ANNES '95 Proceedings of the 2nd New Zealand Two-Stream International Conference on Artificial Neural Networks and Expert Systems
CNSR '04 Proceedings of the Second Annual Conference on Communication Networks and Services Research
Principles and Applications of Electrical Engineering
Principles and Applications of Electrical Engineering
Trust Inference in Web-Based Social Networks Using Resistive Networks
ICIW '08 Proceedings of the 2008 Third International Conference on Internet and Web Applications and Services
Scientific paper summarization using citation summary networks
COLING '08 Proceedings of the 22nd International Conference on Computational Linguistics - Volume 1
Hi-index | 0.00 |
Finding scientific papers and journals relevant to a particular area of research is a concern for many people including students, professors, and researchers. A subject classification of papers facilitates the search process. That is, having a list of subjects in a research field, we try to find out to which subject(s) a given paper is more related. This task can be done manually by, for example, asking authors to assign one or more categories at submit time. However, categorizing a large collection of resources manually is a time consuming process. In automatic methods, a naive strategy is to do a keyword-based search for the subject term in paper's title, keywords, and even its fulltext. Nonetheless, this approach fails for resources employing semantically equivalent terms but not exactly the same subject words. Besides, processing the whole text of a paper takes a long time. In this paper, we introduce a novel supervised approach for subject classification of scientific articles based on analysis of their interrelationships. We exploit links such as citations, common authors, and common references to assign subject to papers. Our experimental results show that our approach works well especially when the graph of relationships is dense enough, i.e., there are significant number of links among papers.