Research of User Profile Model in Personalized Search

Abstract:

For personalized service, existing user interest model primarily through the select weights Highest N keywords to represent the user interest model based on space vector method. The method of establishing the model is tend to content-based analysis methods and there is a serious "cold start" problem, cannot meet the demand for personalized services. Therefore, this paper add collaborative filtering factor in the process of establishing user interest model, and verified by experiment, after adding personalization features which make the service more obvious. In a certain extent, solve the new user's "cold start" problem.

• Periodical: Applied Mechanics and Materials (Volumes 543-547)
• Pages: 3364-3368
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.543-547.3364
• Online since: March 2014
• Authors: Yu Yang He*, Yan Tang
• Keywords: LSH-Minhash, Personalized Search, SimHash, Slope One, User Interest Model

References

[1] Daniel Lemire, Anna Maclachlan: Slope One Predictors for Online Rating-Based Collaborative Filtering (2005).

DOI: 10.1137/1.9781611972757.43

[2] Das, A. Datar, M. Garg, A. and Rajaram: Online Collaborative Filtering ( 2007).

[3] C Sadowski, G Levin Simhash: Hash-based similarity detection (2007).

[4] Nakano Y, Sagawa H, TAmmi H.: Proc of the 2005 Intemational Conference on Active Media Tiechnology (New York, USA, 2005), p.147-l 51.

[5] J. P. Zeng, S. Y. Zhang, C. R. Wu: Knowledge Based System, Vol. 21 (2008) No. 8, pp.905-910.

[6] Aleksandra Korolova, Krishnaram Kenthapadi, et al.: Proceedings of 18th International World Wide Web Conference, ( 2009), pp.171-180.