推薦系統在各種實際應用中扮演著重要角色，為使用者提供個性化建議。然而，無論是傳統的推薦系統還是序列推薦，冷啟動問題仍然是一個挑戰。當新使用者或新品加入系統時，由於缺乏先前的互動，精準的推薦受到阻礙。為了解決這個問題，許多研究都探討了meta-leaning的方法解決。其中一個值得注意的框架，metaDyGNN，利用MAML和Dynamic graph 來解決冷啟動問題。然而，對於新用戶與項目互動的準確時間預測仍然是一個挑戰。HINTS模型是用於預測學術論文引用的模型，可以提供預測新論文未來每年引用的數量。在本研究中，我們旨在通過整合metaDyGNN和HINTS的優勢，預測新用戶與項目之間的未來互動時間。我們提出了一個流程，利用metaDyGNN 作為編碼器，HINTS作為解碼器。我們在用於序列推薦和連結預測的基準資料集上評估我們的模型，展示了其在推薦準確性和克服冷啟動問題方面的優越表現。我們還突出了它利用動態圖建模來預測下一個時間間隔的能力。通過全面評估，我們的模型顯示了在處理序列推薦和連結預測方面的效能，為推薦系統的發展做出了貢獻。

Recommendation systems play a crucial role in various real-world scenarios, providing personalized suggestions to users. However, the cold-start problem remains a challenge for both traditional recommender systems and sequential recommendations. When new users or items join the system, accurate recommendations are hindered by the lack of prior interactions. To address this problem, meta-learning techniques have been explored to conquer it. One notable framework, metaDyGNN, leverages Model-Agnostic Meta-Learning and dynamic graph structures to mitigate the cold-start problem. However, accurately predicting the timing of interactions for new users remains a challenge. The HINTS model, developed for predicting academic paper citations, offers insights into timing prediction. In this paper, we aim to optimize the timing of new user-item interactions by integrating the strengths of metaDyGNN and HINTS. We propose a heuristic pipeline that utilizes metaDyGNN as an encoder and HINTS as a decoder. We evaluate our model on benchmark datasets for sequential recommendations and link prediction, demonstrating its superior performance in recommendation accuracy and overcoming the cold-start problem. We also highlight its ability to predict the next intervals using dynamic graph modeling. Through comprehensive evaluations, our proposed model showcases its efficacy in handling sequential recommendations and link prediction, contributing to the advancement of recommendation systems.

論文審定書.............................................................................................. i
誌謝......................................................................................................... ii
摘要........................................................................................................ iii
Abstract.................................................................................................. iv
Table of contents................................................................................... v
Table of figures................................................................................... vii
Table of Tables...................................................................................viii
1. INTRODUCTION............................................................................ 1
2. RELATED WORK............................................................................ 6
2.1 Graph model (Graph Neural Network).......................................... 6
2.2 Graph-based Sequence Recommendation Model ............................. 9
2.3 Meta-Learning Approaches for the Cold-Start Issue........................... 10
3. METHODOLOGY.............................................................................12
3.1 Problem Definition..................................................................12
3.1.1 The first-time prediction................................................ 12
3.1.2 Continuous dynamic graph............................................. 13
3.2 Model Architecture..................................................................16
3.2.1 Node encoder..............................................................17
3.2.2 Interval decoder...........................................................19
3.2.3 Loss computation........................................................ 20
3.2.4 Meta-learning..............................................................22
4. EXPERIMENT.................................................................................27
4.1 Experiment Setup.................................................................. 28
4.1.1 Dataset.................................................................... 28
4.1.2 Baselines..................................................................29
4.1.3 Evaluation Settings......................................................30
4.1.4 Hyperparameter Settings...............................................31
4.2 Result.................................................................................32
4.2.1 Next item predictions...................................................32
4.2.2 Next interval predictions............................................... 34
5. DISCUSSION AND LIMITATIONS......................................................36
5.1 Discussion.......................................................................... 36
5.1.1 Data related discussion.................................................36
5.1.2 Model related discussion...............................................36
5.2 Limitations..........................................................................38
5.2.1 The Distribution of the Decoder......................................38
5.2.2 The Number of Layers and Neighbors..............................39
5.3 Applications........................................................................39
6. CONCLUSION AND FUTURE WORK.................................................40
7. REFERENCE................................................................................ 42

Alareeni, B. A., & Hamdan, A. (2020). ESG impact on performance of US S&P 500-listed firms. Corporate Governance: The International Journal of Business in Society, 20(7), 1409–1428. https://doi.org/10.1108/CG-06-2020-0258
Antoniou, A., Edwards, H., & Storkey, A. (2018). How to train your MAML. https://doi.org/10.48550/ARXIV.1810.09502
Atwood, J., & Towsley, D. (n.d.). Diffusion-Convolutional Neural Networks. 9.
Bharadhwaj, H. (2019). Meta-Learning for User Cold-Start Recommendation. 2019 International Joint Conference on Neural Networks (IJCNN), 1–8. https://doi.org/10.1109/IJCNN.2019.8852100
Cai, D., Qian, S., Fang, Q., Hu, J., & Xu, C. (2022). User Cold-start Recommendation via Inductive Heterogeneous Graph Neural Network. ACM Transactions on Information Systems, 3560487. https://doi.org/10.1145/3560487
Chang, X., Liu, X., Wen, J., Li, S., Fang, Y., Song, L., & Qi, Y. (2020). Continuous-Time Dynamic Graph Learning via Neural Interaction Processes. Proceedings of the 29th ACM International Conference on Information & Knowledge Management, 145–154. https://doi.org/10.1145/3340531.3411946
Chen, J., Wang, X., & Xu, X. (2021). GC-LSTM: Graph Convolution Embedded LSTM for Dynamic Link Prediction (arXiv:1812.04206). arXiv. http://arxiv.org/abs/1812.04206
Chen, Z., Zhang, W., Yan, J., Wang, G., & Wang, J. (2021). Learning Dual Dynamic Representations on Time-Sliced User-Item Interaction Graphs for Sequential Recommendation (arXiv:2109.11790). arXiv. http://arxiv.org/abs/2109.11790
Defferrard, M., Bresson, X., & Vandergheynst, P. (n.d.). Convolutional Neural Networks on Graphs with Fast Localized Spectral Filtering. 9.
Du, Z., Wang, X., Yang, H., Zhou, J., & Tang, J. (2019). Sequential Scenario-Specific Meta Learner for Online Recommendation (arXiv:1906.00391). arXiv. http://arxiv.org/abs/1906.00391
Fan, Z., Liu, Z., Zhang, J., Xiong, Y., Zheng, L., & Yu, P. S. (2021). Continuous-Time Sequential Recommendation with Temporal Graph Collaborative Transformer. Proceedings of the 30th ACM International Conference on Information & Knowledge Management, 433–442. https://doi.org/10.1145/3459637.3482242
Feng, X., Chen, C., Li, D., Zhao, M., Hao, J., & Wang, J. (2021). CMML: Contextual Modulation Meta Learning for Cold-Start Recommendation (arXiv:2108.10511). arXiv. http://arxiv.org/abs/2108.10511
Feng, Y., You, H., Zhang, Z., Ji, R., & Gao, Y. (2019). Hypergraph Neural Networks (arXiv:1809.09401). arXiv. http://arxiv.org/abs/1809.09401
Finn, C., Abbeel, P., & Levine, S. (n.d.). Model-Agnostic Meta-Learning for Fast Adaptation of Deep Networks. 10.
Goyal, P., Kamra, N., He, X., & Liu, Y. (2018). DynGEM: Deep Embedding Method for
Dynamic Graphs (arXiv:1805.11273). arXiv. http://arxiv.org/abs/1805.11273
Hamilton, W., Ying, Z., & Leskovec, J. (2017). Inductive Representation Learning on Large Graphs. 11.
He, X., Deng, K., Wang, X., Li, Y., Zhang, Y., & Wang, M. (2020). LightGCN: Simplifying and Powering Graph Convolution Network for Recommendation. Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval, 639–648. https://doi.org/10.1145/3397271.3401063
He, X., Liao, L., Zhang, H., Nie, L., Hu, X., & Chua, T.-S. (2017). Neural Collaborative Filtering. Proceedings of the 26th International Conference on World Wide Web, 173–182. https://doi.org/10.1145/3038912.3052569
Hidasi, B., & Karatzoglou, A. (2018). Recurrent Neural Networks with Top-k Gains for Session-based Recommendations. Proceedings of the 27th ACM International Conference on Information and Knowledge Management, 843–852. https://doi.org/10.1145/3269206.3271761
Hidasi, B., Karatzoglou, A., Baltrunas, L., & Tikk, D. (2015). Session-based Recommendations with Recurrent Neural Networks. https://doi.org/10.48550/ARXIV.1511.06939
Holme, P. (2015). Modern temporal network theory: A colloquium. The European Physical Journal B, 88(9), 234. https://doi.org/10.1140/epjb/e2015-60657-4
Hsu, C., & Li, C.-T. (2021). RetaGNN: Relational Temporal Attentive Graph Neural Networks for Holistic Sequential Recommendation. Proceedings of the Web Conference 2021, 2968–2979. https://doi.org/10.1145/3442381.3449957
Huang, X., Sang, J., Yu, J., & Xu, C. (2022). Learning to Learn a Cold-start Sequential Recommender. ACM Transactions on Information Systems, 40(2), 1–25. https://doi.org/10.1145/3466753
Jiang, S., Koch, B., & Sun, Y. (2021). HINTS: Citation Time Series Prediction for New Publications via Dynamic Heterogeneous Information Network Embedding. Proceedings of the Web Conference 2021, 3158–3167. https://doi.org/10.1145/3442381.3450107
Kang, W.-C., & McAuley, J. (2018). Self-Attentive Sequential Recommendation. https://doi.org/10.48550/ARXIV.1808.09781
Kazemi, S. M., Goel, R., Eghbali, S., Ramanan, J., Sahota, J., Thakur, S., Wu, S., Smyth, C., Poupart, P., & Brubaker, M. (2019). Time2Vec: Learning a Vector Representation of Time. https://doi.org/10.48550/ARXIV.1907.05321
Kipf, T. N., & Welling, M. (2016). Semi-Supervised Classification with Graph Convolutional Networks. https://doi.org/10.48550/ARXIV.1609.02907
Koren, Y., Bell, R., & Volinsky, C. (2009). Matrix Factorization Techniques for Recommender Systems. Computer, 42(8), 30–37. https://doi.org/10.1109/MC.2009.263
Lee, H., Im, J., Jang, S., Cho, H., & Chung, S. (2019). MeLU: Meta-Learned User Preference Estimator for Cold-Start Recommendation. Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, 1073–1082. https://doi.org/10.1145/3292500.3330859
Levie, R., Monti, F., Bresson, X., & Bronstein, M. M. (2019). CayleyNets: Graph Convolutional Neural Networks With Complex Rational Spectral Filters. IEEE Transactions on Signal Processing, 67(1), 97–109. https://doi.org/10.1109/TSP.2018.2879624
Li, J., Wang, Y., & McAuley, J. (2020). Time Interval Aware Self-Attention for Sequential Recommendation. Proceedings of the 13th International Conference on Web Search and Data Mining, 322–330. https://doi.org/10.1145/3336191.3371786
Li, X., Zhang, M., Wu, S., Liu, Z., Wang, L., & Yu, P. S. (2021). Dynamic Graph Collaborative Filtering (arXiv:2101.02844). arXiv. http://arxiv.org/abs/2101.02844
Liu, F., Cheng, Z., Zhu, L., Gao, Z., & Nie, L. (2021). Interest-aware Message-Passing GCN for Recommendation. Proceedings of the Web Conference 2021, 1296–1305. https://doi.org/10.1145/3442381.3449986
Liu, S., Ounis, I., Macdonald, C., & Meng, Z. (2020). A Heterogeneous Graph Neural Model for Cold-start Recommendation. Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval, 2029–2032. https://doi.org/10.1145/3397271.3401252
Lu, Y., Fang, Y., & Shi, C. (2020). Meta-learning on Heterogeneous Information Networks for Cold-start Recommendation. Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, 1563–1573. https://doi.org/10.1145/3394486.3403207
Ma, C., Ma, L., Zhang, Y., Sun, J., Liu, X., & Coates, M. (2020). Memory Augmented Graph Neural Networks for Sequential Recommendation. Proceedings of the AAAI Conference on Artificial Intelligence, 34(04), 5045–5052. https://doi.org/10.1609/aaai.v34i04.5945
Ma, Y., Guo, Z., Ren, Z., Tang, J., & Yin, D. (2020). Streaming Graph Neural Networks. Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval, 719–728. https://doi.org/10.1145/3397271.3401092
Micheli, A. (2009). Neural Network for Graphs: A Contextual Constructive Approach. IEEE Transactions on Neural Networks, 20(3), 498–511. https://doi.org/10.1109/TNN.2008.2010350
Neupane, K. P., Zheng, E., Kong, Y., & Yu, Q. (2022). A Dynamic Meta-Learning Model for Time-Sensitive Cold-Start Recommendations. Proceedings of the AAAI Conference on Artificial Intelligence, 36(7), 7868–7876. https://doi.org/10.1609/aaai.v36i7.20756
Pareja, A., Domeniconi, G., Chen, J., Ma, T., Suzumura, T., Kanezashi, H., Kaler, T., Schardl, T., & Leiserson, C. (2020). EvolveGCN: Evolving Graph Convolutional Networks for Dynamic Graphs. Proceedings of the AAAI Conference on Artificial Intelligence, 34(04), 5363–5370. https://doi.org/10.1609/aaai.v34i04.5984
Quadrana, M., Karatzoglou, A., Hidasi, B., & Cremonesi, P. (2017). Personalizing Session-based Recommendations with Hierarchical Recurrent Neural Networks. Proceedings of the Eleventh ACM Conference on Recommender Systems, 130–137. https://doi.org/10.1145/3109859.3109896
Rippel, O., Snoek, J., & Adams, R. P. (n.d.). Spectral Representations for Convolutional Neural Networks. 9.
Sankar, A., Wu, Y., Gou, L., Zhang, W., & Yang, H. (2020). DySAT: Deep Neural Representation Learning on Dynamic Graphs via Self-Attention Networks. Proceedings of the 13th International Conference on Web Search and Data Mining, 519–527. https://doi.org/10.1145/3336191.3371845
Seo, Y., Defferrard, M., Vandergheynst, P., & Bresson, X. (2018). Structured Sequence Modeling with Graph Convolutional Recurrent Networks. In L. Cheng, A. C. S. Leung, & S. Ozawa (Eds.), Neural Information Processing (Vol. 11301, pp. 362–373). Springer International Publishing. https://doi.org/10.1007/978-3-030-04167-0_33
Shi, C., Han, X., Song, L., Wang, X., Wang, S., Du, J., & Yu, P. S. (2021). Deep Collaborative Filtering with Multi-Aspect Information in Heterogeneous Networks. IEEE Transactions on Knowledge and Data Engineering, 33(4), 1413–1425. https://doi.org/10.1109/TKDE.2019.2941938
Shi, C., Hu, B., Zhao, W. X., & Yu, P. S. (2019). Heterogeneous Information Network Embedding for Recommendation. IEEE Transactions on Knowledge and Data Engineering, 31(2), 357–370. https://doi.org/10.1109/TKDE.2018.2833443
Tan, Y. K., Xu, X., & Liu, Y. (2016). Improved Recurrent Neural Networks for Session-based Recommendations. Proceedings of the 1st Workshop on Deep Learning for Recommender Systems, 17–22. https://doi.org/10.1145/2988450.2988452
Trivedi, R., Farajtabar, M., Biswal, P., & Zha, H. (2019). DYREP: LEARNING REPRESENTATIONS OVER DYNAMIC GRAPHS. 25.
Velicˇkovic, P., Cucurull, G., Casanova, A., Romero, A., Lio, P., & Bengio, Y. (2018). GRAPH ATTENTION NETWORKS. 12.
Wang, J., Ding, K., & Caverlee, J. (2021). Sequential Recommendation for Cold-start Users with Meta Transitional Learning. Proceedings of the 44th International ACM SIGIR Conference on Research and Development in Information Retrieval, 1783–1787. https://doi.org/10.1145/3404835.3463089
Wang, J., Ding, K., Hong, L., Liu, H., & Caverlee, J. (2020). Next-item Recommendation with Sequential Hypergraphs. Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval, 1101–1110. https://doi.org/10.1145/3397271.3401133
Wang, X., He, X., Wang, M., Feng, F., & Chua, T.-S. (2019). Neural Graph Collaborative Filtering. Proceedings of the 42nd International ACM SIGIR Conference on Research and Development in Information Retrieval, 165–174. https://doi.org/10.1145/3331184.3331267
Xu, D., Ruan, C., Korpeoglu, E., Kumar, S., & Achan, K. (2020). INDUCTIVE REPRESENTATION LEARNING ON TEMPORAL GRAPHS. 19.
Xu, K., Hu, W., Leskovec, J., & Jegelka, S. (2019). HOW POWERFUL ARE GRAPH NEURAL NETWORKS? 17.
Yang, C., Wang, C., Lu, Y., Gong, X., Shi, C., Wang, W., & Zhang, X. (2022). Few-shot Link Prediction in Dynamic Networks. Proceedings of the Fifteenth ACM International Conference on Web Search and Data Mining, 1245–1255. https://doi.org/10.1145/3488560.3498417
Ying, R., He, R., Chen, K., Eksombatchai, P., Hamilton, W. L., & Leskovec, J. (2018). Graph Convolutional Neural Networks for Web-Scale Recommender Systems. Proceedings of the 24th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, 974–983. https://doi.org/10.1145/3219819.3219890
Zhang, M., & Chen, Y. (2020). INDUCTIVE MATRIX COMPLETION BASED ON GRAPH NEURAL NETWORKS. 14.
Zhang, M., Wu, S., Yu, X., Liu, Q., & Wang, L. (2021). Dynamic Graph Neural Networks for Sequential Recommendation (arXiv:2104.07368). arXiv. http://arxiv.org/abs/2104.07368
Zheng, Y., Liu, S., Li, Z., & Wu, S. (2021). Cold-start Sequential Recommendation via Meta Learner. Proceedings of the AAAI Conference on Artificial Intelligence, 35(5), 4706–4713. https://doi.org/10.1609/aaai.v35i5.16601