Document Driven Exploration of Citation Usage via Contextualized Neural Language Models (DSI-SRP)
This DSI-SRP fellowship funded Joseph DeRose to work in the laboratory of Professor Matthew Berger in the Department of Electrical Engineering and Computer Science during the summer of 2019. Joseph graduated in 2020 with majors in Computer Science and Mathematics.
As document collections in conferences and journals have grown, documents for research have become more publicly available. However, this growth and accessibility has led to challenges in peer reviewing, as paper reviewers must keep up with the fast pace of research publications. Thus, understanding how a given article compounds upon prior research has become challenging. The research project funded by this fellowship aims to ease the burden of peer reviewing through the development of automated, and semi-automated, machine reading techniques to classify citation usage: how a cited document is used in a given research article. Documents are used for a variety of reasons, e.g. supporting evidence, evaluation methodology, comparison purposes. Therefore, automated techniques to detect such usage can help reduce the cognitive load of peer reviewing. Our proposed research will build off contextualized neural language models used in natural language processing (NLP), with an emphasis on aligning a document’s citation with its corresponding article.
In addition to receiving support through a DSI-SRP fellowship, this project was supported and facilitated by the DSI Data Science Team through their regular summer workshops and demo sessions.