Skip to main content

Improving Personalized Search on the Social Web Based on Similarities Between Users

Zhenghua Xu‚ Thomas Lukasiewicz and Oana Tifrea−Marciuska


To characterize a user's preference and the social summary of a document, the user profile and the general document profile are widely adopted in the existing personalized ranking functions. However, in many real world situations, using these two profiles can not primely personalize the search results on the social Web because (i) different people usually have different perceptions on a same document and (ii) the information contained in the user profile is usually not comprehensive enough to characterize a user's preference. Therefore, we propose a dual personalized ranking (D-PR) function to improve the personalized search on the social Web by introducing two novel profiles: the extended user profile and the personalized document profile, to better characterize a user's preference and better summarize his personal perception on a document. Instead of using a same general document profile for all users, for each of the documents, our method computes each individual user a personalized document profile to characterize his perception on this document; while the extended user profile is defined as the sum of all of the user's personalized document profiles. Moreover, how to obtain the personalized document profile is a challenge, we propose a method to estimate it utilizing the perception similarities between users. A method used to quantify the perception similarity is also presented. The experimental results show that our D-PR ranking function achieves better personalized ranking on the social Web than the state-of-the-art method.

Book Title
Proceedings of the 8th International Conference on Scalable Uncertainty Management‚ SUM 2014‚ Oxford‚ UK‚ September 15−17‚ 2014
Umberto Straccia and Andrea Calì
Lecture Notes in Computer Science