Multigraph Clustering for Unsupervised Coreference Resolution
Article Structure
Abstract
We present an unsupervised model for coreference resolution that casts the problem as a clustering task in a directed labeled weighted multigraph.
Introduction
Coreference resolution is the task of determining which mentions in a text refer to the same entity.
Related Work
Graph-based coreference resolution.
A Multigraph Model
We aim for a model which directly represents the relations between mentions in a graph structure.
Relations
The graph model described in Section 3 is based on expressing relations between pairs of mentions via edges built from such relations.
Evaluation
5.1 Data and Evaluation Metrics
Error Analysis
In order to understand weaknesses of our model we perform an error analysis on the development data.
Conclusions and Future Work
We presented an unsupervised graph-based model for coreference resolution.
Topics
coreference
Appears in 30 sentences as: Coreference (3) coreference (26) coreferent (4) coreferent, (1)
In Multigraph Clustering for Unsupervised Coreference Resolution
- We present an unsupervised model for coreference resolution that casts the problem as a clustering task in a directed labeled weighted multigraph.Page 1, “Abstract”
- Coreference resolution is the task of determining which mentions in a text refer to the same entity.Page 1, “Introduction”
- Quite recently, however, rule-based approaches regained popularity due to Stanford’s multi-pass sieve approach which exhibits state-of-the-art performance on many standard coreference data sets (Raghunathan et al., 2010) and also won the CoNLL-2011 shared task on coreference resolution (Lee et al., 2011; Pradhan et al., 2011).Page 1, “Introduction”
- In this paper we present a graph-based approach for coreference resolution that models a document to be processed as a graph.Page 1, “Introduction”
- Coreference resolution is performed via graph clustering.Page 1, “Introduction”
- Our approach belongs to a class of recently proposed graph models for coreference resolution (Cai and Strube, 2010;Page 1, “Introduction”
- Graph-based coreference resolution.Page 1, “Related Work”
- Nicolae and Nicolae (2006) phrase coreference resolution as a graph clustering problem: they first perform pairwise classification and then construct a graph using the derived confidence values as edge weights.Page 1, “Related Work”
- (2010) and Cai and Strube (2010) perform coreference resolution in one step using graph partitioning approaches.Page 1, “Related Work”
- Unsupervised coreference resolution.Page 2, “Related Work”
- Cardie and Wagstaff (1999) present an early approach to unsupervised coreference resolution based on a straightforward clustering approach.Page 2, “Related Work”
See all papers in Proc. ACL 2013 that mention coreference.
See all papers in Proc. ACL that mention coreference.
Back to top.
coreference resolution
Appears in 19 sentences as: Coreference Resolution (1) Coreference resolution (2) coreference resolution (16)
In Multigraph Clustering for Unsupervised Coreference Resolution
- We present an unsupervised model for coreference resolution that casts the problem as a clustering task in a directed labeled weighted multigraph.Page 1, “Abstract”
- Coreference resolution is the task of determining which mentions in a text refer to the same entity.Page 1, “Introduction”
- Quite recently, however, rule-based approaches regained popularity due to Stanford’s multi-pass sieve approach which exhibits state-of-the-art performance on many standard coreference data sets (Raghunathan et al., 2010) and also won the CoNLL-2011 shared task on coreference resolution (Lee et al., 2011; Pradhan et al., 2011).Page 1, “Introduction”
- In this paper we present a graph-based approach for coreference resolution that models a document to be processed as a graph.Page 1, “Introduction”
- Coreference resolution is performed via graph clustering.Page 1, “Introduction”
- Our approach belongs to a class of recently proposed graph models for coreference resolution (Cai and Strube, 2010;Page 1, “Introduction”
- Graph-based coreference resolution .Page 1, “Related Work”
- Nicolae and Nicolae (2006) phrase coreference resolution as a graph clustering problem: they first perform pairwise classification and then construct a graph using the derived confidence values as edge weights.Page 1, “Related Work”
- (2010) and Cai and Strube (2010) perform coreference resolution in one step using graph partitioning approaches.Page 1, “Related Work”
- Unsupervised coreference resolution .Page 2, “Related Work”
- Cardie and Wagstaff (1999) present an early approach to unsupervised coreference resolution based on a straightforward clustering approach.Page 2, “Related Work”
See all papers in Proc. ACL 2013 that mention coreference resolution.
See all papers in Proc. ACL that mention coreference resolution.
Back to top.
shared task
Appears in 16 sentences as: shared task (14) shared tasks (2) shared task’s (1)
In Multigraph Clustering for Unsupervised Coreference Resolution
- The model outperforms most systems participating in the English track of the CoNLL’ 12 shared task .Page 1, “Abstract”
- Quite recently, however, rule-based approaches regained popularity due to Stanford’s multi-pass sieve approach which exhibits state-of-the-art performance on many standard coreference data sets (Raghunathan et al., 2010) and also won the CoNLL-2011 shared task on coreference resolution (Lee et al., 2011; Pradhan et al., 2011).Page 1, “Introduction”
- On the English data of the CoNLL’ 12 shared task the model outperforms most systems which participated in the shared task .Page 1, “Introduction”
- These approaches participated in the recent CoNLL’ll shared task (Pradhan et al., 2011; Sapena et al., 2011; Cai et al., 2011b) with excellent results.Page 1, “Related Work”
- (2012) and ranked second in the English track at the CoNLL’ 12 shared task (Pradhan et al., 2012).Page 1, “Related Work”
- The top performing system at the CoNLL’ 12 shared task (Femandes et al., 2012)Page 1, “Related Work”
- (2011), which in turn won the CoNLL’ 11 shared task .Page 2, “Related Work”
- We use the data provided for the English track of the CoNLL’ l2 shared task on multilingual coreference resolution (Pradhan et al., 2012) which is a subset of the upcoming OntoNotes 5.0 release and comes with various annotation layers provided by state-of-the-art NLP tools.Page 4, “Evaluation”
- We evaluate the model in a setting that corresponds to the shared task’s closed track, i.e.Page 4, “Evaluation”
- We evaluate our system with the coreference resolution evaluation metrics that were used for the CoNLL shared tasks on coreference, which are MUC (Vilain et al., 1995), B3 (Bagga and Baldwin, 1998) and CEAFe (Luo, 2005).Page 4, “Evaluation”
- We also report the unweighted average of the three scores, which was the official evaluation metric in the shared tasks .Page 4, “Evaluation”
See all papers in Proc. ACL 2013 that mention shared task.
See all papers in Proc. ACL that mention shared task.
Back to top.
CoNLL’
Appears in 12 sentences as: CoNLL (1) CoNLL’ (11)
In Multigraph Clustering for Unsupervised Coreference Resolution
- The model outperforms most systems participating in the English track of the CoNLL’ 12 shared task.Page 1, “Abstract”
- On the English data of the CoNLL’ 12 shared task the model outperforms most systems which participated in the shared task.Page 1, “Introduction”
- (2012) and ranked second in the English track at the CoNLL’ 12 shared task (Pradhan et al., 2012).Page 1, “Related Work”
- The top performing system at the CoNLL’ 12 shared task (Femandes et al., 2012)Page 1, “Related Work”
- (2011), which in turn won the CoNLL’ 11 shared task.Page 2, “Related Work”
- We use the data provided for the English track of the CoNLL’ l2 shared task on multilingual coreference resolution (Pradhan et al., 2012) which is a subset of the upcoming OntoNotes 5.0 release and comes with various annotation layers provided by state-of-the-art NLP tools.Page 4, “Evaluation”
- We evaluate our system with the coreference resolution evaluation metrics that were used for the CoNLL shared tasks on coreference, which are MUC (Vilain et al., 1995), B3 (Bagga and Baldwin, 1998) and CEAFe (Luo, 2005).Page 4, “Evaluation”
- CoNLL’ 12 shared task, which are denoted as best and median respectively.Page 5, “Evaluation”
- We use the official CoNLL’ 12 English training set for training.Page 5, “Evaluation”
- Our unsupervised model performs considerably better than the median system from the CoNLL’ 12 shared task on both data sets according to all metrics.Page 5, “Evaluation”
- While we observe a decrease of 1 point average score when evaluating on test data the model still would have ranked fourth in the English track of the CoNLL’ 12 shared task with only 0.2 points difference in average score to the second ranked system.Page 5, “Evaluation”
See all papers in Proc. ACL 2013 that mention CoNLL’.
See all papers in Proc. ACL that mention CoNLL’.
Back to top.
graph models
Appears in 5 sentences as: graph model (1) graph modeling (1) graph models (3)
In Multigraph Clustering for Unsupervised Coreference Resolution
- Our approach belongs to a class of recently proposed graph models for coreference resolution (Cai and Strube, 2010;Page 1, “Introduction”
- Figure 1: An example graph modeling relations between mentions.Page 2, “A Multigraph Model”
- Many graph models for coreference resolution operate on A = V x V. Our multigraph model allows us to have multiple edges with different labels between mentions.Page 2, “A Multigraph Model”
- In contrast to previous work on similar graph models we do not learn any edge weights from training data.Page 3, “A Multigraph Model”
- The graph model described in Section 3 is based on expressing relations between pairs of mentions via edges built from such relations.Page 3, “Relations”
See all papers in Proc. ACL 2013 that mention graph models.
See all papers in Proc. ACL that mention graph models.
Back to top.
graph-based
Appears in 5 sentences as: Graph-based (1) graph-based (4)
In Multigraph Clustering for Unsupervised Coreference Resolution
- In this paper we present a graph-based approach for coreference resolution that models a document to be processed as a graph.Page 1, “Introduction”
- Graph-based coreference resolution.Page 1, “Related Work”
- While not developed within a graph-based framework, factor-based approaches for pronoun resolution (Mitkov, 1998) can be regarded as greedy clustering in a multigraph, where edges representing factors for pronoun resolution have negative or positive weight.Page 1, “Related Work”
- Work on graph-based models similar to ours report robustness with regard to the amount of training data used (Cai et al., 2011b; Cai et al., 2011a; Martschat et al., 2012).Page 3, “A Multigraph Model”
- We presented an unsupervised graph-based model for coreference resolution.Page 6, “Conclusions and Future Work”
See all papers in Proc. ACL 2013 that mention graph-based.
See all papers in Proc. ACL that mention graph-based.
Back to top.
edge weights
Appears in 4 sentences as: edge weights (4)
In Multigraph Clustering for Unsupervised Coreference Resolution
- In contrast to previous models belonging to this class we do not learn any edge weights but perform inference on the graph structure only which renders our model unsupervised.Page 1, “Introduction”
- Nicolae and Nicolae (2006) phrase coreference resolution as a graph clustering problem: they first perform pairwise classification and then construct a graph using the derived confidence values as edge weights .Page 1, “Related Work”
- In contrast to previous work on similar graph models we do not learn any edge weights from training data.Page 3, “A Multigraph Model”
- We aim to employ a simple and efficient clustering scheme on this graph and therefore choose 1-nearest-neighbor clustering: for every m, we choose as antecedent m’s child n such that the sum of edge weights is maximal and positive.Page 3, “A Multigraph Model”
See all papers in Proc. ACL 2013 that mention edge weights.
See all papers in Proc. ACL that mention edge weights.
Back to top.
evaluation metrics
Appears in 4 sentences as: evaluation metric (1) Evaluation Metrics (1) evaluation metrics (2)
In Multigraph Clustering for Unsupervised Coreference Resolution
- 5.1 Data and Evaluation MetricsPage 4, “Evaluation”
- We evaluate our system with the coreference resolution evaluation metrics that were used for the CoNLL shared tasks on coreference, which are MUC (Vilain et al., 1995), B3 (Bagga and Baldwin, 1998) and CEAFe (Luo, 2005).Page 4, “Evaluation”
- We also report the unweighted average of the three scores, which was the official evaluation metric in the shared tasks.Page 4, “Evaluation”
- o the evaluation metrics employed are to be questioned (certainly),Page 6, “Conclusions and Future Work”
See all papers in Proc. ACL 2013 that mention evaluation metrics.
See all papers in Proc. ACL that mention evaluation metrics.
Back to top.
rule-based
Appears in 3 sentences as: rule-based (3)
In Multigraph Clustering for Unsupervised Coreference Resolution
- With the advent of machine learning and the availability of annotated corpora in the mid 1990s the research focus shifted from rule-based approaches to supervised machine learning techniques.Page 1, “Introduction”
- Quite recently, however, rule-based approaches regained popularity due to Stanford’s multi-pass sieve approach which exhibits state-of-the-art performance on many standard coreference data sets (Raghunathan et al., 2010) and also won the CoNLL-2011 shared task on coreference resolution (Lee et al., 2011; Pradhan et al., 2011).Page 1, “Introduction”
- These results show that carefully crafted rule-based systems which employ suitable inference schemes can achieve competitive performance.Page 1, “Introduction”
See all papers in Proc. ACL 2013 that mention rule-based.
See all papers in Proc. ACL that mention rule-based.
Back to top.
