個人化推薦系統已成為現代商業中吸引和留住客戶的關鍵策略。然而，現有的推薦系統方法如協同過濾和矩陣分解等，往往難以充分挖掘深層特徵，而基於圖神經網路的方法雖有所改進，但在解決資料稀疏性和時間序列問題上仍有不足。針對這些問題，本論文提出一種基於知識圖譜及用戶偏好演化的個人化推薦系統。
    本研究之架構，利用知識圖譜和圖神經網路技術，提取長尾關係，挖掘用戶深層隱性偏好；其次，本論文採用對比學習技術，有效緩解資料稀疏性問題；再者，透過時間窗口分割用戶互動記錄，並引入動態權重機制，識別用戶興趣的長期趨勢；最後，結合位置編碼和注意力機制，智能權衡不同時間段偏好的重要性。
    本研究的主要貢獻包括：提出了一種新的方法來處理資料稀疏和時間序列問題，有效提升了個人化推薦的效能。實驗結果顯示，本研究提出的方法在提高推薦準確性和用戶滿意度方面取得了顯著成效。

Personalized recommendation systems have become a key strategy in modern business for attracting and retaining customers. However, existing recommendation system methods such as collaborative filtering and matrix factorization often struggle to fully extract deep features. While graph neural network-based methods have shown improvements, they still fall short in addressing data sparsity and time series issues. To tackle these problems, this paper proposes a personalized recommendation system based on knowledge graphs and user preference evolution.
    The framework of this study utilizes knowledge graph and graph neural network technologies to extract long-tail relationships and mine users' deep implicit preferences. Additionally, this paper adopts contrastive learning techniques to effectively mitigate data sparsity issues. Furthermore, by segmenting user interaction records through time windows and introducing a dynamic weighting mechanism, it identifies long-term trends in user interests. Finally, it intelligently balances the importance of preferences from different time periods by combining positional encoding and attention mechanisms.
    The main contributions of this research include: proposing a novel method to address data sparsity and time series issues, effectively enhancing the performance of personalized recommendations. Experimental results demonstrate that the proposed method in this study has achieved significant improvements in recommendation accuracy and user satisfaction.

目錄
目錄	VI
圖目錄	IX
表目錄	XI
第一章、簡介	1
第二章、相關研究	7
2-1 協同過濾	7
2-2 圖神經網路	9
2-3 多技術整合	12
2-4總覽	15
第三章、背景知識	18
3-1 知識圖譜	18
3-2  LIGHTGCN	20
3-3 對比學習	22
3-4 位置編碼	24
3-5 注意力機制	27
第四章、系統架構	30
4-1 整體架構	30
4-2 資料蒐集和前處理	31
4-2-1 資料來源	31
4-2-2 資料前處理	32
4-3 建立知識圖譜	35
4-3-1 用戶特徵感知圖	35
4-3-2 商品特徵感知圖	36
4-3-3 用戶-商品互動圖	37
4-4 計算各時序偏好向量	38
4-4-1 圖神經網路	39
4-4-2 對比學習	40
4-5 多時序向量聚合	43
第五章、實驗分析	47
5-1 資料集	47
5-2 環境與系統參數設定	47
5-3 實驗結果	49
5-3-1 各時間區間模型之效能	49
5-3-2 參數敏感性分析	50
5-3-3 消融實驗	53
5-3-4 整體模型之效能	54
第六章、結論	56
參考文獻	58

圖目錄
圖1、整體架構	3
圖2、知識圖譜示意圖	18
圖3、LightGCN架構	20
圖4、LightGCN公式	20
圖5、對比學習概念	23
圖6、Position Encoding公式	26
圖7、注意力機制算法	28
圖8、整體架構	30
圖9、時間數列變化方法	34
圖10、時間區間設定	34
圖11、用戶特徵感知圖示意圖	36
圖12、商品特徵感知圖示意圖	37
圖13、用戶-商品互動圖示意圖	38
圖14、計算各時序偏好向量	39
圖15、對比學習標籤設計	42
圖16、訓練期標籤調整	43
圖17、多時序向量聚合	44
圖18、MovieLens和Last.fm資料集資料數量	47
圖19、圖神經網路和注意力機制的損失值	49
圖20、相似度閾值對應精確度	52
圖21、Attention層數對應精確度	53
圖22、效能比較	55

表目錄
表1、相關研究比較表	17
表2、實驗環境	48
表3、各參數設置	48
表4、知識圖譜提取特徵做法之比較	50
表5、消融實驗	54

[1]	J. L. Herlocker et al., "An algorithmic framework for performing collaborative filtering," in Proc. 22nd Annu. Int. ACM SIGIR Conf. Res. Dev. Inf. Retrieval, 1999, pp. 230-237.
[2]	B. Sarwar et al., "Item-based collaborative filtering recommendation algorithms," in Proc. 10th Int. Conf. World Wide Web, 2001, pp. 285-295.
[3]	Y. Koren, R. Bell, and C. Volinsky, "Matrix factorization techniques for recommender systems," Computer, vol. 42, no. 8, pp. 30-37, 2009.
[4]	X. Wang et al., "Neural graph collaborative filtering," in Proc. 42nd Int. ACM SIGIR Conf. Res. Dev. Inf. Retrieval, 2019, pp. 165-174.
[5]	X. He et al., "Lightgcn: Simplifying and powering graph convolution network for recommendation," in Proc. 43rd Int. ACM SIGIR Conf. Res. Dev. Inf. Retrieval, 2020, pp. 639-648.
[6]	X. Wang et al., "Kgat: Knowledge graph attention network for recommendation," in Proc. 25th ACM SIGKDD Int. Conf. Knowl. Discovery Data Mining, 2019, pp. 950-956.
[7]	W. Ning et al., "Automatic meta-path discovery for effective graph-based recommendation," in Proc. 31st ACM Int. Conf. Inf. Knowl. Manage., 2022, pp. 1563-1572.
[8]	D. Zou et al., "Multi-level cross-view contrastive learning for knowledge-aware recommender system," in Proc. 45th Int. ACM SIGIR Conf. Res. Dev. Inf. Retrieval, 2022, pp. 1358-1368.
[9]	H. Wang et al., "Knowledge-aware graph neural networks with label smoothness regularization for recommender systems," in Proc. 25th ACM SIGKDD Int. Conf. Knowl. Discovery Data Mining, 2019, pp. 968-977.
[10]	H. Chen et al., "Temporal meta-path guided explainable recommendation," in Proc. 14th ACM Int. Conf. Web Search Data Mining, 2021, pp. 1056-1064.
[11]	X. He et al., "Neural collaborative filtering," in Proc. 26th Int. Conf. World Wide Web, 2017, pp. 173-182.
[12]	D. Liang et al., "Variational autoencoders for collaborative filtering," in Proc. 2018 World Wide Web Conf., 2018, pp. 689-698.
[13]	W. Wang et al., "Denoising implicit feedback for recommendation," in Proc. 14th ACM Int. Conf. Web Search Data Mining, 2021, pp. 373-381.
[14]	T. Ebesu, B. Shen, and Y. Fang, "Collaborative memory network for recommendation systems," in Proc. 41st Int. ACM SIGIR Conf. Res. Dev. Inf. Retrieval, 2018, pp. 515-524.
[15]	X. Wang et al., "Neural graph collaborative filtering," in Proc. 42nd Int. ACM SIGIR Conf. Res. Dev. Inf. Retrieval, 2019, pp. 165-174.
[16]	X. Wang et al., "Disentangled graph collaborative filtering," in Proc. 43rd Int. ACM SIGIR Conf. Res. Dev. Inf. Retrieval, 2020, pp. 1001-1010.
[17]	M. Ren et al., "LR-GCN: Latent relation-aware graph convolutional network for conversational emotion recognition," IEEE Trans. Multimedia, vol. 24, pp. 4422-4432, 2021.
[18]	K. Mao et al., "UltraGCN: ultra simplification of graph convolutional networks for recommendation," in Proc. 30th ACM Int. Conf. Inf. Knowl. Manage., 2021, pp. 1253-1262.
[19]	L. Procopio, R. Tripodi, and R. Navigli, "SGL: Speaking the graph languages of semantic parsing via multilingual translation," in Proc. 2021 Conf. North Amer. Chapter Assoc. Comput. Linguistics: Human Language Technologies, 2021, pp. 325-337.
[20]	D. Zou et al., "Multi-level cross-view contrastive learning for knowledge-aware recommender system," in Proc. 45th Int. ACM SIGIR Conf. Res. Dev. Inf. Retrieval, 2022, pp. 1358-1368.
[21]	L. Huang et al., "Position-enhanced and time-aware graph convolutional network for sequential recommendations," ACM Trans. Inf. Syst., vol. 41, no. 1, pp. 1-32, 2023.
[22]	C. Xiong, R. Power, and J. Callan, "Explicit semantic ranking for academic search via knowledge graph embedding," in Proc. 26th Int. Conf. World Wide Web, 2017, pp. 1271-1279.
[23]	X. Huang et al., "Knowledge graph embedding based question answering," in Proc. 12th ACM Int. Conf. Web Search Data Mining, 2019, pp. 105-113.
[24]	Q. Guo et al., "A survey on knowledge graph-based recommender systems," IEEE Trans. Knowl. Data Eng., vol. 34, no. 8, pp. 3549-3568, 2020.
[25]	F. Schroff, D. Kalenichenko, and J. Philbin, "Facenet: A unified embedding for face recognition and clustering," in Proc. IEEE Conf. Comput. Vision Pattern Recognit., 2015, pp. 815-823.
[26]	T. Chen et al., "A simple framework for contrastive learning of visual representations," in Proc. Int. Conf. Mach. Learn., 2020, pp. 1597-1607.
[27]	S. Hochreiter and J. Schmidhuber, "Long short-term memory," Neural Comput., vol. 9, no. 8, pp. 1735-1780, 1997.
[28]	A. Vaswani et al., "Attention is all you need," in Adv. Neural Inf. Process. Syst., 2017, vol. 30.
[29]	A. Dosovitskiy et al., "An image is worth 16x16 words: Transformers for image recognition at scale," arXiv preprint arXiv:2010.11929, 2020.
[30]	C. Zhou et al., "ATRank: An attention-based user behavior modeling framework for recommendation," in Proc. 32nd AAAI Conf. Artif. Intell., 2018, pp. 4564-4571.
[31]	H. Wang et al., "Ripplenet: Propagating user preferences on the knowledge graph for recommender systems," in Proc. 27th ACM Int. Conf. Inf. Knowl. Manage., 2018, pp. 417-426.