Similarity or inference for assessing relevance in information retrieval

The primary objective of an information retrieval system is to provide the user with relevant documents given the user's query. Numerous approaches to determining a document's relevance have been proposed based on the underlying retrieval model. In the paper, similarity-based and inference-based IR models are examined. A fuzzy information retrieval model is presented such that assessing document relevancy is based on the user's interpretation of the query term weights. The close relationship between the similarity-based and inference-based IR models is investigated in the context of the fuzzy information retrieval model.