Identifying Reference Spans: Topic Modeling and Word Embeddings help IR

TL;DR

This paper explores how topic models and word embeddings can improve the identification of reference spans in scientific documents, advancing the state-of-the-art in citation analysis tasks.

Contribution

It introduces a novel approach combining topic modeling and word embeddings that outperforms previous systems in reference span identification.

Findings

Topic models and word embeddings enhance performance

Achieved surpassing previous best system

Provides insights into citation span detection

Abstract

The CL-SciSumm 2016 shared task introduced an interesting problem: given a document D and a piece of text that cites D, how do we identify the text spans of D being referenced by the piece of text? The shared task provided the first annotated dataset for studying this problem. We present an analysis of our continued work in improving our system's performance on this task. We demonstrate how topic models and word embeddings can be used to surpass the previously best performing system.