Learning to Refine Expansion Terms for Biomedical Information Retrieval Using Semantic Resources

摘要: With the rapid development of biomedicine, number biomedical articles has increased accordingly, which presents a great challenge for biologists trying to keep up with latest research. Information retrieval seeks meet this by searching among large based on given queries and providing most relevant ones fulfill information needs. As an effective technique, query expansion some room improvement achieve desired performance when directly applied because there exist many domain-related terms both in users' related articles. To solve problem, we propose framework learning-to-rank methods, refine candidate training term-ranking models select terms. train models, first pseudo-relevance feedback method MeSH then represent as feature vectors defining corpus-based term features resource-based features. Experimental results obtained TREC genomics datasets show that our can capture more expand original effectively improve performance.