Abstract

MotivationThe rapidly increasing size of biomedical databases such as MEDLINE requires the use of intelligent data mining methods for information extraction and summarization. Existing biomedical text-mining tools have limited capabilities for inferring topological and network relationships between biomedical terms. Very often too much is returned during summarization leading to information overload.\n\nResultsWe present herein SEACOIN 2.0, an interactive knowledge discovery and hypothesis generation tool for biomedical literature.SEACOIN generates k-ary relational networks of biomedical terms using a novel term ranking scheme to facilitate efficient information retrieval, summarization, and visual data exploration. Summarization is presented via multiple dynamic visualization panels. We evaluate the system performance in information retrieval and features extraction using the BioCreative 2013 Track 3 learning corpus. An average F-measure of 94% was achieved for document retrieval and an average precision of 88% was achieved for identification of top co-occurrence terms. The system allows interactive mining of complex implicit and explicit relationships among biomedical entities (genes, chemicals, diseases/disorders, mutations, etc.) and provides a framework for hypothesis generation. It also improves our understanding of various biological processes and disease mechanisms.\n\nContacteva.lee@gatech.edu