Personalized Web Search using Machine Learning Technique

Abstract: Machine learning is a field which deals with the creation and analysis of algorithms that can follow a reinforced methodology of learning. Such algorithms work on the basis of predictions made based on inputs, rather than following programmed instructions. In the existing search engines, if the user wants to know his likeness based on his previous searches, nothing except the history is present in the browser. In the proposed system, a framework has been designed to get a user specific re-ranking for the various websites and pages from an already present search engine and to implement Support Vector Machine Algorithm to find out the weightage of each concept. The main objective of the proposed system is to provide personalized results to the users based on their individual interests. It re- ranks the results for a given query obtained from existing search engine based on the user need. Profile database is got from a search engine for a given user query. Topical preferences are symbolized by words that occur regularly, referred as concepts. Weights are assigned to these concepts and this shows the user�s degree of interest in that concept. To show the relations among the different concepts, required information is saved along with its strength. For every query�s outcome, a score is given to each snippet based on weights of different concepts that they have and new ranking is done based on the scores. Currently web search personalization is used by social networking sites, search engines, banking services and online shopping sites.

Keywords: Machine Learning, Personalized Web Search Engine, Webpage Re-ranking, Supervised and Unsupervised Machine Learning, Support Vector Mechanism.