Leveraging Annotators’ Gaze Behaviour for Coreference Resolution

This paper aims at utilizing cognitive information obtained from the eye movements behavior of annotators for automatic coreference resolution. We first record eye-movement behavior of multiple annotators resolving coreferences in 22 documents selected from MUC dataset. By inspecting the gaze-regression profiles of our participants, we observe how regressive saccades account for selection of potential antecedents for a certain anaphoric mention. Based on this observation, we then propose a heuristic to utilize gaze data to prune mention pairs in mention-pair model, a popular paradigm for automatic coreference resolution. Consistent improvement in accuracy across several classifiers is observed with our heuristic, demonstrating why cognitive data can be useful for a difficult task like coreference resolution.

[1]  K. Rayner,et al.  Eye movements in reading: Psycholinguistic studies. , 1994 .

[2]  Pushpak Bhattacharyya,et al.  Automatically Predicting Sentence Translation Difficulty , 2013, ACL.

[3]  Hwee Tou Ng,et al.  A Machine Learning Approach to Coreference Resolution of Noun Phrases , 2001, CL.

[4]  Dan Klein,et al.  Easy Victories and Uphill Battles in Coreference Resolution , 2013, EMNLP.

[5]  Noam Chomsky.  Some Concepts and Consequences of the Theory of Government and Binding , 1982 .

[6]  Claire Gardent,et al.  Improving Machine Learning Approaches to Coreference Resolution , 2002, ACL.

[7]  Branimir Boguraev,et al.  Anaphora for Everyone: Pronominal Anaphora Resolution without a Parser , 1996, COLING.

[8]  Pushpak Bhattacharyya,et al.  More than meets the eye: Study of Human Cognition in Sense Annotation , 2013, NAACL.

[9]  P. Elango.  Coreference Resolution : A Survey , 2006 .

[10]  Vincent Ng,et al.  Syntactic Parsing for Ranking-Based Coreference Resolution , 2011, IJCNLP.

[11]  Ruslan Mitkov,et al.  Robust Pronoun Resolution with Limited Knowledge , 1998, ACL.

[12]  Richárd Farkas,et al.  Data-driven Multilingual Coreference Resolution using Resolver Stacking , 2012, EMNLP-CoNLL Shared Task.

[13]  W. Vonk,et al.  Eye Movements During Comprehension of Pronouns , 1984 .

[14]  M. G. Calvo,et al.  Working memory and inferences: Evidence from eye fixations during reading , 2001, Memory.

[15]  Sarah Brown-Schmidt,et al.  The rapid use of gender information: evidence of the time course of pronoun resolution from eyetracking , 2000, Cognition.

[16]  Heeyoung Lee,et al.  A Multi-Pass Sieve for Coreference Resolution , 2010, EMNLP.

[17]  Pushpak Bhattacharyya,et al.  Measuring Sentiment Annotation Complexity of Text , 2014, ACL.

[18]  Christopher D. Manning,et al.  Entity-Centric Coreference Resolution with Model Stacking , 2015, ACL.

[19]  Shravan Vasishth,et al.  What is the scanpath signature of syntactic reanalysis , 2011 .

[20]  Ian Cunnings,et al.  Variable Binding and Coreference in Sentence Comprehension: Evidence from Eye-Movements. , 2014 .

[21]  Yuchen Zhang,et al.  CoNLL-2012 Shared Task: Modeling Multilingual Unrestricted Coreference in OntoNotes , 2012, EMNLP-CoNLL Shared Task.

[22]  K. Rayner,et al.  Lexical complexity and fixation times in reading: Effects of word frequency, verb complexity, and lexical ambiguity , 1986, Memory & cognition.

[23]  Jonas Kuhn,et al.  Learning Structured Perceptrons for Coreference Resolution with Latent Antecedents and Non-local Features , 2014, ACL.

[24]  Michael Strube,et al.  Plug Latent Structures and Play Coreference Resolution , 2015, ACL.