作者: Fidel Cacheda , Iadh Ounis , Vassilis Plachouras
DOI:
关键词: Relevance (information retrieval) 、 Vector space model 、 Adversarial information retrieval 、 Data mining 、 Query expansion 、 Information retrieval 、 Data retrieval 、 Computer science 、 Human–computer information retrieval 、 Document retrieval 、 Term Discrimination
摘要: The combination of evidence for Information Retrieval has been studied extensively in order to increase effectiveness. In this paper, we study the selective application different retrieval approaches on a per-query basis Web Retrieval. Our methodology is based assumption that not all queries benefit from uniform approach. select most appropriate each query, use hyperlink structure retrieved documents and also distribution aggregates, groups same domain. experimental results show it possible obtain important improvements effectiveness by using simple statistical decision mechanisms basis.