User Web Access Prediction Based On Web Services And User Profile

Abstract

With the growing use of web services in social networks, user behavior prediction for web access becomes significant and can minimize the perceived latency. The profile of the web user is an essential element in this prediction. However, this profile may contain several attributes that remain more or less significant and negatively influence this prediction. This paper presents a strategy for classifying web users and predicting their web services access behavior. The method uses neural networks as a database optimizer, removing irrelevant descriptors from the database using a new filtering technique called UPDS (User Profile Descriptors Selection), and as a classifier, with the predicted class representing the available web services. The proposed strategy appears to be promising, according to a case study.