Topic Model based Privacy Protection in Personalized Web Search
Switch branches/tags
Nothing to show
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Type Name Latest commit message Commit time
Failed to load latest commit information.

Topic Model based Privacy Protection in Personalized Web Search

In modern search engines, personalization introduces potential risk of revealing user privacy by identifying their underlying interests from their logged search behaviors. As a solution of this problem we proposed a client-centered approach to hide user intents. We prevent search engine from building user profile by injecting cover queries with the original user issued query in a controlled way. Each user’s query is submitted with a set of cover queries which carry similar amount of information to generalize user profile while still having some personalization. We used LDA-C topic modeling to generate topic proportion of the original query. By using this topic proportion we used language model to generate cover queries of similar entropy but on unrelated topics. The performance of our proposed model is evaluated in terms of user profile variation from the initial user profile in privacy settings and non-privacy settings.

Figure: Topic-based Privacy Protection (TPP) Framework


Wasi Uddin Ahmad, Md Masudur Rahman and Hongning Wang. Topic Model based Privacy Protection in Personalized Web Search. The 39th International ACM SIGIR Conference on Research and Development in Information Retrieval (SIGIR'16), p1025-1028, 2016. (PDF)


If you use this work for academic work, please cite:

  title={Topic Model based Privacy Protection in Personalized Web Search},
  author={Ahmad, Wasi Uddin and Rahman, Md Masudur and Wang, Hongning},
  booktitle={Proceedings of the 39th International ACM SIGIR conference},