Improving Mining Performance in Bigdata

Abstract

The World Wide Web (WWW) has become a well-known source of information. newlineHowever, the information overload over the internet has created a noteworthy challenge in newlineproviding accurate information. Sometimes, the browsing of appropriate information newlineproves time-consuming or ineffective. When users interact with the web, digital footprints newlinesuch as web sites are recorded on the web servers in the form of web server log files in newlinechronological order. These log files consist of a significant amount of information about newlineweb users previous browsing activities and are shown to be a great source of information. newlineBased on the stored log files, users behavior prediction plays a significant role in web newlineusage mining. In recent days, several web pattern classification methods have been newlineemployed to predict web user behavior. But, the performance of the conventional technique newlinewas not efficient since the complexity was too high. In order to overcome the above newlinelimitations, research work is concentrated on developing novel IoT based machine learning newlineand deep learning techniques as follows. newlineFirstly, an IoT based Target Projective Pursuit Pre-processed Modest Adaptive newlineBoost Clustering (IoT-TPPPMABC) method was developed to enhance the web user newlinebehavior prediction performance. The main aim of the IoT-TPPPMABC Method is to newlinecollect and discover the frequently accessed patterns from weblogs depending on the newlinenavigational behavior of users. In the IoT-TPPPMABC method, a Target Projective Pursuit newlinePre-processing is performed to eliminate unwanted patterns and select appropriate web newlinepatterns from the weblogs. Afterward, Modest Adaptive Boost Clustering (MABC) was newlineintroduced in the IoT-TPPPMABC Method to find the frequently accessed web patterns in newlineii newlinea particular session with less time complexity. MABC is a collection of brown clusters to newlineform the final strong clustering results with a minimum error rate. This, in turn, helps to newlineimprove the web usage behavior pattern mining performance. newlineNext, an IoT Aware Margin Truncative S