簡易檢索 / 詳目顯示
| 研究生： |
馬崇堯 Ma, Chung-Yao |
|---|---|
| 論文名稱： |
藉由堆疊式注意力網路實現適應性情境感知之推薦系統 Adaptive Context-Aware Recommendation System via Stacked Attention Network |
| 指導教授： |
蔣榮先
Chiang, Jung-Hsien |
| 學位類別： |
碩士 Master |
| 系所名稱： |
電機資訊學院 - 資訊工程學系 Department of Computer Science and Information Engineering |
| 論文出版年： | 2020 |
| 畢業學年度： | 108 |
| 語文別： | 英文 |
| 論文頁數： | 48 |
| 中文關鍵詞： | 推薦系統 、情境感知推薦系統 、神經網路推薦系統 、注意力神經網路 、深度學習 、卷積神經網絡 |
| 外文關鍵詞： | Recommendation System, Context-Aware Recommendation System, Neural Recommendation System, Attention Networks, Deep Learning, Convolutional Neural Networks |
| 相關次數： | 點閱：218 下載：2 |
| 分享至: |
| 查詢本校圖書館目錄 查詢臺灣博碩士論文知識加值系統 勘誤回報 |
推薦系統在現今社會中，隨著網路越來越普及而變得越來越重要。現在可以在網路上供使用者選擇的選項也越來越多，此時使用者們很難順利找到自己想要的東西。然而使用者的喜好與需求卻可能會隨時間或是周遭社群的影響而不停地改變，要如何掌握使用者在不同時間的需求及喜好便成為在推薦系統中一個很重要的課題。因此我們提出一個推薦系統來適應並學習使用者在不同時間點的需求及喜好。
情境感知推薦系統可以在推薦系統中加入許多的情境因子的資訊來幫助推薦系統更加了解使用者，因此我們提出一個情境感知推薦系統來學習使用者喜好並適應使用者在不同時間點的需求。並且在所有的情境因子中，時間與使用者的喜好變化最為相關且最容易取得，我們也因此基於時間來建構我們的情境。此外，深度注意力機制不考慮使用者與商品間的互動順序且可以學習商品的重要度等特性很適合來抓取使用者變動的喜好，所以我們使用商品層面以及使用者層面兩種不同的深度注意力機制來學習並適應使用者變動的需求與喜好。並且商品與使用者間互動的順序資訊可助於我們抓取並判別使用者的喜好變化，因此我們額外提供兩種抓取商品間時間順序資訊的方式來加強我們推薦系統適應使用者變動的需求與喜好的能力。
在本研究中，我們採用了三個收集自現實世界的資料集來進行推薦結果的評估實驗並與其他先前研究方法來做比較。實驗數據顯示我們提出的推薦模型可以比先前的方法在推薦的精確度以及推薦序上還要有更好的表現，並且以個案探討來證明以注意力機制學習時間相關的情境因子可以更加適應使用者的喜好變化。此外，加入兩種不同商品與使用者間互動的時間資訊的方式，皆可以對實驗結果帶來顯著的增長。最後，希望我們的研究未來可以應用在現實生活的推薦系統中，並期許可以帶給人們更多的幫助和更適合的推薦結果。
Recommendation systems become more and more important since the world become more and more convenient with the widespread internet. With enormous amounts of items, it is difficult for users to find out what they really need. However, users’ preferences will change over time because of the age and the impacts of social networks. In a recommendation system, how to capture users’ preferences and how we recommend suitable items to users is a big issue and difficult challenge. Therefore, we proposed a novel recommendation system to adapt to users’ changing preferences. By learning and adapting to users’ changing preferences, we can know user better and recommend more precisely.
In this study, we applied context factors in our recommendation system since context-aware recommendation systems consider features as additional information to model users’ preferences more precisely. Within the context factors, time becomes more important and can capture the changing patterns in users’ preferences. And we applied the context which based on users’ interacted items in each defined time interval in our recommendation system. Moreover, the deep attention mechanism is an appropriate technique to capture the changing patterns in users’ preferences that learned the weight between items. In addition, the deep attention mechanism learned the importance between items without considering the order information of the interacted record. Therefore, we applied two different attention mechanisms, user-item attention and user multi-head attention, in our recommendation system to adapt to users’ dynamic changing preferences. In addition, we enhanced the performances with two additional item’s temporal information to model the contextual item representation precisely.
In our experiment, we conducted three real-world datasets and evaluated the performances compared to state-of-the-art recommendation method. The experimental results demonstrated that our proposed context-aware recommendation model outperformed the traditional methods and showed the effectiveness of contextual information and attention mechanism. And the case studies demonstrated that our recommendation system can adapt to users’ changing preferences better than the traditional methods. Moreover, the performances also increased with two additional item’s temporal information. Finally, we hope that our context-aware recommendation system with two attention mechanisms can be applied to real-world recommendation scenarios.
Abbasi, R., Grzegorzek, M., &Staab, S. "Large scale tag recommendation using different image representations". Lecture Notes in Computer Science (Including Subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 5887 LNCS(June 2014), 65–76, 2009.
Adomavicius, G., &Tuzhilin, A. "Incorporating Contextual Information in Recommender Systems Using a Multidimensional Approach University of Minnesota University of Connecticut". ACM Transactions on Information Systems, 23(1), 103–145, 2005.
Adrian, B., Sauermann, L., &Roth-berghofer, T. "ConTag : A Semantic Tag Recommendation System". Artificial Intelligence, 7(October), 297–304, 2007.
Al, R. E. T., Rendle, S., Freudenthaler, C., Gantner, Z., &Schmidt-thieme, L. "BPR:Bayesian Personalized Ranking from Implicit Feedback". 452–461, 2009.
Bahdanau, D., Cho, K., &Bengio, Y. "Neural Machine Translation by Jointly Learning to Align and Translate". ArXiv, 1–15, 2014.
Chorowski, J. "Attention-Based Models for Speech Recognition". 1–9, 2014.
Chung, J., Gulcehre, C., Cho, K., &Bengio, Y. "Empirical Evaluation of Gated Recurrent Neural Networks on Sequence Modeling". ArXiv, 1–9, 2014.
DaCosta, F. S., &Dolog, P. "Collective embedding for neural context-aware recommender systems". RecSys 2019 - 13th ACM Conference on Recommender Systems, 201–209, 2019.
Dong, X., Yu, L., Wu, Z., Sun, Y., Yuan, L., &Zhang, F. "A hybrid collaborative filtering model with deep structure for recommender systems". 31st AAAI Conference on Artificial Intelligence, AAAI 2017, 1309–1315, 2017.
Dourish, P. "What we talk about when we talk about context". Personal and Ubiquitous Computing, 8(1), 19–30, 2004.
Gong, Y., &Zhang, Q. "Hashtag Recommendation Using Attention-Based Convolutional Neural Network". 2782–2788, 2013.
Guo, H., Tang, R., Ye, Y., Li, Z., &He, X. "DeepFM: A factorization-machine based neural network for CTR prediction". IJCAI International Joint Conference on Artificial Intelligence, 1725–1731, 2017.
He, X., He, Z., Song, J., Liu, Z., Jiang, Y., &Chua, T. "NAIS : Neural Attentive Item Similarity Model for Recommendation". 30(12), 2354–2366, 2018.
He, X., Liao, L., Zhang, H., Nie, L., Hu, X., &Chua, T. S. "Neural collaborative filtering". 26th International World Wide Web Conference, WWW 2017, 173–182, 2017.
Hidasi, B., Karatzoglou, A., Baltrunas, L., &Tikk, D. "Session-based Recommendations with Recurrent Neural Networks". Proceedings of the 4th. International Conference on Learning Representations, 425–434, 2016.
Hidasi, B., Quadrana, M., Milano, P., &Karatzoglou, A. "Parallel Recurrent Neural Network Architectures for Feature-rich Session-based Recommendations". 241–248, 2016.
Jannach, Di., &Ludewig, M. "When recurrent neural networks meet the neighborhood for session-based recommendation". RecSys 2017 - Proceedings of the 11th ACM Conference on Recommender Systems, 306–310, 2017.
Kabbur, S., Ning, X., &Karypis, G. "FISM: Factored item similarity models for Top-N recommender systems". Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, Part F1288(August), 659–667, 2013.
Kingma, D. P., &Ba, J. L. "Adam: A method for stochastic optimization". 3rd International Conference on Learning Representations, ICLR 2015 - Conference Track Proceedings, 1–15, 2015.
Koren, Y., Bell, R., &Volinsky, C. "Matrix Factorization Techniques for Recommender Systems". Computer, 42(8), 30–37, 2009.
Li, J., Li, L., Wu, Y., &Chen, S. "An Improved Recommender Based on Hidden Markov Model". In P.Anthony, M.Ishizuka, &D.Lukose (Eds.), PRICAI 2012: Trends in Artificial Intelligence (pp. 870–874), 2012.
Qiu, J., Tang, J., Ma, H., Dong, Y., Wang, K., &Tang, J. "DeepInf: Social influence prediction with deep learning". Proceedings of the ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 2110–2119, 2018.
Rendle, S. "Factorization machines". Proceedings - IEEE International Conference on Data Mining, ICDM, 995–1000, 2010.
Salakhutdinov, R., Mnih, A., &Hinton, G. "Restricted Boltzmann Machines for Collaborative Filtering". Proceedings of the 24th ACM International Conference on Machine Learning (ICML), 791–798, 2007.
Smirnova, E., &Vasile, F. "Contextual sequence modeling for recommendation with Recurrent Neural Networks". ACM International Conference Proceeding Series, Part F1301, 2–9, 2017.
Tan, Y. K. "Improved Recurrent Neural Networks for Session-based Recommendations". 0–5, 2016.
Tu, C., Liu, H., Liu, Z., &Sun, M. "CANE: Context-aware network embedding for relation modeling". ACL 2017 - 55th Annual Meeting of the Association for Computational Linguistics, Proceedings of the Conference (Long Papers), 1, 1722–1731, 2017.
Vaswani, A. "Attention Is All You Need". (Nips), 2017.
Veličković, P., Cucurull, G., Casanova, A., Romero, A., Liò, P., &Bengio, Y. "Graph Attention Networks". 1–12, 2017.
Vinyals, O., &Toshev, A. "Show and Tell : A Neural Image Caption Generator".
Wang, J., &Caverlee, J. "Recurrent recommendation with local coherence". WSDM 2019 - Proceedings of the 12th ACM International Conference on Web Search and Data Mining, 564–572, 2019.
Wang, S., Hu, L., Cao, L., Huang, X., Lian, D., &Liu, W. "Attention-based transactional context embedding for next-item recommendation". 32nd AAAI Conference on Artificial Intelligence, AAAI 2018, 2532–2539, 2018.
Ying, H., Zhuang, F., Zhang, F., Liu, Y., Xu, G., Xie, X., …Wu, J. "Sequential Recommender System based on Hierarchical Attention Network". 3926–3932, 2016.
Zhang, S., Tay, Y., Yao, L., &Sun, A. "Next Item Recommendation with Self-Attention", 2018.
Zhang, Y., Dai, H., Xu, C., Feng, J., Wang, T., Bian, J., …Liu, T. Y. "Sequential click prediction for sponsored search with recurrent neural networks". Proceedings of the National Conference on Artificial Intelligence, 2, 1369–1375, 2014.
Zheng, L., Noroozi, V., &Yu, P. S. "Joint Deep Modeling of Users and Items Using Reviews for Recommendation". Proceedings of the Tenth ACM International Conference on Web Search and Data Mining, 425–433, 2017.
校內：2022-09-01公開校外：2022-09-01公開