隨著購物網站的蓬勃發展，可提供使用者合適商品選項的推薦系統之技術也有大幅的進展。其中，基於商品資訊與使用者資訊進行推薦的內容導向推薦系統可針對各使用者進行個人化的推薦。然而，多數的推薦系統雖然可以進行精準的推薦，但是無法解釋推薦該商品的原因。
　　我們設計一個可發掘使用者喜好並提供推薦原因的推薦系統，該系統可根據商品描述與使用者的其他評論，預測使用者對商品的偏好以及提取使用者所關注的商品特色，推薦適合使用者的商品並指出符合使用者喜好的商品特色。本研究為推薦系統引入注意力機制處理文字資訊，使系統不只可以提供精準的推薦，還可以將商品資訊與使用者資訊之間的關聯以可理解的方式呈現出來。
　　藉由探討參數變化的實驗與評估系統推薦能力的實驗，我們發現引入注意力機制的推薦系統的確可以達到更好的推薦效果。透過視覺化商品描述與使用者評論之注意力圖譜的個案探討，我們理解到系統如何解析其中的關聯性，並透過解釋注意力圖譜的特徵來提供使用者可理解的推薦原因。
　　本研究改善過往推薦系統難以解釋推薦原因的限制，並透過實驗驗證了推薦系統結合注意力機制之架構的可行性。除了可進行精準的個人化推薦之外，也可以向使用者說明提供該選項的原因。希望透過系統提供的精準推薦與原因說明，可以提升使用者的購物體驗。

As online shopping flourishes, commercial recommendation systems provide suitable products for users. For personalized recommendation, some recommendation systems use product information and user profile. However, recommendation systems are able to do precision recommendation, but they cannot explain the reasons for recommendation.
We design a personalized recommendation system to discover user preference and provide recommendation reasons. It predicts user preference and captures product features from the product description and user profile. We apply attention mechanism to the recommendation system for modeling sentence pairs. With attention mechanism, the system provides not only precision recommendations but also visualized relationships between products and user profiles.
We found the recommendation system with attention mechanism performs well according to the experiments of model performance evaluation. We also understand how the recommendation system deals with the relationship between the product and the user by visualizing the attention map of the product-user pair in the case study, and we interpret features in the attention map to provide recommendation reasons for the user.
Our research improves the recommendation system for the personalized recommendation and recommendation reasons, and we verify the system by experiments and case studies. With the system, we hope it is able to improve the buying experience.

[1] M. J.Pazzani and D.Billsus, “Content-Based Recommendation Systems,” The adaptive web, pp. 325–341, 2007.
[2] G.Ganu, N.Elhadad, and A.Marian, “Beyond the Stars : Improving Rating Predictions using Review Text Content,” WebDB, no. 9, pp. 1–6, 2009.
[3] Y.Zhang, G.Lai, M.Zhang, Y.Zhang, Y.Liu, and S.Ma, “Explicit factor models for explainable recommendation based on phrase-level sentiment analysis,” Proceedings of the 37th international ACM SIGIR conference on Research & development in information retrieval, pp. 83–92, 2014.
[4] N.Hariri, B.Mobasher, R.Burke, and Y.Zheng, “Context-aware recommendation based on review mining,” CEUR Workshop Proceedings, vol. 756, pp. 30–36, 2010.
[5] J.McAuley and J.Leskovec, “Hidden Factors and Hidden Topics : Understanding Rating Dimensions with Review Text,” Proceedings of the 7th ACM conference on Recommender systems, pp. 165–172, 2013.
[6] G.Ling, M. R.Lyu, and I.King, “Ratings meet reviews, a combined approach to recommend,” Proceedings of the 8th ACM Conference on Recommender systems, pp. 105–112, 2014.
[7] A.Almahairi, K.Kastner, K.Cho, and A.Courville, “Learning Distributed Representations from Reviews for Collaborative Filtering,” Proceedings of the 9th ACM Conference on Recommender Systems, pp. 147–154, 2015.
[8] S.Hochreiter and J.Schmidhuber, “Long Short-Term Memory,” Neural Computation, vol. 9, no. 8, pp. 1735–1780, 1997.
[9] Y.Koren, P.Ave, F.Park, H. D.Management, and D.Applications, “Factorization meets the neighborhood: a multifaceted collaborative filtering model,” Proceedings of the 14th ACM SIGKDD international conference on Knowledge discovery and data mining, pp. 426–434, 2008.
[10] L.Zheng, V.Noroozi, and P. S.Yu, “Joint Deep Modeling of Users and Items Using Reviews for Recommendation,” Proceedings of the Tenth ACM International Conference on Web Search and Data Mining, pp. 425–433, 2017.
[11] Y.Kim, “Convolutional Neural Networks for Sentence Classification,” Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing, pp. 1746–1751, 2014.
[12] D.Tang, B.Qin, and T.Liu, “Learning Semantic Representations of Users and Products for Document Level Sentiment Classification,” Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing, vol. 1, pp. 1014–1023, 2015.
[13] J.Chen, H.Zhang, X.He, L.Nie, W.Liu, and T.-S.Chua, “Attentive Collaborative Filtering,” Proceedings of the 40th International ACM SIGIR Conference on Research and Development in Information Retrieval. pp. 335–344, 2017.
[14] C.Zhou et al., “ATRank: An Attention-Based User Behavior Modeling Framework for Recommendation,” arXiv, 2017.
[15] S.Seo, J.Huang, H.Yang, and Y.Liu, “Interpretable Convolutional Neural Networks with Dual Local and Global Attention for Review Rating Prediction,” Proceedings of the Eleventh ACM Conference on Recommender Systems, pp. 297–305, 2017.
[16] T.Mikolov, K.Chen, G.Corrado, and J.Dean, “Efficient Estimation of Word Representations in Vector Space,” arXiv, pp. 1–12, 2013.
[17] Q.V.Le and T.Mikolov, “Distributed Representations of Sentences and Documents,” International Conference on Machine Learning, vol. 32, 2014.
[18] W.Yin, H.Schütze, B.Xiang, and B.Zhou, “ABCNN: Attention-Based Convolutional Neural Network for Modeling Sentence Pairs,” arXiv, 2015.
[19] J.McAuley, C.Targett, Q.Shi, and A.van denHengel, “Image-based Recommendations on Styles and Substitutes,” Proceedings of the 38th International ACM SIGIR Conference on Research and Development in Information Retrieval, pp. 43–52, 2015.
[20] R.He and J.McAuley, “Ups and Downs: Modeling the Visual Evolution of Fashion Trends with One-Class Collaborative Filtering,” proceedings of the 25th international conference on world wide web, 2016.
[21] K.Bauman, B.Liu, and A.Tuzhilin, “Aspect Based Recommendations,” Proceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, pp. 717–725, 2017.