Promoting Diversity in Top Hits for Biomedical Passage Retrieval

摘要: With the volume of biomedical literature exploding, such as BMC or PubMed, it is paramount importance to have scalable passage retrieval systems that allow researchers quickly find desired information. While topical relevance most important factor in text retrieval, an effective system needs also cover diverse aspects topic. Aspect-level performance means top-ranked passages for a topic should aspects. methods often involve clustering retrieved on basis textual similarity. We propose HIERDENC ranks passages, achieving scalability and improved aspect-level over other methods. runtimes scale large datasets, PubMed BMC. The consistently better than cosine similarity Hamming Distance-based comparable biclustering separation relevant improves topics where many are involved. Converting GO/MeSH ontological terms performance.