Using Statistical Machine Translation Model to Improve Domain-Specific Metasearch Engines

In order to improve the recall of the domain-specific information retrieval, an efficient query expansion mechanism is proposed for the metasearch engines. This mechanism uses the statistical machine translation model to compute the relevance between general query words and domain-relevant query words and dispatches the expanded queries to component search engines. The key ingredient of translation model is the expectation maximization (EM) algorithm. The experimental results show that the proposed expansion mechanism is a desirable and efficient method to improve the domain-relevance of the pages returned by a metasearch engine.