One of the major challenges in the web is accessing relevant information according to the needs of the user. With each user having different information needs in relation to his/her query, the search results should be personalized according to the information needs of the users. Personalized ontology model for knowledge representation and reasoning over user profiles fails to match the local instance repository queries (i.e.,) users with global information base (i.e.,) web database. Although the application of ontology model has been underway for many years and many algorithms related to ontology have been developed, it is not applicable to the majority of the existing web documents. Query planning for Weighted Additive Aggregation Queries (WAAQ) obtained optimal set of sub queries with incoherency bounds and least number of refresh messages were sent from aggregators to the client. But WAAQ is not effective in developing the cost model for complex queries. To overcome the issues related to complex queries, Predictive User Preference Identification (PUPI) technique is developed based on the relational users and stored queries. PUPI technique searches the results according to each user’s need based on their relevant information with little effort from the side of the user, followed by it the effectiveness with complex queries are verified. PUPI then extends to Condition Redefine Query which redefines the queries according to the user relational profile and stored query context. In order to learn user’s preference, PUPI make use of both semantic and lexical information. PUPI take into account the web queries for the current user task and engender a new query language model for redefining based on the user query model and the considering the user profile. PUPI technique performed experiment on Weka tool using the MSNBC.com Anonymous Web Data. By analyzing the results, PUPI technique is better than using the state-of-art methods. Experiments were conducted on the factors such as execution time, micro averaged accuracy, query user relationship ratio, query frequency similarity and true positive rate to reveal the efficiency of the technique, query frequency similarity and true positive rate. © 2005 - 2014 JATIT & LLS. All rights reserved.