Relation based term weighting regularization

摘要: Traditional retrieval models compute term weights based on only the information related to individual terms such as TF and IDF. However, query are related. Intuitively, these relations could provide useful about importance of a in context other terms. For example, "perl tutorial" specifies that user look for relevant both perl tutorial. Thus, document containing should have higher relevance score than ones with one them. if IDF value "tutorial" is much smaller "perl", existing may assign lower those multiple occurrences "perl". It clear be dependent not collection statistics but also In this work, we study how utilize semantic among regularize weighting. Experiment results over TREC collections show proposed strategy effective improve performance.