APPLYING ROUGH SET THEORY TO GENETIC ALGORITHM FOR WEB SERVICE COMPOSITION

Rough set theory is a very efficient tool for imperfect data analysis, especially to resolve ambiguities, classify raw data and generate rules based on the input data. It can be applied to multiple domains such as banking, medicine etc., wherever it is essential to make decisions dynamically and generate appropriate rules. In this paper, we have focused on the travel and tourism domain, specifically, Web-based applications, whose business processes are run by Web Services. At present, the trend is towards deploying business processes as composed web services, thereby providing value-added services to the application developers, who consumes these composed services. In this paper, we have used Genetic Algorithm (GA), an evolutionary computing technique, for composing web services. GA suffers from the innate problem of larger execution time when the initial population (input data) is high, as well as lower hit rate (success rate). In this paper, we present implementation results of a new technique of solving this problem by applying two key concepts of rough set theory, namely, lower and upper approximation and equivalence class to generate if-then decision support rules, which will restrict the initial population of web services given to the genetic algorithm for composition.