Soft approaches to distributed information retrieval Original Research Article

In this paper we consider the context of information retrieval in a network. Two distinct situations of distributed retrieval are considered: the case in which we have distinct databases residing on distinct servers having their own IRSs, and the case in which we have a unique, huge and distributed information repository (like the WWW), and distinct IRSs (search engines), which can be used to inquiry the same collection. In both situations we have the problem of merging the results produced by the distinct IRS into a unique ordered list, which constitutes the unified answer to the user query. In solving the problem of list fusion two main issues are involved. The first one concerns the decision of how many retrieved documents to select from each individual documents list. The second aspect is to define a procedure to fuse the individual lists of retrieved documents into an overall ordered list. In this paper some fuzzy approaches to solve the two above-mentioned problems are presented.