S2GA: A soft structured genetic algorithm, and its application in Web mining


We present a soft structured genetic algorithm (s2GA) that inherits all the advantages of its crisp (non-fuzzy) counterpart (sGA), but possesses several additional unique features compared to the sGA and other GA based techniques. We outline several strengths of the s2GA approach with regard to several emerging problems, such as its ability to address the scalability issue in a very eloquent manner for most data and Web mining problems. We also illustrate the use if s2GA for multimodal optimization by using it within a Deterministic Crowding framework, when used to find an unknown number of clusters underlying a data set Even though the proposed techniques inherit as legacy from the GA an almost unlimited number of different applications in all areas of science and engineering, we focus on an application of vital importance in today's networked environment-that of analyzing usage patterns on Web sites.

Publication Title

Annual Conference of the North American Fuzzy Information Processing Society - NAFIPS