Learning Structured Perceptrons for Coreference Resolution with Latent Antecedents and Non-local Features

By Anders Björkelund, Jonas Kuhn. 2014

132 citations recorded by the source corpus.

We investigate different ways of learning structured perceptron models for coreference resolution when using non-local features and beam search. Our experimental results indicate that standard techniques such as early updates or Learning as Search Optimization (LaSO) perform worse than a greedy baseline that only uses local features. By modifying LaSO to delay updates until the end of each instance we obtain significant improvements over the baseline. Our model obtains the best results to date on recent shared task data for Arabic, Chinese, and English.

Subfields: Artificial Intelligence, Natural Language Processing

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Natural Language Processing

Citations in the corpus

Gender Bias in Coreference Resolution (2018, methodology): “Statistical Statistical methods, often with millions of parameters, ultimately surpassed the performance of rule-based systems on shared task data (Durrett and Klein, 2013; Björkelund and Kuhn, 2014). The system of Durrett and Klein (2013) replaced hand-written rules with simple feature templates.”
End-to-end Neural Coreference Resolution (2017, neutral): “…entitylevel models (Haghighi and Klein, 2010; Clark and Manning, 2015, 2016b; Wiseman et al., 2016), (3) latent-tree models (Fernandes et al., 2012; Björkelund and Kuhn, 2014; Martschat and Strube, 2015), or (4) mention-ranking models (Durrett and Klein, 2013; Wiseman et al., 2015; Clark and…”
Deep Reinforcement Learning for Mention-Ranking Coreference Models (2016, background): “…are typically trained with heuristic loss functions that assign costs to different error types, as in the heuristic loss we describe in Section 3.1 (Fernandes et al., 2012; Durrett et al., 2013; Bj¨orkelund and Kuhn, 2014; Wiseman et al., 2015; Martschat and Strube, 2015; Wiseman et al., 2016).”
Sequence-to-Sequence Learning as Beam-Search Optimization (2016, methodology): “We also note that because we do not update the parameters until after the T ’th search step, our training procedure differs slightly from LaSO (which is online), and in this aspect is essentially equivalent to the “delayed LaSO update” of Björkelund and Kuhn (2014). Algorithm 1 Seq2seq Beam-Search O”
Higher-Order Coreference Resolution with Coarse-to-Fine Inference (2018, background): “Since global features are notoriously difﬁcult to deﬁne (Wiseman et al., 2016), they often depend heavily on existing pairwise features or architectures (Bj¨orkelund and Kuhn, 2014; Clark and Manning, 2015, 2016b).”
Dynamic Entity Representations in Neural Language Models (2017, background): “In previous work, such information has been added as features (Luo et al., 2004; Bj¨orkelund and Kuhn, 2014) or by computing distributed entity representations (Wiseman et al., 2016; Clark and Manning, 2016b).”
Learning Global Features for Coreference Resolution (2016, methodology): “This approach contrasts in particular with the work of Bj¨orkelund and Kuhn (2014) — who also incorporate global information in mention-ranking — in that they train against latent trees , which are not annotated and must be searched for during training.”
Improving Coreference Resolution by Learning Entity-Level Distributed Representations (2016, methodology): “Other works use structured perceptron models for the same purpose (Stoyanov and Eisner, 2012; Fer-nandes et al., 2012; Bj¨orkelund and Kuhn, 2014).”
Equation Parsing : Mapping Sentences to Grounded Equations (2016, background): “A similar formulation can be found in Bj¨orkelund and Kuhn (2014).”
A Joint Model for Entity Analysis: Coreference, Typing, and Linking (2014, methodology): “On OntoNotes, we match the performance of the best published coreference system (Björkelund and Kuhn, 2014) and outperform two strong NER systems (Ratinov and Roth, 2009; Passos et al.”
Entity-Centric Coreference Resolution with Model Stacking (2015, methodology): “, 2014); the HOTCoref system (Björkelund and Kuhn, 2014); the CPL3M sytem (Chang et al.”
Unsupervised Visual-Linguistic Reference Resolution in Instructional Videos (2017, background): “ing [6, 30], and joint vision and language domains [27, 32, 45, 47].”