Abstract：Review data in e-commerce websites implicates items’ features and users’ sentiment. Most existing recommendation researches based on aspect-level sentiment analysis capture users’ aspect preference for items by extracting users’ sentiment towards different aspects of items in the review data of a same category, ignoring that different category items have different aspects and that users’ aspect preference varies by time. A temporal-aware multi-category products recommendation model is proposed based on aspect-level sentiment analysis, which jointly models user, category, item, aspect, aspect-sentiment and time in order to find how users’ aspect preferences vary by time on different category items. This model is able to infer users’ aspect preferences for items at any time, which can provide users with explainable recommendations. Experiment results on two real-world data sets show that, in comparison to other recommendation models based on time or aspect-level sentiment analysis, the proposed model achieves significant improvement in the precision and recall for the top-N recommendation.
丁永刚,李石君,付星,刘梦君. 面向时序感知的多类别商品方面情感分析推荐模型[J]. 电子与信息学报, 2018, 40(6): 1453-1460.
DING Yonggang, LI Shijun, FU Xing, LIU Mengjun. Temporal-aware Multi-category Products Recommendation Model Based on Aspect-level Sentiment Analysis. JEIT, 2018, 40(6): 1453-1460.
LEUNG W K, CHAN C F and CHUNG F L. Integrating collaborative filtering and sentiment analysis: A rating inference approach[C]. Proceedings of European Conference on Artificial Intelligence Workshop, Riva del Garda, 2006: 300-307.
[2]
PEROŠ and HORVAT T. Opinion-driven matrix factorization for rating prediction[C]. Proceedings of User Mode-ling, Adaptation, and Personalization, Rome, Italy, 2013: 1-13. doi: 10.1007/978-3-642-38844-6_1.
[3]
SCHOUTEN K and FRASINCAR F. Survey on aspect-level sentiment analysis[J]. IEEE Transactions on Knowledge and Data Engineering, 2016, 28(3): 813-830. doi: 10.1109/TKDE. 2015.2485209.
[4]
WANG Feng and CHEN Li. Review mining for estimating users’ ratings and weights for product aspects[J]. Web Intelligence, 2015, 13(3): 137-152. doi: 10.3233/WEB-150317.
[5]
OU W and HUYNH V N. Rating supervised latent topic model for aspect discovery and sentiment classification in on-line review mining[C]. Proceedings of 13th International Conference on Modeling Decisions for Artificial Intelligence, Sant Julià de Lòria, Andorra, 2016: 151-164. doi: 10.1007 /978-3- 319-45656-0_13.
[6]
MUSTO C, DE GEMMIS M, SEMERARO G, et al. A multi-criteria recommender system exploiting aspect-based sentiment analysis of users, reviews[C]. Proceedings of the Eleventh ACM Conference on Recommender Systems, Como, Italy, 2017: 321-325. doi: 10.1145/3109859.3109905.
[7]
ZHANG Yongfeng, ZHANG Haochen, and ZHANG Min, et al. Do users rate or review? boost phrase-level sentiment labeling with review-level sentiment classification[C]. Proceedings of the 37th International ACM SIGIR Conference on Research and Development in Information Retrieval, Gold Coast, Australia, 2014: 1027-1030.
[8]
WANG Shuai, CHEN Zhiyuan, and LIU Bing. Mining aspect- specific opinion using a holistic lifelong topic model[C]. Proceedings of the 25th International Conference on World Wide Web, Montreal, Canada, 2016: 167-176. doi: 10.1145/ 2872427.2883086.
[9]
ZHANG Yongfeng, LAI Guokun, ZHANG Min, et al. Explicit factor models for explainable recommendation based on phrase-level sentiment analysis[C]. Proceeding of The 37th International ACM SIGIR Conference on Research & Development in Information Retrieval. Gold Coast, Australia, 2014: 83-92. doi: 10.1145/2600428.2609579.
[10]
CHEN Xu, XU Tao, ZHANG Yongfeng, et al. Learning to rank features for recommendation over multiple categories[C]. Proceedings of the 39th International ACM SIGIR conference on Research and Development in Information Retrieval, Pisa, Italy, 2016: 305-314. doi: 10.1145/2911451.2911549.
[11]
WU Yao and ESTER M. FLAME: A probabilistic model combining aspect based opinion mining and collaborative filtering[C]. Proceedings of the Eighth ACM International Conference on Web Search and Data Mining, Shanghai, China, 2015: 199-208. doi: 10.1145/2684822.2685291.
[12]
BAUMAN K, LIU Bing, and TUZHILIN A. Aspect based recommendations: recommending items with the most valuable aspects based on user reviews[C]. Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, Halifax, Canada, 2017: 717-725. doi: 10.1145 /3097983.3098170.
[13]
HE Yulan, LIN Chenghua, GAO Wei, et al. Dynamic joint sentiment-topic model[J]. ACM Transactions on Intelligent Systems & Technology, 2014, 5(1): 1-21.
[14]
DERMOUCHE M, VELCIN J, KHOUAS L, et al. A Joint model for topic-sentiment evolution over time[C]. Proceedings of 2014 IEEE International Conference on Data Mining, Shenzhen, China, 2014: 773-778. doi: 10.1109/ICDM. 2014.82.
[15]
HU Yan, XU Xiaofei, and LI Li. Analyzing topic-sentiment and topic evolution over time from social media[C]. Proceedings of the 9th International Conference on Knowledge Science, Engineering and Management, Passau, Germany, 2016: 97-109. doi: 10.1007/978-3-319-47650-6_8.
[16]
KOREN Y. Collaborative filtering with temporal dynamics [C]. Proceedings of the 15th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, Paris, France, 2009: 447-456. doi: 10.1145/1557019.1557072.
[17]
XIONG Liang, CHEN Xi, and HUANG Tzukuo, et al. Temporal collaborative filtering with bayesian probabilistic tensor factorization[C]. Proceedings of the SIAM International Conference on Data Mining, Columbus, USA, 2010: 211-222.
[18]
RAFAILIDIS D and NANOPOULOS A. Modeling users preference dynamics and side information in recommender systems[J]. IEEE Transactions on Systems, Man, & Cybernetics: Systems, 2016, 46(6): 782-792.
[19]
LIU Xin. Modeling users, dynamic preference for personalized recommendation[C]. Proceedings of the Twenty- Fourth International Joint Conference on Artificial Intelligence, Buenos Aires, Argentina, 2015: 1785-1791. doi: 10.1137/1.9781611972801.19.
[20]
SHANG Yanmin, XU Kefu, ZHANG Chuang, et al. FTM: Recommending the right items for user temporal interests with matrix factorization through topic model[C]. Proceedings of the IEEE First International Conference on Data Science in Cyberspace, Changsha, China, 2017: 189-198. doi: 10.1109/DSC.2016.20.
[21]
MNIH A and SALAKHUTDINOV R. Probabilistic matrix factorization[C]. Proceedings of the Twenty-First Annual Conference on Neural Information Processing Systems, Vancouver, Canada, 2007: 1257-1264.
