Abstract: Search engine aims to produce relevant and correct results to a user for their query that they have requested. In this paper, we propose to represent a flexible tag management scheme with automated textual descriptors in userís profile. We develop a recommendation based personalized system that operates on clustering tags based on the category. In this scheme we implement weighted k means clustering approach to effectively cluster the history of user interests which provides flexibility among multiple user environments. We use a synthetic dataset to show our system performance .The main objective of this system is to develop interest based recommendation system using userís past rendition. This methodology is developed to produce relevant query results which are based on the userís previous historical data and to generate personalized user profile with that generated data. Here we implement effective word embedding model to extract the similar words from the extracted corpus. We use LDA scheme with incremental learning algorithm for effective data processing. Incremental algorithm is used for accurate query retrieval even the data changes occurred in user behavior. Then we calculate rank for each repeated words. This calculation mainly focuses on users profile information and historical data of the userís profile.
Keywords: Clustering, recommendation, query, LDA, Incremental Algorithm.