Semantic Relevance Feedback Based on Local Context

Abstract

Semantic similarity is a measure based on the meaning of the word and represented by the type of semantic relationships between the meanings of two words. The semantic similarity between words or sentences can be calculated by following the method, WordNet-based similarity. In our suggested method, the proposed solution depends on the two-step to return the target document collection. Once the top d1, d2…, dn documents have been retrieved using the initial query. The second compares these retrieved documents with the query, this step takes the most frequent senses of sentences in d1, d2…, dn with the senses of the query sentences. Query enrichment is then carried out adding terms semantically related to these senses the aim is to resolve the issue of incompatibility and the ambiguity problem affecting the results of information retrieval through the use of techniques to Word sense disambiguation (WSD) the word and the results be more semantically related with the query. Our experiment, carried out using part of the fire2011 data set.