Searching the Web in Context: Genetic Algorithms for Exploring Query Space

CECCHINI R.L.; LORENZETTI C.M.; MAGUITMAN A.G.; BRIGNOLE, N.B.

Mar del Plata, Buenos Aires, Argentina

Congreso; SSI: 5º Simposio Sobre la Sociedad de la Información - 36 JAIIO: 36th Conference on Computer Science and Operational Research; 2007

How to enable effective use of on-line information is an increasingproblem, as technology enhances the ability to publish informationrapidly and large quantities of information are instantly availablefor retrieval. In text-based web search, users’ information needs and candidatetext resources are typically characterized by terms. Therefore,the quality of web retrieval is highly dependant on the vocabulary usedto generate the search queries. Selecting good query terms is extremelydifficult and presents many interesting research challenges. This paperapplies Genetic Algorithms to develop a framework for the dynamic identificationof “good query terms” to aid web search in the context of atheme. Starting with a population of terms from a thematic context, thetechniques presented in this paper incrementally identify good candidatesearch queries. After a few generations, the population of queries evolvestowards a set of queries that allows to retrieve material relevant to thegiven context. The proposed techniques have the potential to identifygood queries even if some of the terms in the queries do not occur in theoriginal context.