Improving Machine Learning Approaches to Coreference Resolution

By Vincent Ng, Claire Gardent. 2002

793 citations recorded by the source corpus.

We present a noun phrase coreference system that extends the work of Soon et al. (2001) and, to our knowledge, produces the best results to date on the MUC-6 and MUC-7 coreference resolution data sets --- F-measures of 70.4 and 63.4, respectively. Improvements arise from two sources: extra-linguistic changes to the learning framework and a large-scale expansion of the feature set to include more sophisticated linguistic knowledge.

Subfields: Artificial Intelligence, Molecular Biology, 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 Molecular Biology

In Natural Language Processing

Citations in the corpus

A Hybrid Neural Network Model for Commonsense Reasoning (2019, background): “WSC and PDP differ from other co-reference resolution tasks (Soon et al., 2001; Ng and Cardie, 2002; Peng et al., 2016) in that commonsense knowledge, which cannot be explicitly decoded from the given text, is needed to solve the problem, as illustrated in the examples in Table 1.”
Learning Structured Embeddings of Knowledge Bases (2011, methodology): “in (Ng and Cardie 2002; Snow et al. 2007)), this remains a small contribution compared to what could be achieved with such gigantic knowledge quantities.”
Modeling Local Coherence: An Entity-Based Approach (2005, methodology): “Table 2: Summary augmented with syntactic annotations for grid computation. we employ a state-of-the-art noun phrase coreference resolution system (Ng and Cardie, 2002) trained on the MUC (6–7) data sets.”
Solving Hard Coreference Problems (2019, methodology): “We then incorporate all constraints into a general coreference system (Chang et al., 2013) utilizing the mention-pair model (Ng and Cardie, 2002b; Bengt-son and Roth, 2008; Stoyanov et al., 2010).”
Bootstrapping Path-Based Pronoun Resolution (2006, background): “Since no such corpus exists, researchers have used coarser features learned from smaller sets through supervised learning (Soon et al., 2001; Ng and Cardie, 2002), manually-de(cid:2)ned coreference patterns to mine speci(cid:2)c kinds of data (Bean and Riloff, 2004; Bergsma, 2005), or accepted the…”
Entity-Centric Coreference Resolution with Model Stacking (2015, methodology): “A widely used approach for this is bestfirst clustering (Ng and Cardie, 2002).”
Unsupervised Deep Structured Semantic Models for Commonsense Reasoning (2019, background): “Although both tasks are based on pronoun disambiguation, a subtask of coreference resolution (Soon et al., 2001; Ng and Cardie, 2002; Peng et al., 2016), PDP and WSC differ from normal pronoun disambiguation due to their unique properties, which are based on commonsense, selecting the most likely an”
Learning Structured Perceptrons for Coreference Resolution with Latent Antecedents and Non-local Features (2014, background): “The use of latent antecedents goes back to the work of Yu and Joachims (2009), although the idea of determining meaningful antecedents for mentions can be traced back to Ng and Cardie (2002) who used a rule-based approach.”
Active Learning for Coreference Resolution (2012, methodology): “The feature set of this system is similar to that used by Ng and Cardie (2002).”
Active Learning for Coreference Resolution (2012, methodology): “The link classiﬁer is a decision tree and the clustering algorithm a variant of best-ﬁrst clustering (Ng and Cardie, 2002).”
Event Detection and Co-reference with Minimal Supervision (2016, methodology): “Similar to the mention-pair model in entity co-reference (Ng and Cardie, 2002; Bengtson and Roth, 2008; Stoyanov et al., 2010), we use cosine similarities computed from pairs of event mentions: S ( e 1 , e 2 ) (as in Eq.”
Adapting Coreference Resolution for Processing Violent Death Narratives (2021, background): “Coreference resolution (Soon et al., 2001; Ng and Cardie, 2002) is the task of identifying denotative phrases in text that refer to the same entity.”
Simple Coreference Resolution with Rich Syntactic and Semantic Features (2009, background): “While much research (Ng and Cardie, 2002; Cu-lotta et al., 2007; Haghighi and Klein, 2007; Poon and Domingos, 2008; Finkel and Manning, 2008) has explored how to reconcile pairwise decisions to form coherent clusters, we simply take the transitive closure of our pairwise decision (as in Ng and…”
Coreference Semantics from Web Features (2012, background): “It includes over 80 core features that exploit various automatically generated annotations such as named entity tags, syntactic parses, and WordNet classes, inspired by Soon et al. (2001), Ng and Cardie (2002), and Bengtson and Roth (2008).”
Self-Paced Learning for Latent Variable Models (2010, background): “Formally, the input vector x consists of the pairwise features xij suggested in [16] between all pairs of noun phrases i and j in the document.”
Coreference Resolution with World Knowledge (2011, background): “Despite the usefulness of world knowledge for coreference resolution, early learning-based coreference resolvers have relied mostly on morpho-syntactic features (e.g., Soon et al. (2001), Ng and Cardie (2002), Yang et al. (2003)).”
Understanding the Value of Features for Coreference Resolution (2008, methodology): “This is similar to the technique of Ng and Cardie (2002b).”
Conundrums in Noun Phrase Coreference Resolution: Making Sense of the State-of-the-Art (2009, methodology): “Finally, several coreference systems have successfully incorporated anaphoricity determination modules (e.g. Ng and Cardie (2002a) and Bean and Riloff (2004)).”