Methods for accurate retrieval of MEDLINE citations

The lack of discipline and consistency in gene naming poses a formidable challenge to researchers in locating relevant information sources in the genomics literature. The research presented here primarily focuses on how to find the MEDLINE citations that describe functions of particular genes. We developed new methods and extended current techniques that may help researchers to retrieve such citations accurately. We further evaluated several machine learning and optimization algorithms to identify the sentences describing gene functions in given citations. Keywords: Genomics; MEDLINE; MeSH; Information Retrieval; Propositional Logic; Decision Lists; Machine Learning; Bayesian Networks; Model Averaging; Probabilistic Inference