Personalized web search (PWS) has established its effectiveness in increasing the quality of different search services at the internet. But, evidences show that users’ reluctance to disclose their private statistics at some point of search has emerge as a major barrier for the wide proliferation of PWS. We observe privacy safety in PWS applications that model consumer options as hierarchical cusumer profiles. We advise a PWS system called UPS that could adaptively generalize profiles through queries even as respecting user-specified privateness necessities. Our runtime generalization goals at strikeing a stability among two predictive metrics that compare the utility of personalization and the privateness threat of revealing the generalized profile. We present two greedy algorithms, namely GreedyDP and GreedyIL, for runtime generalization. We additionally provide an internet prediction mechanism for identifying whether or not personalizing a query is useful. large experiments show the effectiveness of our framework. The experimental results also monitor that GreedyIL significantly outperforms GreedyDP in terms of performance.