Recognition and Application of Target Issue Entities in Biomedical Literature
Given a research issue about specific biomedical entities (e.g., chemicals, genes, and diseases), biomedical researchers often need to spend much effort searching for highly related articles, and then carefully reading many articles for cross validation. In this project, we plan to explore the development and application of a technique that, given the title and the abstract of a biomedical article a, extracts those entities (in the title and the abstract) that are related to the target research issue of a. An entity related to the target issue is named target issue entity (TIE), and the technique to be developed is named TIER (TIE Recognizer). TIER can serve as a front-end processor for several applications in which the input can be a target biomedical entity, article, or pair of associated entities, while the output can be a ranked list of entities, articles, or association pairs that are highly related to the target input. TIER is the first technique that recognizes TIEs from article titles and abstracts, which are publicly obtainable on the Internet, making it able to recommend information more comprehensively and properly. Performance of TIER will be evaluated in two applications, with several state-of-the-art techniques as the baselines. Contributions of this project are of both technical and practical significance to the dissemination, curation, and cross-validation of the highly related evidence already published in biomedical literature.
Keywords: biomedical literature, biomedical entities; target issue entity, entity recognition.