NCBI at the 2014 BioASQ Challenge Task: Large-scale Biomedical Semantic Indexing and Question Answering

In this paper we report our participation in the 2014 BioASQ chal- lenge tasks on biomedical semantic indexing and question answering. For the biomedical semantic indexing task (Task 2a) where participating teams are pro- vided with PubMed articles and asked to return relevant MeSH terms, we built on our previous learning-to-rank framework with a special focus on systemati- cally incorporating results of complementary methods for improved perfor- mance. For the question answering task (Task 2b) where teams are provided with natural language questions and asked to return responses in the format of documents, snippets, concepts and RDF triplets (Phase A) and direct answers (Phase B), we relied on PubMed search engines and our state-of-the-art named entity recognition tools such as DNorm and tmVar in Phases A and B, respec- tively. The official challenge results demonstrate that we consistently per- formed better than the baseline approaches for Task 2a and Task 2b (Phase B), and ranked among the top tier systems in the 2014 challenge.