Solving Hard Coreference Problems

By Haoruo Peng, Daniel Khashabi, Dan Roth. 2019

90 citations recorded by the source corpus.

Coreference resolution is a key problem in natural language understanding that still escapes reliable solutions. One fundamental difficulty has been that of resolving instances involving pronouns since they often require deep language understanding and use of background knowledge. In this paper we propose an algorithmic solution that involves a new representation for the knowledge required to address hard coreference problems, along with a constrained optimization framework that uses this knowledge in coreference decision making. Our representation, Predicate Schemas, is instantiated with knowledge acquired in an unsupervised way, and is compiled automatically into constraints that impact the coreference decision. We present a general coreference resolution system that significantly improves state-of-the-art performance on hard, Winograd-style, pronoun resolution cases, while still performing at the stateof-the-art level on standard coreference resolution datasets.

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

Experience Grounds Language (2020, background): “ferences, such as hard coreference problems (Peng et al., 2015).”
WinoGrande (2019, methodology): “This IR-oriented approach comes from a line of work in coreference resolution (Kobdani et al. 2011; Ratinov and Roth 2012; Bansal and Klein 2012; Zheng et al. 2013; Peng, Khashabi, and Roth 2015).”
A Simple Method for Commonsense Reasoning (2018, methodology): “ally commonsense information. Previous attempts on solving the Winograd Schema Challenge usually involve heavy utilization of annotated knowledge bases, rule-based reasoning, or hand-crafted features [22, 23, 24]. In particular, Rahman and Ng [25] employ human annotators to build more supervised tra”
Gender Bias in Coreference Resolution: Evaluation and Debiasing Methods (2018, methodology): “This dataset follows the winograd format (Hirst, 1981; Rah-man and Ng, 2012; Peng et al., 2015b), and contains references to people using a vocabulary of 40 occupations.”
Gender Bias in Contextualized Word Embeddings (2019, methodology): “We evaluate bias with respect to the WinoBias dataset (Zhao et al., 2018a), a benchmark of paired male and female coreference resolution examples following the Winograd format (Hirst, 1981; Rah-man and Ng, 2012; Peng et al., 2015).”
Natural Perturbation for Robust Question Answering (2020, background): “Minimal-pairs in NLP. Datasets with minimalpair instances are relatively well-established in certain tasks; for instance, Winograd schema datasets (Levesque et al., 2011; Peng et al., 2015; Sakaguchi et al., 2020), or the recently-published contrast sets (Gardner et al., 2020).”
More Bang for Your Buck: Natural Perturbation for Robust Question Answering (2020, background): “Minimal-pairs in NLP. Datasets with minimal-pair instances are relatively well-established in certain tasks, such as Winograd schema datasets (Levesque et al., 2011; Peng et al., 2015; Sakaguchi et al., 2020), or the recent contrast sets (Gardner et al., 2020).”
Contrastive Self-Supervised Learning for Commonsense Reasoning (2020, unknown): “…et al., 2019) 63.0 % Ensemble LMs (Trinh and Le, 2018) 63.8 % CSS (Proposed Method) 69.6 % DPR (Rahman and Ng, 2012) (Rahman and Ng, 2012) 73.0% (Peng et al., 2015) 76.4 % CSS (Proposed Method) 80.1 % KnowRef (Emami et al., 2019) E2E (Emami et al., 2019) 58.0 % BERT-ft (Emami et al., 2019) 61.0…”
On GAP Coreference Resolution Shared Task: Insights from the 3rd Place Solution (2019, methodology): “Besides GAP data, the presented solution uses four external data sources: Winobias (Zhao et al., 2018), Wino-gender (Rudinger et al., 2018), The Deﬁnite Pro-noun Resolution (DPR) Dataset (Rahman and Ng, 2012; Peng et al., 2015) and Ontonotes 5.0 (Prad-han et al., 2012).”
The KnowRef Coreference Corpus: Removing Gender and Number Cues for Difficult Pronominal Anaphora Resolution (2018, background): “Although the WSC is an important step in evaluating systems en route to human-like language understanding, its size and other characteristics are a bottleneck for progress in pronoun disambiguation (Trichelair et al., 2018).”
An Adversarial Winograd Schema Challenge at Scale (2019, background): “This IR-oriented approach comes from an important line of work in coreference resolution (Kobdani et al., 2011; Ratinov and Roth, 2012; Bansal and Klein, 2012; Zheng et al., 2013; Peng et al., 2015; Sharma et al., 2015).”
WikiCREM: A Large Unsupervised Corpus for Coreference Resolution (2019, methodology): “We decide not to use the CONLL2012 and WINOCOREF (Pradhan et al., 2012; Peng et al., 2015) datasets, because they contain more general coreference examples than just pronouns.”
A Generalized Knowledge Hunting Framework for the Winograd Schema Challenge (2018, background): “Systems directed exclusively towards the expanded Winograd corpus include Rahman and Ng’s system itself (Rahman and Ng, 2012), reporting 73% accuracy on Winograd-like sentences, and Peng et al.’s system that improves accuracy to 76% (Peng et al., 2015).”
Addressing the Winograd Schema Challenge as a Sequence Ranking Task (2018, methodology): “We also downloaded the state of the art system of Peng et al. (2015a), which the authors made publicly available 8 .”
A Knowledge Hunting Framework for Common Sense Reasoning (2018, methodology): “Other systems (Kobdani et al., 2011; Ratinov and Roth, 2012; Bansal and Klein, 2012; Zheng et al., 2013; Peng et al., 2015; Sharma et al., 2015) rely on similar techniques, i.”
Unsupervised Deep Structured Semantic Models for Commonsense Reasoning (2019, unknown): “Previous efforts on solving the Winograd Schema Challenge and Pronoun Disambiguation Problems mostly rely on human-labeled data, sophisticated rules, hand-crafted features, or external knowledge bases (Peng et al., 2015; Bailey et al., 2015; Schüller, 2014).”