Hybrid Fuzzy Inference System for Web RobotDetection

Nowadays, due to the dramatic expansion of theInternet, in various spheres, and dependence on informationresources it needs to create a secure environment, accessible tomultiple users, human users and intelligent agents, soft feel bemore. Web robots, there are plans for extracting knowledgefrom Web pages, that have begun their work, the number ofpages and views are all available - documentation available fromthese pages, recursively. Web robots, with destructive and nondestructiveconduct, along with human users, are consideredamong the web visitors. These intelligent agents are designed fornon-malicious purposes, such as collecting information, for thesearch engines, or malicious purposes, such as distributedattacks, which can leave the negative effects on web server. Inthis paper, a method is proposed, based on fuzzy inferencesystem for detection and classification of web visitors. Thepurpose of this paper is to introduce a hybrid fuzzy inferencesystem with instance based learning and a decision tree that canprovide in the context of Web robot detection acceptableperformance compared to other classification methods, such asneural networks, Support vector machine, Bayesian networks,and decision trees. Experiments have been carried out based onfour main criteria, accuracy, recall rate, F-measure and theROC curve.I. INTRODUCTION