Rewarding Coreference Resolvers for Being Consistent with World Knowledge

Aralikatte, Rahul; Lent, Heather; Gonzalez, Ana Valeria; Herschcovich, Daniel; Chen, Qiu; Sandholm, Anders; Ringaard, Michael; Søgaard, Anders

doi:10.18653/v1/d19-1118

Cited by 11 publications

(15 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Precision vs. recall Precision often increases after finetuning whereas recall decreases. Similar effects have been reported for knowledge-base grounding of coreference resolvers (Aralikatte et al 2019).…”

Section: Error Analysissupporting

confidence: 82%

“…Previous work has augmented Coreference resolvers with syntax information (Wiseman, Rush, and Shieber 2016;Clark and Manning 2016a,b), external world knowledge (Rahman and Ng 2011;Emami et al 2018;Aralikatte et al 2019) and a variety of other linguistic features (Ng 2007;Haghighi and Klein 2009;Zhang, Song, and Song 2019). Similarly, Ponzetto and Strube (2006a,b) used features from SRL and external sources for a non-neural coreference resolver.…”

Section: Related Work Augmented Coreference Resolutionmentioning

confidence: 99%

See 1 more Smart Citation

Joint Semantic Analysis with Document-Level Cross-Task Coherence Rewards

Aralikatte

Abdou

Lent

et al. 2021

AAAI

Self Cite

View full text Add to dashboard Cite

Coreference resolution and semantic role labeling are NLP tasks that capture different aspects of semantics, indicating respectively, which expressions refer to the same entity, and what semantic roles expressions serve in the sentence. However, they are often closely interdependent, and both generally necessitate natural language understanding. Do they form a coherent abstract representation of documents? We present a neural network architecture for joint coreference resolution and semantic role labeling for English, and train graph neural networks to model the 'coherence' of the combined shallow semantic graph. Using the resulting coherence score as a reward for our joint semantic analyzer, we use reinforcement learning to encourage global coherence over the document and between semantic annotations. This leads to improvements on both tasks in multiple datasets from different domains, and across a range of encoders of different expressivity, calling, we believe, for a more holistic approach for semantics in NLP.

show abstract

Section: Error Analysissupporting

confidence: 82%

Section: Related Work Augmented Coreference Resolutionmentioning

confidence: 99%

Joint Semantic Analysis with Document-Level Cross-Task Coherence Rewards

Aralikatte

Abdou

Lent

et al. 2021

AAAI

Self Cite

View full text Add to dashboard Cite

show abstract

“…One example of such data would be elevation maps [49], which can be integrated into MapLUR in a way similar to Experiment 4, where we provided the CNN with additional map image layers. Beyond this, there is a wide variety of methods to provide deep learning methods with additional information which holds great potential to further improve our results [4,39].…”

Section: Discussionmentioning

confidence: 99%

MapLUR: Exploring a new Paradigm for Estimating Air Pollution using Deep Learning on Map Images

Steininger,

Kobs,

Zehe

et al. 2020

Preprint

View full text Add to dashboard Cite

“…Conisdering CR approaches that use external knowledge, Aralikatte et al (2019) solved CR task by incorporate knowledge or information in reinforcement learning models. Emami et al (2018) solved the binary choice coreference-resolution task by leveraging information retrieval results from search engines.…”

Section: Related Workmentioning

confidence: 99%

A Knowledge-Driven Approach to Classifying Object and Attribute Coreferences in Opinion Mining

Chen¹,

Wang

Mazumder

et al. 2020

Findings of the Association for Computational Linguistics: EMNLP 2020

View full text Add to dashboard Cite

Classifying and resolving coreferences of objects (e.g., product names) and attributes (e.g., product aspects) in opinionated reviews is crucial for improving the opinion mining performance. However, the task is challenging as one often needs to consider domain-specific knowledge (e.g., iPad is a tablet and has aspect resolution) to identify coreferences in opinionated reviews. Also, compiling a handcrafted and curated domain-specific knowledge base for each domain is very time consuming and arduous. This paper proposes an approach to automatically mine and leverage domain-specific knowledge for classifying objects and attribute coreferences. The approach extracts domain-specific knowledge from unlabeled review data and trains a knowledgeaware neural coreference classification model to leverage (useful) domain knowledge together with general commonsense knowledge for the task. Experimental evaluation on realworld datasets involving five domains (product types) shows the effectiveness of the approach.

show abstract

Rewarding Coreference Resolvers for Being Consistent with World Knowledge

Cited by 11 publications

References 22 publications

Joint Semantic Analysis with Document-Level Cross-Task Coherence Rewards

Joint Semantic Analysis with Document-Level Cross-Task Coherence Rewards

MapLUR: Exploring a new Paradigm for Estimating Air Pollution using Deep Learning on Map Images

A Knowledge-Driven Approach to Classifying Object and Attribute Coreferences in Opinion Mining

Contact Info

Product

Resources

About