Identifying the Conceptual Space of Citation Contexts using Coreferences.

The study of citation contexts is an important element in understanding the function of citations and categorizing the relationships between works. One of the problems in this field is defining the size of citation contexts. In this paper we propose the definition of citation blocks (CB) that are citation contexts composed of one or more sentences that are linked by coreference clusters. We describe the methodology for the automatic processing and determining the boundaries of CB and observe the different sizes of CB in the different sections of the IMRaD structure of articles. The results are obtained from a sample of 70,000 citation contexts extracted from the PLOS dataset.