Genetic Algorithms for Query Optimization in Information Retrieval: Relevance Feedback

Evolutionary improvement of search queries and its parameters

“…Two techniques are being used for query enhancement query term weighting using numeric weights or linguistic variables and Boolean conjunction parameterization for expressing relationships among the extremes of AND, OR, NOT etc. [2].…”

“…Donald Kraft proposed in [2] the usage of Gerard Saltons tf × idf t indexing formula introduced for vector space IR model as document indexing mechanism in extended Boolean IR model.…”

“…Query language is in extended Boolean IR model enhanced with the possibility of weighting query terms in order to express different importance of those in search request and by weighting (parameterizing) aggregation operators to soften or blur their impact on query evaluation [1], [2]. Consider Q to be the set of user queries over a collection then the weight of term t in query q is denoted as a(q, t) satisfying…”

“…For RSV enumeration is crucial the interpretation of query term weight a. The most used interpretations are to see query term weight as importance weight, threshold or ideal document description [1], [2].…”

Implicit User Modelling Using Hybrid Meta-Heuristics

“…To support human level IR task, user query was evolved over an IR model describing the collection of retrieved documents taking the document summary s as content of the document. The IR model was created by the means of extended Boolean IR model featuring document representation as fuzzy set of index terms and Boolean search queries [8,9,10].…”