlap -- with a bundle adjustment which uses only the homographies computed in the previous steps .,209,3
210,Global registration is achieved by combining a [[ graph-based technique ]] -- that exploits the topological structure of the sequence induced by the spatial overlap -- with a << bundle adjustment >> which uses only the homographies computed in the previous steps .,210,0
211,<< Global registration >> is achieved by combining a graph-based technique -- that exploits the topological structure of the sequence induced by the spatial overlap -- with a [[ bundle adjustment ]] which uses only the homographies computed in the previous steps .,211,3
212,Global registration is achieved by combining a graph-based technique -- that exploits the topological structure of the sequence induced by the spatial overlap -- with a << bundle adjustment >> which uses only the [[ homographies ]] computed in the previous steps .,212,3
213,Experimental comparison with other << techniques >> shows the effectiveness of our [[ approach ]] .,213,5
214,The main of this project is << computer-assisted acquisition and morpho-syntactic description of verb-noun collocations >> in [[ Polish ]] .,214,3
215,"We present methodology and resources obtained in three main project << phases >> which are : [[ dictionary-based acquisition of collocation lexicon ]] , feasibility study for corpus-based lexicon enlargement phase , corpus-based lexicon enlargement and collocation description .",215,2
216,"We present methodology and resources obtained in three main project phases which are : [[ dictionary-based acquisition of collocation lexicon ]] , << feasibility study >> for corpus-based lexicon enlargement phase , corpus-based lexicon enlargement and collocation description .",216,0
217,"We present methodology and resources obtained in three main project << phases >> which are : dictionary-based acquisition of collocation lexicon , [[ feasibility study ]] for corpus-based lexicon enlargement phase , corpus-based lexicon enlargement and collocation description .",217,2
218,"We present methodology and resources obtained in three main project phases which are : dictionary-based acquisition of collocation lexicon , [[ feasibility study ]] for << corpus-based lexicon enlargement phase >> , corpus-based lexicon enlargement and collocation description .",218,3
219,"We present methodology and resources obtained in three main project << phases >> which are : dictionary-based acquisition of collocation lexicon , feasibility study for corpus-based lexicon enlargement phase , [[ corpus-based lexicon enlargement and collocation description ]] .",219,2
220,"We present methodology and resources obtained in three main project phases which are : dictionary-based acquisition of collocation lexicon , << feasibility study >> for corpus-based lexicon enlargement phase , [[ corpus-based lexicon enlargement and collocation description ]] .",220,0
221,The presented here [[ corpus-based approach ]] permitted us to triple the size the << verb-noun collocation dictionary >> for Polish .,221,3
222,The presented here corpus-based approach permitted us to triple the size the << verb-noun collocation dictionary >> for [[ Polish ]] .,222,1
223,"Along with the increasing requirements , the [[ hash-tag recommendation task ]] for << microblogs >> has been receiving considerable attention in recent years .",223,3
224,"Motivated by the successful use of [[ convolutional neural networks -LRB- CNNs -RRB- ]] for many << natural language processing tasks >> , in this paper , we adopt CNNs to perform the hashtag recommendation problem .",224,3
225,"To incorporate the << trigger words >> whose effectiveness have been experimentally evaluated in several previous works , we propose a novel [[ architecture ]] with an attention mechanism .",225,3
226,"To incorporate the trigger words whose effectiveness have been experimentally evaluated in several previous works , we propose a novel << architecture >> with an [[ attention mechanism ]] .",226,1
227,The results of experiments on the [[ data ]] collected from a real world microblogging service demonstrated that the proposed << model >> outperforms state-of-the-art methods .,227,6
228,The results of experiments on the data collected from a real world microblogging service demonstrated that the proposed [[ model ]] outperforms << state-of-the-art methods >> .,228,5
229,"By incorporating trigger words into the consideration , the relative improvement of the proposed [[ method ]] over the << state-of-the-art method >> is around 9.4 % in the F1-score .",229,5
230,"By incorporating trigger words into the consideration , the relative improvement of the proposed method over the << state-of-the-art method >> is around 9.4 % in the [[ F1-score ]] .",230,6
231,"In this paper , we improve an << unsupervised learning method >> using the [[ Expectation-Maximization -LRB- EM -RRB- algorithm ]] proposed by Nigam et al. for text classification problems in order to apply it to word sense disambiguation -LRB- WSD -RRB- problems .",231,3
232,"In this paper , we improve an unsupervised learning method using the [[ Expectation-Maximization -LRB- EM -RRB- algorithm ]] proposed by Nigam et al. for << text classification problems >> in order to apply it to word sense disambiguation -LRB- WSD -RRB- problems .",232,3
233,"In this paper , we improve an unsupervised learning method using the Expectation-Maximization -LRB- EM -RRB- algorithm proposed by Nigam et al. for text classification problems in order to apply [[ it ]] to << word sense disambiguation -LRB- WSD -RRB- problems >> .",233,3
234,"In experiments , we solved 50 noun WSD problems in the [[ Japanese Dictionary Task ]] in << SENSEVAL2 >> .",234,1
235,"Furthermore , our [[ methods ]] were confirmed to be effective also for << verb WSD problems >> .",235,3
236,"[[ Dividing sentences in chunks of words ]] is a useful preprocessing step for << parsing >> , information extraction and information retrieval .",236,3
237,"[[ Dividing sentences in chunks of words ]] is a useful preprocessing step for parsing , << information extraction >> and information retrieval .",237,3
238,"[[ Dividing sentences in chunks of words ]] is a useful preprocessing step for parsing , information extraction and << information retrieval >> .",238,3
239,"Dividing sentences in chunks of words is a useful preprocessing step for [[ parsing ]] , << information extraction >> and information retrieval .",239,0
240,"Dividing sentences in chunks of words is a useful preprocessing step for parsing , [[ information extraction ]] and << information retrieval >> .",240,0
241,"-LRB- Ramshaw and Marcus , 1995 -RRB- have introduced a `` convenient '' [[ data representation ]] for << chunking >> by converting it to a tagging task .",241,3
242,In this paper we will examine seven different [[ data representations ]] for the problem of << recognizing noun phrase chunks >> .,242,3
243,"However , equipped with the most suitable [[ data representation ]] , our << memory-based learning chunker >> was able to improve the best published chunking results for a standard data set .",243,3
244,"However , equipped with the most suitable data representation , our << memory-based learning chunker >> was able to improve the best published chunking results for a standard [[ data set ]] .",244,6
245,"We focus on << FAQ-like questions and answers >> , and build our [[ system ]] around a noisy-channel architecture which exploits both a language model for answers and a transformation model for answer/question terms , trained on a corpus of 1 million question/answer pairs collected from the Web .",245,3
246,"We focus on FAQ-like questions and answers , and build our << system >> around a [[ noisy-channel architecture ]] which exploits both a language model for answers and a transformation model for answer/question terms , trained on a corpus of 1 million question/answer pairs collected from the Web .",246,3
247,"We focus on FAQ-like questions and answers , and build our system around a [[ noisy-channel architecture ]] which exploits both a << language model >> for answers and a transformation model for answer/question terms , trained on a corpus of 1 million question/answer pairs collected from the Web .",247,3
248,"We focus on FAQ-like questions and answers , and build our system around a [[ noisy-channel architecture ]] which exploits both a language model for answers and a << transformation model >> for answer/question terms , trained on a corpus of 1 million question/answer pairs collected from the Web .",248,3
249,In this paper we evaluate four objective [[ measures of speech ]] with regards to << intelligibility prediction >> of synthesized speech in diverse noisy situations .,249,6
250,In this paper we evaluate four objective measures of speech with regards to << intelligibility prediction >> of [[ synthesized speech ]] in diverse noisy situations .,250,3
251,In this paper we evaluate four objective measures of speech with regards to intelligibility prediction of << synthesized speech >> in [[ diverse noisy situations ]] .,251,1
252,"We evaluated three [[ intel-ligibility measures ]] , the Dau measure , the glimpse proportion and the Speech Intelligibility Index -LRB- SII -RRB- and a << quality measure >> , the Perceptual Evaluation of Speech Quality -LRB- PESQ -RRB- .",252,0
253,"We evaluated three << intel-ligibility measures >> , the [[ Dau measure ]] , the glimpse proportion and the Speech Intelligibility Index -LRB- SII -RRB- and a quality measure , the Perceptual Evaluation of Speech Quality -LRB- PESQ -RRB- .",253,2
254,"We evaluated three intel-ligibility measures , the [[ Dau measure ]] , the << glimpse proportion >> and the Speech Intelligibility Index -LRB- SII -RRB- and a quality measure , the Perceptual Evaluation of Speech Quality -LRB- PESQ -RRB- .",254,0
255,"We evaluated three << intel-ligibility measures >> , the Dau measure , the [[ glimpse proportion ]] and the Speech Intelligibility Index -LRB- SII -RRB- and a quality measure , the Perceptual Evaluation of Speech Quality -LRB- PESQ -RRB- .",255,2
256,"We evaluated three intel-ligibility measures , the Dau measure , the [[ glimpse proportion ]] and the << Speech Intelligibility Index -LRB- SII -RRB- >> and a quality measure , the Perceptual Evaluation of Speech Quality -LRB- PESQ -RRB- .",256,0
257,"We evaluated three << intel-ligibility measures >> , the Dau measure , the glimpse proportion and the [[ Speech Intelligibility Index -LRB- SII -RRB- ]] and a quality measure , the Perceptual Evaluation of Speech Quality -LRB- PESQ -RRB- .",257,2
258,"We evaluated three intel-ligibility measures , the Dau measure , the glimpse proportion and the Speech Intelligibility Index -LRB- SII -RRB- and a << quality measure >> , the [[ Perceptual Evaluation of Speech Quality -LRB- PESQ -RRB- ]] .",258,2
259,For the << generation of synthesized speech >> we used a state of the art [[ HMM-based speech synthesis system ]] .,259,3
260,The << noisy conditions >> comprised four [[ additive noises ]] .,260,4
261,The [[ measures ]] were compared with << subjective intelligibility scores >> obtained in listening tests .,261,5
262,"The results show the [[ Dau ]] and the << glimpse measures >> to be the best predictors of intelligibility , with correlations of around 0.83 to subjective scores .",262,0
263,"The results show the [[ Dau ]] and the glimpse measures to be the best << predictors of intelligibility >> , with correlations of around 0.83 to subjective scores .",263,2
264,"The results show the [[ Dau ]] and the glimpse measures to be the best predictors of intelligibility , with correlations of around 0.83 to << subjective scores >> .",264,5
265,"The results show the Dau and the [[ glimpse measures ]] to be the best << predictors of intelligibility >> , with correlations of around 0.83 to subjective scores .",265,2
266,"The results show the Dau and the [[ glimpse measures ]] to be the best predictors of intelligibility , with correlations of around 0.83 to << subjective scores >> .",266,5
267,"The results show the << Dau >> and the glimpse measures to be the best predictors of intelligibility , with [[ correlations ]] of around 0.83 to subjective scores .",267,6
268,"The results show the Dau and the << glimpse measures >> to be the best predictors of intelligibility , with [[ correlations ]] of around 0.83 to subjective scores .",268,6
269,All [[ measures ]] gave less accurate << predictions of intelligibility >> for synthetic speech than have previously been found for natural speech ; in particular the SII measure .,269,6
270,All measures gave less accurate << predictions of intelligibility >> for [[ synthetic speech ]] than have previously been found for natural speech ; in particular the SII measure .,270,3
271,All measures gave less accurate predictions of intelligibility for [[ synthetic speech ]] than have previously been found for << natural speech >> ; in particular the SII measure .,271,5
272,All << measures >> gave less accurate predictions of intelligibility for synthetic speech than have previously been found for natural speech ; in particular the [[ SII measure ]] .,272,2
273,"In additional experiments , we processed the << synthesized speech >> by an [[ ideal binary mask ]] before adding noise .",273,3
274,The [[ Glimpse measure ]] gave the most accurate << intelligibility predictions >> in this situation .,274,3
275,"A [[ '' graphics for vision '' approach ]] is proposed to address the problem of << reconstruction >> from a large and imperfect data set : reconstruction on demand by tensor voting , or ROD-TV .",275,3
276,"A '' graphics for vision '' approach is proposed to address the problem of << reconstruction >> from a [[ large and imperfect data set ]] : reconstruction on demand by tensor voting , or ROD-TV .",276,3
277,"A '' graphics for vision '' approach is proposed to address the problem of reconstruction from a large and imperfect data set : << reconstruction >> on demand by [[ tensor voting ]] , or ROD-TV .",277,3
278,"A '' graphics for vision '' approach is proposed to address the problem of reconstruction from a large and imperfect data set : reconstruction on demand by [[ tensor voting ]] , or << ROD-TV >> .",278,0
279,"A '' graphics for vision '' approach is proposed to address the problem of reconstruction from a large and imperfect data set : << reconstruction >> on demand by tensor voting , or [[ ROD-TV ]] .",279,3
280,"<< ROD-TV >> simultaneously delivers good [[ efficiency ]] and robust-ness , by adapting to a continuum of primitive connectivity , view dependence , and levels of detail -LRB- LOD -RRB- .",280,6
281,"<< ROD-TV >> simultaneously delivers good efficiency and [[ robust-ness ]] , by adapting to a continuum of primitive connectivity , view dependence , and levels of detail -LRB- LOD -RRB- .",281,6
282,"ROD-TV simultaneously delivers good << efficiency >> and [[ robust-ness ]] , by adapting to a continuum of primitive connectivity , view dependence , and levels of detail -LRB- LOD -RRB- .",282,0
283,"ROD-TV simultaneously delivers good efficiency and robust-ness , by adapting to a continuum of << primitive connectivity >> , [[ view dependence ]] , and levels of detail -LRB- LOD -RRB- .",283,0
284,"ROD-TV simultaneously delivers good efficiency and robust-ness , by adapting to a continuum of primitive connectivity , << view dependence >> , and [[ levels of detail -LRB- LOD -RRB- ]] .",284,0
285,[[ Locally inferred surface elements ]] are robust to noise and better capture << local shapes >> .,285,3
286,"By inferring [[ per-vertex normals ]] at sub-voxel precision on the fly , we can achieve << interpolative shading >> .",286,3
287,"By inferring << per-vertex normals >> at [[ sub-voxel precision ]] on the fly , we can achieve interpolative shading .",287,1
288,"By relaxing the [[ mesh connectivity requirement ]] , we extend ROD-TV and propose a simple but effective << multiscale feature extraction algorithm >> .",288,3
289,"By relaxing the mesh connectivity requirement , we extend [[ ROD-TV ]] and propose a simple but effective << multiscale feature extraction algorithm >> .",289,3
290,<< ROD-TV >> consists of a [[ hierarchical data structure ]] that encodes different levels of detail .,290,4
291,The << local reconstruction algorithm >> is [[ tensor voting ]] .,291,2
292,"<< It >> is applied on demand to the visible subset of data at a desired level of detail , by [[ traversing the data hierarchy ]] and collecting tensorial support in a neighborhood .",292,3
293,"It is applied on demand to the visible subset of data at a desired level of detail , by [[ traversing the data hierarchy ]] and << collecting tensorial support >> in a neighborhood .",293,0
294,"<< It >> is applied on demand to the visible subset of data at a desired level of detail , by traversing the data hierarchy and [[ collecting tensorial support ]] in a neighborhood .",294,3
295,Both [[ rhetorical structure ]] and << punctuation >> have been helpful in discourse processing .,295,0
296,Both [[ rhetorical structure ]] and punctuation have been helpful in << discourse processing >> .,296,3
297,Both rhetorical structure and [[ punctuation ]] have been helpful in << discourse processing >> .,297,3
298,"Based on a corpus annotation project , this paper reports the discursive usage of 6 [[ Chinese punctuation marks ]] in << news commentary texts >> : Colon , Dash , Ellipsis , Exclamation Mark , Question Mark , and Semicolon .",298,4
299,"Based on a corpus annotation project , this paper reports the discursive usage of 6 << Chinese punctuation marks >> in news commentary texts : [[ Colon ]] , Dash , Ellipsis , Exclamation Mark , Question Mark , and Semicolon .",299,2
300,"Based on a corpus annotation project , this paper reports the discursive usage of 6 Chinese punctuation marks in news commentary texts : [[ Colon ]] , << Dash >> , Ellipsis , Exclamation Mark , Question Mark , and Semicolon .",300,0
301,"Based on a corpus annotation project , this paper reports the discursive usage of 6 << Chinese punctuation marks >> in news commentary texts : Colon , [[ Dash ]] , Ellipsis , Exclamation Mark , Question Mark , and Semicolon .",301,2
302,"Based on a corpus annotation project , this paper reports the discursive usage of 6 Chinese punctuation marks in news commentary texts : Colon , [[ Dash ]] , << Ellipsis >> , Exclamation Mark , Question Mark , and Semicolon .",302,0
303,"Based on a corpus annotation project , this paper reports the discursive usage of 6 << Chinese punctuation marks >> in news commentary texts : Colon , Dash , [[ Ellipsis ]] , Exclamation Mark , Question Mark , and Semicolon .",303,2
304,"Based on a corpus annotation project , this paper reports the discursive usage of 6 Chinese punctuation marks in news commentary texts : Colon , Dash , [[ Ellipsis ]] , << Exclamation Mark >> , Question Mark , and Semicolon .",304,0
305,"Based on a corpus annotation project , this paper reports the discursive usage of 6 << Chinese punctuation marks >> in news commentary texts : Colon , Dash , Ellipsis , [[ Exclamation Mark ]] , Question Mark , and Semicolon .",305,2
306,"Based on a corpus annotation project , this paper reports the discursive usage of 6 Chinese punctuation marks in news commentary texts : Colon , Dash , Ellipsis , [[ Exclamation Mark ]] , << Question Mark >> , and Semicolon .",306,0
307,"Based on a corpus annotation project , this paper reports the discursive usage of 6 << Chinese punctuation marks >> in news commentary texts : Colon , Dash , Ellipsis , Exclamation Mark , [[ Question Mark ]] , and Semicolon .",307,2
308,"Based on a corpus annotation project , this paper reports the discursive usage of 6 Chinese punctuation marks in news commentary texts : Colon , Dash , Ellipsis , Exclamation Mark , [[ Question Mark ]] , and << Semicolon >> .",308,0
309,"Based on a corpus annotation project , this paper reports the discursive usage of 6 << Chinese punctuation marks >> in news commentary texts : Colon , Dash , Ellipsis , Exclamation Mark , Question Mark , and [[ Semicolon ]] .",309,2
310,The [[ rhetorical patterns ]] of these << marks >> are compared against patterns around cue phrases in general .,310,1
311,The [[ rhetorical patterns ]] of these marks are compared against << patterns around cue phrases >> in general .,311,5
312,"Results show that these [[ Chinese punctuation marks ]] , though fewer in number than << cue phrases >> , are easy to identify , have strong correlation with certain relations , and can be used as distinctive indicators of nuclearity in Chinese texts .",312,5
313,"Results show that these [[ Chinese punctuation marks ]] , though fewer in number than cue phrases , are easy to identify , have strong correlation with certain relations , and can be used as distinctive << indicators of nuclearity >> in Chinese texts .",313,3
314,"Results show that these Chinese punctuation marks , though fewer in number than cue phrases , are easy to identify , have strong correlation with certain relations , and can be used as distinctive << indicators of nuclearity >> in [[ Chinese texts ]] .",314,1
315,The << features >> based on [[ Markov random field -LRB- MRF -RRB- models ]] are usually sensitive to the rotation of image textures .,315,3
316,This paper develops an [[ anisotropic circular Gaussian MRF -LRB- ACGMRF -RRB- model ]] for << modelling rotated image textures >> and retrieving rotation-invariant texture features .,316,3
317,This paper develops an [[ anisotropic circular Gaussian MRF -LRB- ACGMRF -RRB- model ]] for modelling rotated image textures and << retrieving rotation-invariant texture features >> .,317,3
318,This paper develops an anisotropic circular Gaussian MRF -LRB- ACGMRF -RRB- model for [[ modelling rotated image textures ]] and << retrieving rotation-invariant texture features >> .,318,0
319,"To overcome the [[ singularity problem ]] of the << least squares estimate -LRB- LSE -RRB- method >> , an approximate least squares estimate -LRB- ALSE -RRB- method is proposed to estimate the parameters of the ACGMRF model .",319,1
320,"To overcome the singularity problem of the least squares estimate -LRB- LSE -RRB- method , an [[ approximate least squares estimate -LRB- ALSE -RRB- method ]] is proposed to estimate the << parameters of the ACGMRF model >> .",320,3
321,The << rotation-invariant features >> can be obtained from the [[ parameters of the ACGMRF model ]] by the one-dimensional -LRB- 1-D -RRB- discrete Fourier transform -LRB- DFT -RRB- .,321,3
322,The << rotation-invariant features >> can be obtained from the parameters of the ACGMRF model by the [[ one-dimensional -LRB- 1-D -RRB- discrete Fourier transform -LRB- DFT -RRB- ]] .,322,3
323,Significantly improved accuracy can be achieved by applying the [[ rotation-invariant features ]] to classify << SAR -LRB- synthetic aperture radar >> -RRB- sea ice and Brodatz imagery .,323,3
324,"Despite much recent progress on accurate << semantic role labeling >> , previous work has largely used [[ independent classifiers ]] , possibly combined with separate label sequence models via Viterbi decoding .",324,3
325,"Despite much recent progress on accurate semantic role labeling , previous work has largely used [[ independent classifiers ]] , possibly combined with separate << label sequence models >> via Viterbi decoding .",325,0
326,"Despite much recent progress on accurate semantic role labeling , previous work has largely used independent classifiers , possibly combined with separate << label sequence models >> via [[ Viterbi decoding ]] .",326,3
327,"We show how to build a joint model of argument frames , incorporating novel [[ features ]] that model these interactions into << discriminative log-linear models >> .",327,4
328,This << system >> achieves an [[ error reduction ]] of 22 % on all arguments and 32 % on core arguments over a state-of-the art independent classifier for gold-standard parse trees on PropBank .,328,6
329,This system achieves an [[ error reduction ]] of 22 % on all arguments and 32 % on core arguments over a state-of-the art << independent classifier >> for gold-standard parse trees on PropBank .,329,6
330,This << system >> achieves an error reduction of 22 % on all arguments and 32 % on core arguments over a state-of-the art [[ independent classifier ]] for gold-standard parse trees on PropBank .,330,5
331,This << system >> achieves an error reduction of 22 % on all arguments and 32 % on core arguments over a state-of-the art independent classifier for [[ gold-standard parse trees ]] on PropBank .,331,6
332,This system achieves an error reduction of 22 % on all arguments and 32 % on core arguments over a state-of-the art << independent classifier >> for [[ gold-standard parse trees ]] on PropBank .,332,6
333,This system achieves an error reduction of 22 % on all arguments and 32 % on core arguments over a state-of-the art independent classifier for [[ gold-standard parse trees ]] on << PropBank >> .,333,4
334,"In order to deal with << ambiguity >> , the [[ MORphological PArser MORPA ]] is provided with a probabilistic context-free grammar -LRB- PCFG -RRB- , i.e. it combines a `` conventional '' context-free morphological grammar to filter out ungrammatical segmentations with a probability-based scoring function which determines the likelihood of each successful parse .",334,3
335,"In order to deal with ambiguity , the << MORphological PArser MORPA >> is provided with a [[ probabilistic context-free grammar -LRB- PCFG -RRB- ]] , i.e. it combines a `` conventional '' context-free morphological grammar to filter out ungrammatical segmentations with a probability-based scoring function which determines the likelihood of each successful parse .",335,3
336,"In order to deal with ambiguity , the MORphological PArser MORPA is provided with a probabilistic context-free grammar -LRB- PCFG -RRB- , i.e. << it >> combines a [[ `` conventional '' context-free morphological grammar ]] to filter out ungrammatical segmentations with a probability-based scoring function which determines the likelihood of each successful parse .",336,3
337,"In order to deal with ambiguity , the MORphological PArser MORPA is provided with a probabilistic context-free grammar -LRB- PCFG -RRB- , i.e. it combines a [[ `` conventional '' context-free morphological grammar ]] to filter out << ungrammatical segmentations >> with a probability-based scoring function which determines the likelihood of each successful parse .",337,3
338,"In order to deal with ambiguity , the MORphological PArser MORPA is provided with a probabilistic context-free grammar -LRB- PCFG -RRB- , i.e. << it >> combines a `` conventional '' context-free morphological grammar to filter out ungrammatical segmentations with a [[ probability-based scoring function ]] which determines the likelihood of each successful parse .",338,3
339,"In order to deal with ambiguity , the MORphological PArser MORPA is provided with a probabilistic context-free grammar -LRB- PCFG -RRB- , i.e. it combines a << `` conventional '' context-free morphological grammar >> to filter out ungrammatical segmentations with a [[ probability-based scoring function ]] which determines the likelihood of each successful parse .",339,0
340,"In order to deal with ambiguity , the MORphological PArser MORPA is provided with a probabilistic context-free grammar -LRB- PCFG -RRB- , i.e. it combines a `` conventional '' context-free morphological grammar to filter out ungrammatical segmentations with a [[ probability-based scoring function ]] which determines the likelihood of each successful << parse >> .",340,3
341,Test performance data will show that a [[ PCFG ]] yields good results in << morphological parsing >> .,341,3
342,[[ MORPA ]] is a fully implemented << parser >> developed for use in a text-to-speech conversion system .,342,2
343,[[ MORPA ]] is a fully implemented parser developed for use in a << text-to-speech conversion system >> .,343,3
344,MORPA is a fully implemented [[ parser ]] developed for use in a << text-to-speech conversion system >> .,344,3
345,This paper describes the framework of a << Korean phonological knowledge base system >> using the [[ unification-based grammar formalism ]] : Korean Phonology Structure Grammar -LRB- KPSG -RRB- .,345,3
346,This paper describes the framework of a Korean phonological knowledge base system using the << unification-based grammar formalism >> : [[ Korean Phonology Structure Grammar -LRB- KPSG -RRB- ]] .,346,2
347,The [[ approach ]] of << KPSG >> provides an explicit development model for constructing a computational phonological system : speech recognition and synthesis system .,347,3
348,The approach of [[ KPSG ]] provides an explicit development model for constructing a computational << phonological system >> : speech recognition and synthesis system .,348,3
349,We show that the proposed [[ approach ]] is more describable than other << approaches >> such as those employing a traditional generative phonological approach .,349,5
350,We show that the proposed approach is more describable than other approaches such as << those >> employing a traditional [[ generative phonological approach ]] .,350,3
351,"In this paper , we study the [[ design of core-selecting payment rules ]] for such << domains >> .",351,3
352,We design two [[ core-selecting rules ]] that always satisfy << IR >> in expectation .,352,3
353,To study the performance of our << rules >> we perform a [[ computational Bayes-Nash equilibrium analysis ]] .,353,3
354,"We show that , in equilibrium , our new [[ rules ]] have better incentives , higher efficiency , and a lower rate of ex-post IR violations than standard << core-selecting rules >> .",354,5
355,"We show that , in equilibrium , our new << rules >> have better incentives , higher efficiency , and a lower [[ rate of ex-post IR violations ]] than standard core-selecting rules .",355,6
356,"We show that , in equilibrium , our new rules have better incentives , higher efficiency , and a lower [[ rate of ex-post IR violations ]] than standard << core-selecting rules >> .",356,6
357,"In this paper , we will describe a [[ search tool ]] for a huge set of << ngrams >> .",357,3
358,This system can be a very useful [[ tool ]] for << linguistic knowledge discovery >> and other NLP tasks .,358,3
359,This system can be a very useful [[ tool ]] for linguistic knowledge discovery and other << NLP tasks >> .,359,3
360,This system can be a very useful tool for [[ linguistic knowledge discovery ]] and other << NLP tasks >> .,360,0
361,This paper explores the role of [[ user modeling ]] in such << systems >> .,361,4
362,"Since acquiring the knowledge for a [[ user model ]] is a fundamental problem in << user modeling >> , a section is devoted to this topic .",362,3
363,"Next , the benefits and costs of implementing a [[ user modeling component ]] for a << system >> are weighed in light of several aspects of the interaction requirements that may be imposed by the system .",363,4
364,"[[ Information extraction techniques ]] automatically create << structured databases >> from unstructured data sources , such as the Web or newswire documents .",364,3
365,"<< Information extraction techniques >> automatically create structured databases from [[ unstructured data sources ]] , such as the Web or newswire documents .",365,3
366,"Information extraction techniques automatically create structured databases from << unstructured data sources >> , such as the [[ Web ]] or newswire documents .",366,2
367,"Information extraction techniques automatically create structured databases from unstructured data sources , such as the [[ Web ]] or << newswire documents >> .",367,0
368,"Information extraction techniques automatically create structured databases from << unstructured data sources >> , such as the Web or [[ newswire documents ]] .",368,2
369,"Despite the successes of these << systems >> , [[ accuracy ]] will always be imperfect .",369,6
370,"The << information extraction system >> we evaluate is based on a [[ linear-chain conditional random field -LRB- CRF -RRB- ]] , a probabilistic model which has performed well on information extraction tasks because of its ability to capture arbitrary , overlapping features of the input in a Markov model .",370,3
371,"The information extraction system we evaluate is based on a [[ linear-chain conditional random field -LRB- CRF -RRB- ]] , a << probabilistic model >> which has performed well on information extraction tasks because of its ability to capture arbitrary , overlapping features of the input in a Markov model .",371,2
372,"The information extraction system we evaluate is based on a linear-chain conditional random field -LRB- CRF -RRB- , a [[ probabilistic model ]] which has performed well on << information extraction tasks >> because of its ability to capture arbitrary , overlapping features of the input in a Markov model .",372,3
373,"The information extraction system we evaluate is based on a linear-chain conditional random field -LRB- CRF -RRB- , a [[ probabilistic model ]] which has performed well on information extraction tasks because of its ability to capture << arbitrary , overlapping features >> of the input in a Markov model .",373,3
374,"The information extraction system we evaluate is based on a linear-chain conditional random field -LRB- CRF -RRB- , a probabilistic model which has performed well on information extraction tasks because of its ability to capture [[ arbitrary , overlapping features ]] of the << input >> in a Markov model .",374,1
375,"The information extraction system we evaluate is based on a linear-chain conditional random field -LRB- CRF -RRB- , a probabilistic model which has performed well on information extraction tasks because of its ability to capture [[ arbitrary , overlapping features ]] of the input in a << Markov model >> .",375,4
376,"We implement several techniques to estimate the confidence of both [[ extracted fields ]] and entire << multi-field records >> , obtaining an average precision of 98 % for retrieving correct fields and 87 % for multi-field records .",376,0
377,"We implement several << techniques >> to estimate the confidence of both extracted fields and entire multi-field records , obtaining an [[ average precision ]] of 98 % for retrieving correct fields and 87 % for multi-field records .",377,6
378,"In this paper , we use the [[ information redundancy in multilingual input ]] to correct errors in << machine translation >> and thus improve the quality of multilingual summaries .",378,3
379,"In this paper , we use the [[ information redundancy in multilingual input ]] to correct errors in machine translation and thus improve the quality of << multilingual summaries >> .",379,3
380,"We demonstrate how errors in the << machine translations >> of the input [[ Arabic documents ]] can be corrected by identifying and generating from such redundancy , focusing on noun phrases .",380,3
381,"In this paper , we propose a new [[ approach ]] to generate << oriented object proposals -LRB- OOPs -RRB- >> to reduce the detection error caused by various orientations of the object .",381,3
382,"In this paper , we propose a new approach to generate << oriented object proposals -LRB- OOPs -RRB- >> to reduce the [[ detection error ]] caused by various orientations of the object .",382,6
383,"To this end , we propose to efficiently locate << object regions >> according to [[ pixelwise object probability ]] , rather than measuring the objectness from a set of sampled windows .",383,3
384,"To this end , we propose to efficiently locate object regions according to [[ pixelwise object probability ]] , rather than measuring the << objectness >> from a set of sampled windows .",384,5
385,"We formulate the << proposal generation problem >> as a [[ generative proba-bilistic model ]] such that object proposals of different shapes -LRB- i.e. , sizes and orientations -RRB- can be produced by locating the local maximum likelihoods .",385,3
386,"We formulate the proposal generation problem as a generative proba-bilistic model such that << object proposals >> of different [[ shapes ]] -LRB- i.e. , sizes and orientations -RRB- can be produced by locating the local maximum likelihoods .",386,1
387,"We formulate the proposal generation problem as a generative proba-bilistic model such that object proposals of different << shapes >> -LRB- i.e. , [[ sizes ]] and orientations -RRB- can be produced by locating the local maximum likelihoods .",387,2
388,"We formulate the proposal generation problem as a generative proba-bilistic model such that object proposals of different shapes -LRB- i.e. , [[ sizes ]] and << orientations >> -RRB- can be produced by locating the local maximum likelihoods .",388,0
389,"We formulate the proposal generation problem as a generative proba-bilistic model such that object proposals of different << shapes >> -LRB- i.e. , sizes and [[ orientations ]] -RRB- can be produced by locating the local maximum likelihoods .",389,2
390,"We formulate the proposal generation problem as a generative proba-bilistic model such that << object proposals >> of different shapes -LRB- i.e. , sizes and orientations -RRB- can be produced by locating the [[ local maximum likelihoods ]] .",390,3
391,"First , it helps the [[ object detector ]] handle objects of different << orientations >> .",391,3
392,"Third , [[ it ]] avoids massive window sampling , and thereby reducing the << number of proposals >> while maintaining a high recall .",392,3
393,"Third , << it >> avoids massive window sampling , and thereby reducing the number of proposals while maintaining a high [[ recall ]] .",393,6
394,Experiments on the [[ PASCAL VOC 2007 dataset ]] show that the proposed << OOP >> outperforms the state-of-the-art fast methods .,394,6
395,Experiments on the PASCAL VOC 2007 dataset show that the proposed [[ OOP ]] outperforms the << state-of-the-art fast methods >> .,395,5
396,Further experiments show that the [[ rotation invariant property ]] helps a << class-specific object detector >> achieve better performance than the state-of-the-art proposal generation methods in either object rotation scenarios or general scenarios .,396,3
397,Further experiments show that the rotation invariant property helps a [[ class-specific object detector ]] achieve better performance than the state-of-the-art << proposal generation methods >> in either object rotation scenarios or general scenarios .,397,5
398,Further experiments show that the rotation invariant property helps a << class-specific object detector >> achieve better performance than the state-of-the-art proposal generation methods in either [[ object rotation scenarios ]] or general scenarios .,398,6
399,Further experiments show that the rotation invariant property helps a class-specific object detector achieve better performance than the state-of-the-art << proposal generation methods >> in either [[ object rotation scenarios ]] or general scenarios .,399,6
400,Further experiments show that the rotation invariant property helps a class-specific object detector achieve better performance than the state-of-the-art proposal generation methods in either [[ object rotation scenarios ]] or << general scenarios >> .,400,0
401,Further experiments show that the rotation invariant property helps a << class-specific object detector >> achieve better performance than the state-of-the-art proposal generation methods in either object rotation scenarios or [[ general scenarios ]] .,401,6
402,Further experiments show that the rotation invariant property helps a class-specific object detector achieve better performance than the state-of-the-art << proposal generation methods >> in either object rotation scenarios or [[ general scenarios ]] .,402,6
403,"This paper describes three relatively [[ domain-independent capabilities ]] recently added to the << Paramax spoken language understanding system >> : non-monotonic reasoning , implicit reference resolution , and database query paraphrase .",403,4
404,"This paper describes three relatively << domain-independent capabilities >> recently added to the Paramax spoken language understanding system : [[ non-monotonic reasoning ]] , implicit reference resolution , and database query paraphrase .",404,2
405,"This paper describes three relatively << domain-independent capabilities >> recently added to the Paramax spoken language understanding system : non-monotonic reasoning , [[ implicit reference resolution ]] , and database query paraphrase .",405,2
406,"This paper describes three relatively << domain-independent capabilities >> recently added to the Paramax spoken language understanding system : non-monotonic reasoning , implicit reference resolution , and [[ database query paraphrase ]] .",406,2
407,"Finally , we briefly describe an experiment which we have done in extending the << n-best speech/language integration architecture >> to improving [[ OCR accuracy ]] .",407,6
408,"We investigate the problem of fine-grained sketch-based image retrieval -LRB- SBIR -RRB- , where [[ free-hand human sketches ]] are used as queries to perform << instance-level retrieval of images >> .",408,3
409,"This is an extremely challenging task because -LRB- i -RRB- visual comparisons not only need to be fine-grained but also executed cross-domain , -LRB- ii -RRB- free-hand -LRB- finger -RRB- sketches are highly abstract , making fine-grained matching harder , and most importantly -LRB- iii -RRB- [[ annotated cross-domain sketch-photo datasets ]] required for training are scarce , challenging many state-of-the-art << machine learning techniques >> .",409,3
410,We then develop a [[ deep triplet-ranking model ]] for << instance-level SBIR >> with a novel data augmentation and staged pre-training strategy to alleviate the issue of insufficient fine-grained training data .,410,3
411,We then develop a [[ deep triplet-ranking model ]] for instance-level SBIR with a novel data augmentation and staged pre-training strategy to alleviate the issue of << insufficient fine-grained training data >> .,411,3
412,We then develop a << deep triplet-ranking model >> for instance-level SBIR with a novel [[ data augmentation ]] and staged pre-training strategy to alleviate the issue of insufficient fine-grained training data .,412,3
413,We then develop a deep triplet-ranking model for instance-level SBIR with a novel [[ data augmentation ]] and << staged pre-training strategy >> to alleviate the issue of insufficient fine-grained training data .,413,0
414,We then develop a << deep triplet-ranking model >> for instance-level SBIR with a novel data augmentation and [[ staged pre-training strategy ]] to alleviate the issue of insufficient fine-grained training data .,414,3
415,Extensive experiments are carried out to contribute a variety of insights into the challenges of [[ data sufficiency ]] and << over-fitting avoidance >> when training deep networks for fine-grained cross-domain ranking tasks .,415,0
416,Extensive experiments are carried out to contribute a variety of insights into the challenges of data sufficiency and over-fitting avoidance when training [[ deep networks ]] for << fine-grained cross-domain ranking tasks >> .,416,3
417,In this paper we target at generating << generic action proposals >> in [[ unconstrained videos ]] .,417,3
418,"Each action proposal corresponds to a << temporal series of spatial bounding boxes >> , i.e. , a [[ spatio-temporal video tube ]] , which has a good potential to locate one human action .",418,2
419,"Each action proposal corresponds to a temporal series of spatial bounding boxes , i.e. , a [[ spatio-temporal video tube ]] , which has a good potential to locate one << human action >> .",419,3
420,"Assuming each action is performed by a human with meaningful motion , both [[ appearance and motion cues ]] are utilized to measure the << ac-tionness >> of the video tubes .",420,3
421,"Assuming each action is performed by a human with meaningful motion , both appearance and motion cues are utilized to measure the [[ ac-tionness ]] of the << video tubes >> .",421,6
422,"After picking those spatiotem-poral paths of high actionness scores , our << action proposal generation >> is formulated as a [[ maximum set coverage problem ]] , where greedy search is performed to select a set of action proposals that can maximize the overall actionness score .",422,3
423,"After picking those spatiotem-poral paths of high actionness scores , our action proposal generation is formulated as a maximum set coverage problem , where [[ greedy search ]] is performed to select a set of << action proposals >> that can maximize the overall actionness score .",423,3
424,"After picking those spatiotem-poral paths of high actionness scores , our action proposal generation is formulated as a maximum set coverage problem , where greedy search is performed to select a set of << action proposals >> that can maximize the overall [[ actionness score ]] .",424,6
425,"Compared with existing [[ action proposal approaches ]] , our << action proposals >> do not rely on video segmentation and can be generated in nearly real-time .",425,5
426,"Experimental results on two challenging [[ datasets ]] , MSRII and UCF 101 , validate the superior performance of our << action proposals >> as well as competitive results on action detection and search .",426,6
427,"Experimental results on two challenging << datasets >> , [[ MSRII ]] and UCF 101 , validate the superior performance of our action proposals as well as competitive results on action detection and search .",427,2
428,"Experimental results on two challenging datasets , [[ MSRII ]] and << UCF 101 >> , validate the superior performance of our action proposals as well as competitive results on action detection and search .",428,0
429,"Experimental results on two challenging << datasets >> , MSRII and [[ UCF 101 ]] , validate the superior performance of our action proposals as well as competitive results on action detection and search .",429,2
430,"Experimental results on two challenging datasets , MSRII and UCF 101 , validate the superior performance of our << action proposals >> as well as competitive results on [[ action detection and search ]] .",430,6
431,This paper reports recent research into [[ methods ]] for << creating natural language text >> .,431,3
432,"<< KDS -LRB- Knowledge Delivery System -RRB- >> , which embodies this [[ paradigm ]] , has distinct parts devoted to creation of the propositional units , to organization of the text , to prevention of excess redundancy , to creation of combinations of units , to evaluation of these combinations as potential sentences , to selection of the best among competing combinations , and to creation of the final text .",432,4
433,The Fragment-and-Compose paradigm and the [[ computational methods ]] of << KDS >> are described .,433,3
434,This paper explores the issue of using different [[ co-occurrence similarities ]] between terms for separating << query terms >> that are useful for retrieval from those that are harmful .,434,3
435,This paper explores the issue of using different co-occurrence similarities between terms for separating [[ query terms ]] that are useful for << retrieval >> from those that are harmful .,435,3
436,This paper explores the issue of using different co-occurrence similarities between terms for separating << query terms >> that are useful for retrieval from [[ those ]] that are harmful .,436,5
437,The hypothesis under examination is that [[ useful terms ]] tend to be more similar to each other than to other << query terms >> .,437,5
438,Preliminary experiments with << similarities >> computed using [[ first-order and second-order co-occurrence ]] seem to confirm the hypothesis .,438,3
439,"We propose a new [[ phrase-based translation model ]] and << decoding algorithm >> that enables us to evaluate and compare several , previously proposed phrase-based translation models .",439,0
440,"Within our framework , we carry out a large number of experiments to understand better and explain why [[ phrase-based models ]] outperform << word-based models >> .",440,5
441,"Our empirical results , which hold for all examined language pairs , suggest that the highest levels of performance can be obtained through relatively simple << means >> : [[ heuristic learning of phrase translations ]] from word-based alignments and lexical weighting of phrase translations .",441,2
442,"Our empirical results , which hold for all examined language pairs , suggest that the highest levels of performance can be obtained through relatively simple means : << heuristic learning of phrase translations >> from [[ word-based alignments ]] and lexical weighting of phrase translations .",442,3
443,"Our empirical results , which hold for all examined language pairs , suggest that the highest levels of performance can be obtained through relatively simple << means >> : heuristic learning of phrase translations from word-based alignments and [[ lexical weighting of phrase translations ]] .",443,2
444,"Traditional [[ methods ]] for << color constancy >> can improve surface re-flectance estimates from such uncalibrated images , but their output depends significantly on the background scene .",444,3
445,"Traditional [[ methods ]] for color constancy can improve << surface re-flectance estimates >> from such uncalibrated images , but their output depends significantly on the background scene .",445,3
446,"Traditional methods for color constancy can improve << surface re-flectance estimates >> from such [[ uncalibrated images ]] , but their output depends significantly on the background scene .",446,3
447,"We introduce the multi-view color constancy problem , and present a [[ method ]] to recover << estimates of underlying surface re-flectance >> based on joint estimation of these surface properties and the illuminants present in multiple images .",447,3
448,"The [[ method ]] can exploit << image correspondences >> obtained by various alignment techniques , and we show examples based on matching local region features .",448,3
449,"The method can exploit << image correspondences >> obtained by various [[ alignment techniques ]] , and we show examples based on matching local region features .",449,3
450,Our results show that [[ multi-view constraints ]] can significantly improve << estimates of both scene illuminants and object color -LRB- surface reflectance -RRB- >> when compared to a baseline single-view method .,450,3
451,Our results show that << multi-view constraints >> can significantly improve estimates of both scene illuminants and object color -LRB- surface reflectance -RRB- when compared to a [[ baseline single-view method ]] .,451,5
452,"Our contributions include a [[ concise , modular architecture ]] with reversible processes of << understanding >> and generation , an information-state model of reference , and flexible links between semantics and collaborative problem solving .",452,3
453,"Our contributions include a [[ concise , modular architecture ]] with reversible processes of understanding and << generation >> , an information-state model of reference , and flexible links between semantics and collaborative problem solving .",453,3
454,"Our contributions include a concise , modular architecture with reversible processes of [[ understanding ]] and << generation >> , an information-state model of reference , and flexible links between semantics and collaborative problem solving .",454,0