Genomics information retrieval using a Bayesian model for learning and re-ranking

摘要: The use of large-scale experimental techniques and biomedical tools has increased the pace at which biologists produce useful information. This promotes us to propose a Bayesian model for learning re-ranking boost genomics information retrieval performance. We first describe general discovering property each passage. Then, we examine Bernoulli distribution as prior provide an efficient way obtain training passages parameter estimation, according characterizations distribution. Later, evaluate our proposed by conducting extensive experiments on TREC 2007 2006 Genomics data sets. results show effectiveness improving performance two years' Furthermore, conclusions future prospects are also discussed.