Reasearch on User Profile Based on User2vec
Personalized services for information overload are becoming more common with the arrival of the era of big data. Massive information also makes the Internet platform pay more attention to the accuracy and efficiency of personalized recommendations. The user’s profile is constructed to describe the user information of the relevant platform more accurately and build virtual user features online through user behavior preference information accumulated on the platform. In this paper we propose a new user mode named user2vec for personalized recommendation. The construction of user2vec relies on platform and extremely targeted. At the same time, user profile is dynamically changing and need to be constantly updated according to the data and date, therefore we define a new time decay function to track time changes. Dynamic description of user behavior and preference information through user vectorization combined with time decay function can provide reference information for the platform more effectively. Finally, we using a layered structure to build an overall user profile system. And the experiment adapts content-based recommendation algorithm to indirectly prove effectiveness of user profile model. After many sets of experiments proved, it can be found that the proposed algorithm is effective and has certain guiding significance.
KeywordsUser profile User2vec Time decay function Personalized recommendation
We would like to thank all colleagues and students who helped for our work. This research was partially supported by (1) National Social Science Fund Project: Research on Principles and Methods of Electronic Document Credential Guarantee in Cloud Computing Environment (Project No. 15BTQ079); (2) Special Project for Civil Aircraft, MIIT; (3) Fund of Shanghai Engineering Research Center of Civil Aircraft Health Monitoring.
- 1.Mulder, S., Yarr, Z.: Win in the User. Machinery Industry Press, Beijing (2007)Google Scholar
- 7.Liang, C.: User profile for personalized web search. In: International Conference on Fuzzy Systems and Knowledge Discovery, Shanghai, IEEE, pp. 1847–1850 (2011)Google Scholar
- 8.Kassak, O., Kompan, M., Bielikova, M.: User preference modeling by global and individual weights for personalized recommendation. Acta Polytech. Hung. 12(8), 27–41 (2015)Google Scholar
- 9.Chen, Y., Yu, Y., Zhang, W., et al.: Analyzing user behavior history for constructing user profile. In: Proceeding of 2008 IEEE International Symposium on IT in Medicine and Education, pp. 343–348. IEEE (2008)Google Scholar
- 12.Tchuente, D., Canut, M., Jessel, N.P.: Visualizing the relevance of social ties in user profile modeling. Web Intell. Agent Syst. Int. J. 10, 261–274 (2012)Google Scholar
- 17.Sahijwani, H., Dasgupta, S.: User Profile Based Research Paper Recommendation. https://uspmes.daiict.ac.in/btpsite. Accessed 5 June 2018
- 18.Ouaftouh, S., Zellou, A., Idri, A.: UPCAR: user profile clustering based approach for recommendation. In: International Conference on Education Technology and Computers, Association for Computing Machinery, Barcelona, pp. 17–21 (2017)Google Scholar
- 19.Mikolov, T., Chen, K., Corrado, G., et al.: Efficient estimation of word representations in vector space, pp. 1–12 (2013)Google Scholar
- 20.Le, Q., Mikolov, T.: Distributed representations of sentences and documents. In: Proceedings of the 31st International Conference on Machine Learning, Beijing, China, JMLR W&CP, vol. 32 (2014)Google Scholar
- 21.Ranking Algorithm Based on User Voting (4): Newton’s Cooling La. https://blog.csdn.net/zhuhengv/article/details/50476306. Accessed 5 June 2018
- 22.Aggarwal, C.C.: Recommender Systems: The Textbook. Springer, Basel (2016)Google Scholar
- 23.Gasparic, M.: Context-based IDE command recommender system. In: RecSys 2016, September 15–19, Boston, MA, USA, pp. 435–438 (2016)Google Scholar