SciCo: Hierarchical Cross-Document Coreference for Scientific Concepts (2021-04-18T00:00:00.000000Z)

TL;DR

SciCo, an expert-annotated dataset for H-CDCR in scientific papers, is created, 3X larger than the prominent ECB+ resource, and is studied to study strong baseline models that are customize for H -CDCR, and highlight challenges for future work.

Abstract

Determining coreference of concept mentions across multiple documents is a fundamental task in natural language understanding. Previous work on cross-document coreference resolution (CDCR) typically considers mentions of events in the news, which seldom involve abstract technical concepts that are prevalent in science and technology. These complex concepts take diverse or ambiguous forms and have many hierarchical levels of granularity (e.g., tasks and subtasks), posing challenges for CDCR. We present a new task of Hierarchical CDCR (H-CDCR) with the goal of jointly inferring coreference clusters and hierarchy between them. We create SciCo, an expert-annotated dataset for H-CDCR in scientific papers, 3X larger than the prominent ECB+ resource. We study strong baseline models that we customize for H-CDCR, and highlight challenges for future work.

Authors

Daniel S. Weld

13 papers

Ido Dagan

19 papers

Iz Beltagy

11 papers

TL;DR

Abstract

Authors

References67 items

Realistic Evaluation Principles for Cross-document Coreference Resolution

Cross-document Coreference Resolution over Predicted Mentions

WEC: Deriving a Large-scale Cross-document Event Coreference dataset from Wikipedia

Extending Multi-Sense Word Embedding to Phrases and Sentences for Unsupervised Semantic Applications

CSFCube - A Test Collection of Computer Science Research Articles for Faceted Query by Example

CDˆ2CR: Co-reference resolution across documents and domains

Cross-Document Language Modeling

Clustering-based Inference for Biomedical Entity Linking

Extracting a Knowledge Base of Mechanisms from COVID-19 Papers

CoRefi: A Crowd Sourcing Suite for Coreference Annotation

Streamlining Cross-Document Coreference Resolution: Evaluation and Modeling

SciSight: Combining faceted navigation and research group detection for COVID-19 exploratory scientific search

SciREX: A Challenge Dataset for Document-Level Information Extraction

Don’t Stop Pretraining: Adapt Language Models to Domains and Tasks

NetTaxo: Automated Topic Taxonomy Construction from Text-Rich Network

Longformer: The Long-Document Transformer

PyTorch: An Imperative Style, High-Performance Deep Learning Library

Scalable Zero-shot Entity Linking with Dense Entity Retrieval

Do Multi-hop Readers Dream of Reasoning Chains?

Entity, Relation, and Event Extraction with Contextualized Span Representations

Sentence-BERT: Sentence Embeddings using Siamese BERT-Networks

RoBERTa: A Robustly Optimized BERT Pretraining Approach

Coreference Resolution with Entity Equalization

Revisiting Joint Modeling of Cross-document Entity and Event Coreference Resolution

SciBERT: A Pretrained Language Model for Scientific Text

Combining Distant and Direct Supervision for Neural Relation Extraction

Smoothing the Geometry of Probabilistic Box Embeddings

Multi-Task Identification of Entities, Relations, and Coreference for Scientific Knowledge Graph Construction

TaxoGen: Unsupervised Topic Taxonomy Construction by Adaptive Term Embedding and Clustering

Hearst Patterns Revisited: Automatic Hypernym Detection from Large Text Corpora

Resolving Event Coreference with Supervised Representation Learning and Clustering-Oriented Regularization

Breaking NLI Systems with Sentences that Require Simple Lexical Inferences

Don’t Annotate, but Validate: a Data-to-Text Method for Capturing Event Data

Neural Models for Reasoning over Multiple Mentions Using Coreference

Concept-Map-Based Multi-Document Summarization using Concept Coreference Resolution and Global Importance Optimization

SemEval-2017 Task 1: Semantic Textual Similarity Multilingual and Crosslingual Focused Evaluation

End-to-end Neural Coreference Resolution

Event Coreference Resolution by Iteratively Unfolding Inter-dependencies among Events

Accelerating Innovation Through Analogy Mining

Multimodal Word Distributions

Billion-Scale Similarity Search with GPUs

Adding Context to Semantic Data-Driven Paraphrasing

Which Coreference Evaluation Metric Do You Trust? A Proposal for a Link-based Entity Aware Metric

Improving Hypernymy Detection with an Integrated Path-based and Distributional Method

Exploring Multiple Feature Spaces for Novel Entity Discovery

Linear Algebraic Structure of Word Senses, with Applications to Polysemy

A large annotated corpus for learning natural language inference

Using a sledgehammer to crack a nut? Lexical diversity and event coreference resolution

Joint Coreference Resolution and Named-Entity Linking with Multi-Pass Sieves

CoNLL-2012 Shared Task: Modeling Multilingual Unrestricted Coreference in OntoNotes

No Noun Phrase Left Behind: Detecting and Typing Unlinkable Entities

Unsupervised Ontology Induction from Text

OntoNotes: A Unified Relational Semantic Representation

On Coreference Resolution Performance Metrics

Entity-Based Cross-Document Coreferencing Using the Vector Space Model

A Model-Theoretic Coreference Scoring Scheme

Information Retrieval Based on Conceptual Distance in is-a Hierarchies

Automatic Acquisition of Hyponyms from Large Text Corpora

Hierarchical Grouping to Optimize an Objective Function

1963] over a pairwise distance matrix populated with scores s(i, j), using the averagelinkage cluster merging criterion

2021a]’s CDCR model Here, we describe the cross-document resolver of [Cattan et al., 2021a] that we use as baseline for SCICO (§4.1)

S2ORC: The Semantic Scholar Open Research Corpus

Pytorch lightning

Wikilinks: A Large-scale Cross-Document Coreference Corpus Labeled via Links to Wikipedia

Proceedings of the 2012 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning

The Unified Medical Language System (UMLS): integrating biomedical terminology

of the Association for Computational Linguistics

Field of Study

Journal Information

Name

Volume

Venue Information

Name

Type

URL

Alternate Names