Categorizing Citation Relations in Scientific Papers Based on the Contributions of Cited Papers

Po-Chun Chen, Hen-Hsen Huang, Hsin-Hsi Chen

Abstract: With the massive increase in the number of research papers, it becomes difficult for researchers to keep track of the current state of research. Unlike the current classification methods that use citation intent, from a reverse perspective, we propose a method to Classify Citation Relationships based on the Contributions of Cited papers. This classification method can count the number of citations for each contribution, which can be used as a feature of a paper summarization system to generate a summary. Since the number of citations changes over time, the generated paper summary is dynamic. It can also generate a citation summary based on the citations of each contribution. We build a dataset for this method called C2RC2. We achieve an accuracy of 0.7896 on the test set using the SciBERT model, which indicates that it is feasible to classify citation relations by the contributions of cited papers.