A Joint Model for Entity Analysis: Coreference, Typing, and Linking

By Greg Durrett, Daniel Klein. 2014

271 citations recorded by the source corpus.

We present a joint model of three core tasks in the entity analysis stack: coreference resolution (within-document clustering), named entity recognition (coarse semantic typing), and entity linking (matching to Wikipedia entities). Our model is formally a structured conditional random field. Unary factors encode local features from strong baselines for each task. We then add binary and ternary factors to capture cross-task interactions, such as the constraint that coreferent mentions have the same semantic type. On the ACE 2005 and OntoNotes datasets, we achieve state-of-the-art results for all three tasks. Moreover, joint modeling improves performance on each task over strong independent baselines.

Subfields: Artificial Intelligence, Management Science and Operations Research, 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 Management Science and Operations Research

In Natural Language Processing

Citations in the corpus

Learning Dense Representations for Entity Retrieval (2019, background): “Recent work such as Durrett and Klein (2014) has shown improvements derived from better, longer-range, context features; similarly, there are many more potentially useful KB-derived features.”
End-to-end Neural Coreference Resolution (2017, neutral): “2 69.3 72.6 66.2 55.8 60.5 59.4 54.9 57.1 63.4 Clark and Manning (2015) 76.1 69.4 72.6 65.6 56.0 60.4 59.4 53.0 56.0 63.0 Martschat and Strube (2015) 76.7 68.1 72.2 66.1 54.2 59.6 59.5 52.3 55.7 62.5 Durrett and Klein (2014) 72.6 69.9 71.2 61.2 56.4 58.7 56.2 54.2 55.2 61.7 Bjo¨rkelund and Kuhn (201”
Posterior calibration and exploratory analysis for natural language processing models (2015, background): “…the goal is to marginalize over structures to calculate and minimize an expected loss function, as in minimum Bayes risk decoding (Goodman, 1996; Kumar and Byrne, 2004), or to perform joint inference between early and later stages of NLP analysis (e.g. Singh et al., 2013; Durrett and Klein, 2014).”
Named Entity Recognition with Bidirectional LSTM-CNNs (2015, methodology): “Finally, the performance characteristics of our model appears to be quite different than the previous CRF models (Finkel and Manning, 2009; Durrett and Klein, 2014), likely because we apply a completely different machine learning method.”
Investigating Entity Knowledge in BERT with Simple Neural End-To-End Entity Linking (2019, background): “Durrett and Klein (2014) were the ﬁrst to propose jointly modelling MD, CG and ED in a graphical model and could show that each of those steps are interdependent and beneﬁt from a joint objective.”
Ultra-Fine Entity Typing (2018, background): “, 2016), query analysis (Balog and Neumayer, 2012), and coreference resolution (Durrett and Klein, 2014).”
A Hierarchical Multi-task Approach for Learning Embeddings from Semantic Tasks (2018, methodology): “For coreference, figures that are comparable with (Durrett and Klein 2014) are tagged with an *.”
Capturing Semantic Similarity for Entity Linking with Convolutional Neural Networks (2016, methodology): “• ACE (NIST, 2005; Bentivogli et al., 2010): This corpus was used in Fahrni and Strube (2014) and Durrett and Klein (2014).”
Improving Entity Linking by Modeling Latent Relations between Mentions (2018, background): “One notable exception is Durrett and Klein (2014): they use a joint model of entity linking and coreference resolution.”
Joint Learning of Named Entity Recognition and Entity Linking (2019, methodology): “(Durrett and Klein, 2014) performed joint learning of entity typing, EL, and coreference using a structured CRF, also with hand-engineered features.”
Learning Anaphoricity and Antecedent Ranking Features for Coreference Resolution (2015, background): “, 2013), as opposed to those that focus on non-local, structured prediction (McCallum and Wellner, 2003; Culotta et al., 2006; Haghighi and Klein, 2010; Fernandes et al., 2012; Stoyanov and Eisner, 2012; Björkelund and Farkas, 2012; Wick et al., 2012; Björkelund and Kuhn, 2014; Durrett and Klein, 20”
Entity Linking via Joint Encoding of Types, Descriptions, and Context (2017, methodology): “Few recent entity linking approaches (Hoffart et al., 2011; Durrett and Klein, 2014; Nguyen et al., 2016; Francis-Landau et al., 2016) use manuallyannotated domain specific training data to learn the linking system.”
Learning Global Features for Coreference Resolution (2016, methodology): “The application of RNNs to modeling (the trajectory of) the state of a cluster is apparently novel, though it bears some similarity to the recent work of Dyer et al. (2015), who use LSTMs to embed the state of a transition based parser’s stack.”
TaggerOne: joint named entity recognition and normalization with semi-Markov Models (2016, background): “For example, Finkel and Manning (2009) learned a joint model for parsing and NER in newswire text, while Durrett and Klein (2014) learned a model for joint coreference resolution, named entity classification and entity linking (disambiguation) when the named entity spans were provided as input.”
Fine-Grained Entity Typing for Domain Independent Entity Linking (2019, methodology): “Recent approaches have used neural techniques (He et al., 2013; Sun et al., 2015; Francis-Landau et al., 2016), various joint models (Durrett and Klein, 2014; Nguyen et al., 2016).”
SpanNER: Named Entity Re-/Recognition as Span Prediction (2021, methodology): “Following previous works (Durrett and Klein, 2014; Ghaddar and Langlais, 2018), we utilize different domains in English to test the robustness of proposed models.”