A comparative study on content-based paper-to-paper recommendation approaches in scientific literature
Bahareh Kazemi, Abdolreza Abhari · Communications and Networking Symposium · 2017
This paper deals with analysis and comparison of two well-known content-based recommendation approaches for scientific papers in biomedical domain. Given a rich set of abstracts for thousands of articles from PUBMED, a series of efficient pre-processing techniques are proposed. Then, for the first approach, a Term-frequency Inverse-document-frequency (TF-IDF) algorithm is formulated to make recommendations for the paper-set. Alternatively, we also use word-embedding to represent papers' abstract text and employ the extracted representation for the recommendation construction. Experimental results will evaluate and compare the efficiency and suitability of any of the proposed frameworks in building a universal paper-to-paper recommendation engine.