融合协同过滤与上下文信息的Bandits推荐算法
|
摘要
随着推荐算法在众多领域的广泛应用,冷启动问题得到了越来越多的关注。针对仅可获得老用户对商品文字评价的场景,提出了一套解决用户冷启动问题的方案与算法。首先通过分析发现了文章主题提取与基于商品评价提取特征的相似性,因此引入自然语言处理领域的LDA(latent Dirichlet allocation)生成模型提取商品潜在特征;然后在传统Bandits算法的基础上融入邻居用户的协同作用提出了COLINBA(collaborativefiltering context linear Bandits)算法,该算法通过相似度权重因子控制邻居用户对推荐结果的贡献,使得协同作用更加精确有效,推荐完成后根据用户真实反馈以及所推荐商品的特征更新用户特征。最后采用真实数据集Delicious和Last.fm将该算法与该领域的最新方法进行比较,实验结果表明该算法对推荐效果有提升作用。
With the wide application of the recommender system in various domains, the cold start problem has gained increasing attention in recent years. For the situation where only text comments of items are available, this paper proposes a set of solution and algorithm that can solve the cold start problem of new users. Firstly, the similarity between the article topic extraction and the feature extraction based on commodity evaluation is found and the latent Dirichlet allocation model is used to extract the features of items innovatively. Then the COLINBA algorithm is proposed which introduces the synergies of neighboring users based on the traditional Bandits strategy.This algorithm adds the neighbors' similarity weight coefficient which determines the influence degree of the neighbors on the target user and makes the synergy more accurate and effective. The user features are updated according to the feedback of the user and the recommended item features. Finally, based on the two real world datasets, Delicious and Last.fm, this paper experimentally compares COLINBA to the state-of-the-art methods, and the results show that COLINBA offers a significant increase in recommendation performance.
With the wide application of the recommender system in various domains, the cold start problem has gained increasing attention in recent years. For the situation where only text comments of items are available, this paper proposes a set of solution and algorithm that can solve the cold start problem of new users. Firstly, the similarity between the article topic extraction and the feature extraction based on commodity evaluation is found and the latent Dirichlet allocation model is used to extract the features of items innovatively. Then the COLINBA algorithm is proposed which introduces the synergies of neighboring users based on the traditional Bandits strategy.This algorithm adds the neighbors' similarity weight coefficient which determines the influence degree of the neighbors on the target user and makes the synergy more accurate and effective. The user features are updated according to the feedback of the user and the recommended item features. Finally, based on the two real world datasets, Delicious and Last.fm, this paper experimentally compares COLINBA to the state-of-the-art methods, and the results show that COLINBA offers a significant increase in recommendation performance.
引文
[1] Lu J, Wu D S, Mao M S, et al. Recommender system application developments:a survey[J]. Decision Support Systems, 2015, 74(3):12-32.
[2] Ma H, King I, Lyu M R. Learning to recommend with social trust ensemble[C]//Proceedings of the 32nd Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, Boston, Jul 19-23, 2009.New York:ACM, 2009:203-210.
[3] Li L H, Chu W, Langford J, et al. A contextual-bandit approach to personalized news article recommendation[C]//Proceedings of the 19th International Conference on World Wide Web, Raleigh, Apr 26-30, 2010. New York:ACM, 2010:661-670.
[4] Zhang J, Lin Y J, Lin M L, et al. An effective collaborative filtering algorithm based on user preference clustering[J].Applied Intelligence, 2016, 45(2):230-240.
[5] Gentile C, Li S, Zappella G. Online clustering of bandits[C]//Proceedings of the 31st International Conference on Machine Learning, Beijing, Jun 21-26, 2014. Red Hook:Curran Associates, 2014:757-765.
[6] Cheng S, Wang B L, Mao L H, et al. Multi-armed bandit recommender algorithm with matrix factorization[J]. Journal of Chinese Computer Systems, 2017, 38(12):2754-2758.
[7] Zhang Z J, Xu G W, Zhang P F, et al. Personalized recommendation algorithm for social networks based on comprehensive trust[J]. Applied Intelligence, 2017, 47(3):659-669.
[8] Li S, Karatzoglou A, Gentile C. Collaborative filtering bandits[C]//Proceedings of the 39th International ACM SIGIR Conference on Research and Development in Information Retrieval, Pisa, Jul 17-21, 2016. New York:ACM, 2016:539-548.
[9] Maatallah M, Seridi-Bouchelaghem H. A fuzzy hybrid approach to enhance diversity in top-N recommendations[J].International Journal of Business Information Systems,2015, 19(4):505-530.
[10] Auer P. Using confidence bounds for exploitation-exploration trade-offs[J]. Journal of Machine Learning Research, 2002,3:397-422.
[11] Blei D M, Ng A Y, Jordan M I. Latent Dirichlet allocation[J]. Journal of Machine Learning Research, 2012, 3:993-1022.
[12] BíróI, Siklósi D, SzabóJ, et al. Linked latent Dirichlet allocation in web spam filtering[C]//Proceedings of the International Workshop on Adversarial Information Retrieval on the Web, Madrid, Apr 21, 2009. New York:ACM, 2009:37-40.
[13] Zhang H Z, Qiu B J, Giles C L, et al. An LDA-based community structure discovery approach for large-scale social networks[C]//Proceedings of the IEEE International Conference on Intelligence and Security Informatics, New Brunswick,May 23-24, 2007. Piscataway:IEEE, 2007:200-207.
[14] Misra H, Yvon F, CappéO, et al. Text segmentation:a topic modeling perspective[J]. Information Processing&Management, 2011, 47(4):528-544.
[15] Huang J J, Zhu K L, Zhong N. A probabilistic inference model for recommender systems[J]. Applied Intelligence,2016, 45(3):686-694.
[16] Zhou X Z, Wu S X. Rating LDA model for collaborative filtering[J]. Knowledge-Based Systems, 2016, 110:135-143.
[17] Griffiths T L, Steyvers M. Finding scientific topics[J]. Proceedings of the National Academy of Sciences of the United States of America, 2004, 101(S1):5228-5235.
[18] Nguyen T T, Lauw H W. Dynamic clustering of contextual multi-armed bandits[C]//Proceedings of the 23rd ACM International Conference on Information and Knowledge Management, Shanghai, Nov 3-7, 2014. New York:ACM, 2014:1959-1962.
[19] Resnick P, Iacovou N, Suchak M, et al. GroupLens:an open architecture for collaborative filtering of netnews[C]//Proceedings of the Conference on Computer Supported Cooperative Work, Chapel Hill, Oct 22-26, 1994. New York:ACM,1994:175-186.
[6]成石,王宝亮,毛陆虹,等.融合矩阵分解的多臂赌博机推荐算法[J].小型微型计算机系统, 2017, 38(12):2754-2758.
[2] Ma H, King I, Lyu M R. Learning to recommend with social trust ensemble[C]//Proceedings of the 32nd Annual International ACM SIGIR Conference on Research and Development in Information Retrieval, Boston, Jul 19-23, 2009.New York:ACM, 2009:203-210.
[3] Li L H, Chu W, Langford J, et al. A contextual-bandit approach to personalized news article recommendation[C]//Proceedings of the 19th International Conference on World Wide Web, Raleigh, Apr 26-30, 2010. New York:ACM, 2010:661-670.
[4] Zhang J, Lin Y J, Lin M L, et al. An effective collaborative filtering algorithm based on user preference clustering[J].Applied Intelligence, 2016, 45(2):230-240.
[5] Gentile C, Li S, Zappella G. Online clustering of bandits[C]//Proceedings of the 31st International Conference on Machine Learning, Beijing, Jun 21-26, 2014. Red Hook:Curran Associates, 2014:757-765.
[6] Cheng S, Wang B L, Mao L H, et al. Multi-armed bandit recommender algorithm with matrix factorization[J]. Journal of Chinese Computer Systems, 2017, 38(12):2754-2758.
[7] Zhang Z J, Xu G W, Zhang P F, et al. Personalized recommendation algorithm for social networks based on comprehensive trust[J]. Applied Intelligence, 2017, 47(3):659-669.
[8] Li S, Karatzoglou A, Gentile C. Collaborative filtering bandits[C]//Proceedings of the 39th International ACM SIGIR Conference on Research and Development in Information Retrieval, Pisa, Jul 17-21, 2016. New York:ACM, 2016:539-548.
[9] Maatallah M, Seridi-Bouchelaghem H. A fuzzy hybrid approach to enhance diversity in top-N recommendations[J].International Journal of Business Information Systems,2015, 19(4):505-530.
[10] Auer P. Using confidence bounds for exploitation-exploration trade-offs[J]. Journal of Machine Learning Research, 2002,3:397-422.
[11] Blei D M, Ng A Y, Jordan M I. Latent Dirichlet allocation[J]. Journal of Machine Learning Research, 2012, 3:993-1022.
[12] BíróI, Siklósi D, SzabóJ, et al. Linked latent Dirichlet allocation in web spam filtering[C]//Proceedings of the International Workshop on Adversarial Information Retrieval on the Web, Madrid, Apr 21, 2009. New York:ACM, 2009:37-40.
[13] Zhang H Z, Qiu B J, Giles C L, et al. An LDA-based community structure discovery approach for large-scale social networks[C]//Proceedings of the IEEE International Conference on Intelligence and Security Informatics, New Brunswick,May 23-24, 2007. Piscataway:IEEE, 2007:200-207.
[14] Misra H, Yvon F, CappéO, et al. Text segmentation:a topic modeling perspective[J]. Information Processing&Management, 2011, 47(4):528-544.
[15] Huang J J, Zhu K L, Zhong N. A probabilistic inference model for recommender systems[J]. Applied Intelligence,2016, 45(3):686-694.
[16] Zhou X Z, Wu S X. Rating LDA model for collaborative filtering[J]. Knowledge-Based Systems, 2016, 110:135-143.
[17] Griffiths T L, Steyvers M. Finding scientific topics[J]. Proceedings of the National Academy of Sciences of the United States of America, 2004, 101(S1):5228-5235.
[18] Nguyen T T, Lauw H W. Dynamic clustering of contextual multi-armed bandits[C]//Proceedings of the 23rd ACM International Conference on Information and Knowledge Management, Shanghai, Nov 3-7, 2014. New York:ACM, 2014:1959-1962.
[19] Resnick P, Iacovou N, Suchak M, et al. GroupLens:an open architecture for collaborative filtering of netnews[C]//Proceedings of the Conference on Computer Supported Cooperative Work, Chapel Hill, Oct 22-26, 1994. New York:ACM,1994:175-186.
[6]成石,王宝亮,毛陆虹,等.融合矩阵分解的多臂赌博机推荐算法[J].小型微型计算机系统, 2017, 38(12):2754-2758.