嶄新的即時POI推薦系統－使用即時事件、圖文／時間內容感知資訊與樹狀卷積網路

Real-Time POI Recommendation Based on Event Embedding, Textual & Visual/Time-Aware Information and Tree Structured CNN

郝沛毅(Pei-Yi Hao)；龔千芬(Chien-Feng Kung)；張俊陽(Chun-Yang Chang)；蔣榮先(Jung-Hsien Chiang)；鄭詠恆(Weng-Hang Cheang)

事件嵌入 ； 興趣點推薦 ； 矩陣分解 ； 深度學習 ； 卷積神經網路 ； event embedding ； POI recommendation ； matrix factorization ； deep learning ； convolutional neural networks

資訊管理學報

27卷4期（2020 / 10 / 31）

495 - 535

繁體中文

基於位置的社群網路（LBSN）近來變得十分流行，這歸功於智慧手機的爆炸式增長，使得用戶可以輕鬆地執行LBSN程序。越來越多使用者在這些平台上與好友分享打卡資訊跟生活點滴。興趣點（POI）推薦系統是LBSN的核心服務，也是最近熱門的研究焦點。目前研究主要是分析用戶的打卡序列，來探勘使用者的偏好，可是，這些方法沒有考量到時間與事件這兩項關鍵因素，我們認為這兩項因素會影響使用者拜訪興趣點的意願。例如，一個平常不運動的使用者，會因為一個演唱會的契機而在體育館打卡，如果系統只考慮使用者對於興趣點的固有偏好，則會忽略了這種情境。本研究的目的是建立一個結合即時事件偵測的興趣點推薦系統。我們的主要貢獻是考慮正在進行的事件、合適的時機與POI的特性，以上述三個基礎來推薦合適的POI。我們的方法可以從大量的具地理標記的推文中，偵測出即時事件，並且透過本研究研發的樹狀卷積神經網路，來學習即時事件與時間感知資訊的嵌入特徵表示。此外，我們的方法可以從標記在興趣點的文字評論與拍攝照片中，捕抓興趣點的圖文內容感知特徵，並且以卷積神經網路來學習興趣點的圖文嵌入特徵向量。最終，這些POI的即時嵌入特徵將融合到矩陣分解式的協同過濾推薦演算法，以建構即時的POI推薦系統。

Purpose - Location-based social networks (LBSN) have recently become popular due to the explosive growth of smartphones, giving users easy access to LBSN applications. More and more users have shared check-in information and daily life with friends on these platforms. The point of interest (POI) recommendation system is one of the core services of the LBSN. Design/methodology/approach - This study proposes a novel real-time POI recommendation system. The proposed approach is capable to detect real-time event from the huge amount of geo-tagged tweets in LBSN, and learn the embedding representation of the real-time event and time-aware information of a given POI. Besides, the proposed approach captures the content characteristic of POI from the text and phots tagged at the corresponding POI, and learn the embedding representation of the textual and visual characteristic of a given POI. Finally, we incorporate the real-time POI embedding into the matrix factorization model to build the real-time POI recommendation system. Findings - Firstly, multimodal embedding with considering various types (spatial, temporal and textual) of features performs well in learning semantic meaning on keywords. Thus it is effective for us to build the event detection from messy short-texts on social media. Besides, the proposed model considers time-aware information of the POIs by extracting ongoing events and recommend for suitable users in different time period. Evaluation on the dataset with geo-tagged tweets in NYC demonstrates the effectiveness of our study. Research limitations/implications - Our POI Intrinsic Embedding metric is trained by textual information. It only considers textual features of POIs. However, other types of information such as geographical features like latitude and longitude, popularity features like the flow at different time period can be applied in the embedding metric in the future work. Practical implications - The POI recommendation applications are of significance in two aspects: helping users explore new interesting places in a city and facilitating business owners to launch advertisements to the target customers. Originality/value - This study is, to the best of our knowledge, the first attempt to apply tree-based CNN for the real time event detection and POI recommendation in Taiwan.

1. Bengio, Y.,Courville, A.,Vincent, P.(2013).Representation learning: A review and new perspectives.IEEE Transactions on Pattern Analysis and Machine Intelligence,35(8),1798-1828.
2. Bokde, D.,Girase, S.,Mukhopadhyay, D.(2015).Role of matrix factorization model in collaborative filtering algorithm: A survey.Procedia Computer Science,49,136-146.
3. Cho, E.,Myers, S.A.,Leskovec, J.(2011).Friendship and mobility: User movement in location-based social networks.Proceedings of the 17th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD’11),San Diego, CA, USA:
4. Gao, H.,Tang, J.,Liu, H.(2015).Addressing the cold-start problem in location recommendation using geo-social correlations.Data Mining Knowledge Discovery,29(2),299-323.
5. Hao, P.-Y.,Cheang, W.-H.,Chiang, J.-H.(2019).Real-time event embedding for POI recommendation.Neurocomputing,349(15),1-11.
6. Hu, B.,Lu, Z.,Li, H.,Chen, Q.(2014).Convolutional neural network architectures for matching natural language sentences.Proceedings of the Advances in Neural Information Processing Systems (NIPS 2014),Montreal, QC, Canada:
7. Iyyer, M.,Manjunatha, V.,Boyd-Graber, J.,Daume, H.(2015).Deep unordered composition rivals syntactic methods for text classification.Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7th International Joint Conference on Natural Language Processing,Beijing, China:
8. Kim, D.,Park, C.,Oh, J.,Lee, S.,Yu, H.(2016).Convolutional matrix factorization for document context-aware recommendation.Proceedings of the 10th ACM Conference on Recommender systems (RecSys ’16),Boston, MA, USA:
9. Kim, Y.(2014).Convolutional neural networks for sentence classification.Proceedings of the 2014 Conference on Empirical Methods in Natural Language Processing (EMNLP),Doha, Qata:
10. Liu, B.,Xiong, H.(2013).Point-of-interest recommendation in location based social networks with topic and location awareness.Proceedings of the 13th SIAM International Conference on Data Mining (SDM 2013),Austin, TX, USA:
11. Liu, W.,Lai, H.,Wang, J.,Ke, G.,Yang, W.,Yin, J.(2019).Mix geographical information into local collaborative ranking for POI recommendation.World Wide Web,23
12. Mikolov, T.,Sutskever, I.,Chen, K.,Corrado, G.,Dean, J.(2013).Distributed representations of words and phrases and their compositionality.Proceedings of the Advances in Neural Information Processing Systems (NIPS 2013),Lake Tahoe, NV, USA:
13. Nguyen, T.H.,Grishman, R.(2015).Relation extraction: Perspective from convolutional neural networks.Proceedings of the Conference of the North American Chapter of the Association for Computational Linguistics-Human Language Technologies (NAACL-HLT 2015),Denver, CO:
14. Pan, R.,Zhou, Y.,Cao, B.,Niu, N.N.,Lukose, R.,Scholz, M.,Yang, Q.(2008).One-class collaborative filtering.Proceedings of the European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases (ECML PKDD 2008),Antwerp, Belgium:
15. Rios, C.,Schiaffino, S.,Godoy, D.(2018).A study of neighbour selection strategies for POI recommendation in LBSNs.Journal of Information Science,44(6),802-817.
16. Shirani-mehr, H.(2015).,CA, USA:Stanford University.
17. Simonyan, K.,Zisserman, A.(2015).Very deep convolutional networks for large-scale image recognition.Proceedings of the 3rd International Conference on Learning Representations (ICLR 2015),San Diego, CA, USA:
18. Taigman, Y.,Yang, M.,Ranzayo, M.A.,Wolf, L.(2014).Deepface: Closing the gap to human-level performance in face verification.Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition (CVPR ’14),Columbus, OH, USA:
19. Wang, F.,Qu, Y.,Zheng, L.,Lu, C.T.,Yu, P.S.(2017).Deep and broad learning on content-aware POI recommendation.Proceedings of the3rd IEEE International Conference on Collaboration and Internet Computing (CIC2017),San Jose, CA, USA:
20. Wang, S.,Wang, Y.,Tang, J.,Shu, K.,Ranganath, S.,Liu, H.(2017).What your images reveal: Exploiting visual contents for point-of-interest recommendation.Proceedings of the 26th International Conference on World Wide Web (WWW’17),Perth, Australia:
21. Wang, X.,Zhao, Y.L.,Nie, L.,Gao, Y.,Nie, W.,Zha, Z.-H.(2015).Semantic-based location recommendation with multimodal venue semantics.IEEE Transactions on Multimedia,17(3),409-419.
22. Wen, Y.,Zhang, W.,Luo, R.,Wang, J.(2016).Learning text representation using recurrent convolutional neural network with highway layers.Proceesings of the Neu-IR ’16 SIGIR Workshop on Neural Information Retrieval,Pisa, Ital:
23. Yan, D.,Zhao, X.,Guo, Z.(2018).Personalized POI recommendation based on subway network features and users’ historical behaviors.Hindawi Wireless Communications and Mobile Computing,2018,Article ID 3698198.
24. Yang, D.,Zhang, D.,Yu, Z.,Wang, Z.(2013).Asentiment-enhanced personalized location recommendation system.Proceedings of the 24th ACM Conference on Hypertext and Social Media,Paris, France:
25. Zhang, C.,Liu, L.,Lei, D.,Yuan, Q.,Zhuang, H.,Hanratty, T.,Han, J.(2017).TrioVecEvent: Embedding-based online local event detection in geo-tagged tweet streams.Proceedings of theProceedings of the 23rd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (KDD’17),Halifax, NS, Canada:
26. Zhang, C.,Zhou, G.,Yuan, Q.,Zhuang, H.,Zheng, Y.,Kaplan, L.,Wang, S.,Han, J.(2016).GeoBurst: Real-time local event detection in geo-tagged tweet streams.Proceedings of the 39th ACM International Conference on Research and Development in Information Retrieval (SIGIR ’16),Pisa, Italy:
27. Zhang, J.,Kong, X.,Yu, P.S.(2014).Transferring heterogeneous links across location-based social networks.Proceedings of 7th ACM International Conference on Web Search and Data Mining (WSDM ’14),New York, NY, USA:
28. Zhang, J.D.,Chow, C.Y.(2013).iGSLR: Personalized geo-social location recommendation: A kernel density estimation approach.Proceedings of the 21st ACM International Conference on Advances in Geographic Information Systems (SIGSPATIAL ’13),Orlando, FL, USA:
29. Zhang, J.D.,Chow, C.Y.,Li, Y.(2014).LORE: Exploiting sequential influence for location recommendations.Proceedings of the 22nd ACM International Conference on Advances in Geographic Information Systems (SIGSPATIAL’14),Dallas/Fort Worth, TX, USA:
30. Zhang, Z.,Zou, C.,Ding, R.,Chen, Z.(2019).VCG: Exploiting visual contents and geographical influence for point-of-interest recommendation.Neurocomputing,357(10),53-65.
31. Zhao, S.,Zhao, T.,King, I.,Lyu, M.R.(2016).,未出版
32. Zhao, S.,Zhao, T.,Yang, H.,Lyu, M.R.,King, I.(2016).STELLAR: Spatial-temporal latent ranking for successive point-of-interest recommendation.Proceedings of the Thirtieth AAAI Conference on Artificial Intelligence(AAAI-16),Phoenix, AZ, USA:
33. Zhu, J.,Wang, C.,Guo, X.,Ming, Q.,Li, J.,Liu, Y.(2019).Friend and POI recommendation based on social trust cluster in location-based social networks.Journal on Wireless Communications

1. (2024)。運用深度學習與主題模型建構歌曲風格和歌詞意涵之整合分析機制。資訊管理學報，31(2)，209-237。