In previous studies, Chinese text retrieval has often been dealt with on the character basis.This approach is not suited to deal with complex queries.We suggest that Chirmse text retrieval should work with words inslead of characters.The crucial problem is to segment originally continuous Chinese texts into words.In this paper, wc Ilrsi propose a hybrid segmentation approach which unifies the commonly used approaches.The systcm SMART is then udaptcd to index the segmented Chinese texls.Finally, wc suggest that Chinese text retrieval should move further to include a thesaurus in order to cope with dle rich vocabulary of Chinese.

Relevance, one of the fundamental notions in information retrieval (IR), has long been studied from a cognitive point of view. It is known that relevance depends not only on the topic of the document and the information need expressed in a query, but also on other ‘situational’ factors in the retrieval situation, such as the user's previous knowledge, background, intentions and so on. Formal models, on the other hand, usually consider relevance from a system point of view, i.e. they isolate relevance in a restricted context in which only the topic matters. One of the reasons for this very partial modeling is due to the inappropriateness of standard formal tools for describing relevance in a general context. This paper is an attempt to identify of more appropriate logical framework for the modeling. Counterfactual conditional logic is examined with respect to the IR requirements, indicating the logic's high potential value for this task. A particular conditional logic is then defined which, in comparison with previous developments on conditional logic, is better suited to IR. The new model gives an insight into the phenomena related to the ‘situational’ factors of relevance judgment which, until now, have not been considered.