Information extraction of extend relation in scientific papers

摘要

Extend relation is one of the papers' relations obtained by using the citation-context based approach where this approach can identify more varied papers relations than two other approaches i.e. content-based and citation-analysis. However, the recent research with citation-context based approach only focuses on identification of relations, and cannot give out information related to this relation in more detail. The information of relation is very necessary for researchers in developing a research. In this paper, the scheme of information extraction process for extend relations is proposed which covers the classification process of extend categories and information extraction process based on the classification of extend categories. The experimental results show that the most optimal process in extend information extraction is reached by classifying extend categories from all sentences directly. The combination between the best extend categories classification and the proposed regular expression for relations information extraction delivered F-Measure more than 50%. This result can be said as good by considering that the proportion of extend sentence in the corpus is very small. The other experiment also shows that the optimal rule is reached for Data and Tool category.