隨著資訊技術的發達與人們的生活習慣改變，為了滿足現代人便捷的需求，各式各樣的網路平台隨之興起，例如影視平台Netflix、Disney+，電商平台Amazon、PC Home、Shopee等。然而，隨著使用者人數與商品的多樣性，過於龐大的資訊量使得商家與消費者皆須付出過多的時間成本來做出選擇，因此，推薦系統幫助人們選出較適合的產品內容，平台也因此節省了行銷的成本達成精準行銷。推薦系統主要利用消費者與商品之間的評分或留言互動來進行預測，然而，現今的商品數量與消費者數量皆有著顯著的提升，間接造成了資料稀疏性的問題，此外，如何將不同型態的消費者行為資料進行結合也是必須解決的問題。
為了解決上述問題，本研究提出了以粒子群最佳化 (Particle Swarm Optimization, PSO) 演算法尋找最合適的消費者評分相似度，避免因為資料稀疏性而導致資料失真的問題，並且利用BERT (Bidirectional Encoder Representations from Transformers) 萃取消費者留言特徵，最後再利用粒子群最佳化找出合適的權重矩陣，結合不同種資料型態之特徵。
根據實驗結果證實，結合評分與留言資料有助於提升推薦系統的表現，接著與標竿方法在Amazon的六個資料集的比較中，其中四個資料集有較好的表現，最後找出最合適的特徵萃取演算法組合。

With the development of information technology and changes in people's living habits, in order to meet the convenient needs of modern people, various online platforms have emerged, such as film and television platforms Netflix, Disney+, e-commerce platforms Amazon, PC Home, Shopee, etc. However, with the rising number of users and the diversity of products, the huge amount of information makes both merchants and consumers have to spend too much time and cost to make choices. Therefore, the recommender system is able to help people choose more suitable product content, and the platform also. Therefore, the cost of marketing is saved to achieve precision marketing. The recommender system mainly uses the rating or interaction message between consumers and products to make predictions. However, the number of products and consumers has increased significantly today, which indirectly causes the problem of data sparsity. The integration of state-of-the-art consumer behavior data is also a problem that must be solved.
In order to solve the above mentioned problems, this study proposes to use the particle swarm optimization (PSO) algorithm to find the most suitable similarity of consumer ratings to avoid the problem of data distortion due to data sparsity, use BERT (Bidirectional Encoder Representations from Transformers) to extract the characteristics of consumer messages, and finally use PSO algorithm to find the appropriate weight matrix, combining the characteristics of different data types.
According to the experimental results, it is confirmed that the combination of rating and message data can help improve the performance of the recommender system. Then, in the comparison with the benchmark methods in Amazon's six data sets, the proposed method can have better performance in four data sets. In addition, this study also finds the best combination of feature extraction methods.

Adomavicius, G., & Tuzhilin, A. (2005). Toward the next generation of recommender systems: a survey of the state-of-the-art and possible extensions. IEEE Transactions on Knowledge and Data Engineering, 17(6), 734-749.
Alijawi, B., & Kilani, Y. (2021) Using genetic algorithms for measuring the similarity values between users in collaborative filtering recommender systems. IEEE/ACIS 15th International Conference on Computer and Information Science (ICIS), 1-6.
Aljunid, M. F., & Dh, M. (2020) An efficient deep learning approach for collaborative filtering recommender system, Proceedia Computer Science, 171, 829-836.
Betru, B. T., Onana, C. A., & Batchakui, B. (2017). Deep learning methods on recommender system: a survey of state-of-the-art. International Journal of Computer Application, 162(10), 17–22.
Bokde, D., Girase, S.,& Mukhopadhyay, D. (2015) Matrix factorization model in collaborative filtering algorithms: A survey. Procedia Computer Science, 49, 136-146.
Catherine, R., Cohen, W. (2017) TransNets: Learning to transform for recommendation. 11th ACM Conference on Recommender Systems, 288–296.
Cordeau, J. F., Gendreau, M., Laporte, G., Potvin, J. Y., & Semet, F. (2002). A guide to vehicle routing heuristics. Journal of the Operation Research Society, 53(5), 512–522.
Deb, K., & Agrawal, S. (1999). Understanding interactions among genetic algorithm parameters. Foundations of genetic algorithms, 5(5), 265-286.
Devlin, J., Chang, M., Lee, K., & Toutanova, K. (2018). BERT: pre-training of deep bidirectional transformers for language understanding. Computation and Language, arXiv preprint arXiv:1810.04805.
Dezfouli, P., & Momtazi, S. (2021). Deep neural review text interaction for recommendation systems. Applied Soft Computing, 100.
Doctor, S., Venayagamoorthy, G. K., & Gudise, V. G. (2004). Optimal PSO for collective robotic search applications. Congress on evolutionary computation, IEEE, 2, 1390-1395.
Eberhart, R., & Kennedt, J. (1995). A new optimizer using particle swarm theory. Sixth International Symposium on Micro Machine and Human Science, IEEE, 39-43.
Ghasemi, N & Momtazi, S. (2020) Neural text similarity of user reviews for improving collaborative filtering recommender systems. Electronic Commerce Research and Applications, 45.
Gridach, M. (2020). Hybrid deep neural networks for recommender systems, Neurocomputing, 413, 23-30.
Himeur, Y., Alsalemi, A., Al-Kababji, A., Bensaali, F., Amira, A., Sardianos, C., Dimitrakopoulos, G. & Varlamis, I. (2021) A survey of recommender systems for energy efficiency in buildings: Principles, challenges and prospects. Information Fusion, 72, 1-21
Holland, J. H. (1975). Genetic Algorithm, Scientific american, 267(1), 66-72.
Hu, R., Dou, W., & Liu, J. (2014). ClubCF: A clustering-based collaborative filtering approach for big data application, IEEE Transactions on Emerging Topics in Computing, 2(3), 302-313
Isinkaye, F.O., Folajimi, Y.O., & Ojokoh, B.A. (2015). Recommendation systems: Principles, methods and evaluation. Egyptian Informatics Journal, 16(3), 261-273.
Kachitvichyanukul, V. (2012). Comparison of three evolutionary algorithms: GA, PSO, and DE. Industrial Engineering and Management Systems, 11(3), 215-223.
Kim, D., Park, C., Oh, J., Lee, S., & Yu, H. (2016). Convolutional matrix factorization for document context-aware recommendation. In Proceedings of the 10th ACM conference on recommender systems, 233-240.
Lauriola, I., Lavelli, A., Aiolli, F. (2021) An introduction to deep learning in natural language processing: models, techniques, and tools. Neurocomputing, 470, 443-456.
Liu, H., Wang, Y., Peng, Q., Wu, F., Gan, L., Pan, L., & Jiao, P. (2020) Hybrid neural recommendation with joint deep representation learning of ratings and reviews, Neurocomputing, 374, 77-85.
Liu, J., & Wu, C. (2017). Deep learning based recommendation: a survey. International Conference on Information Science and Applications, 451–458.
Luo, Y., Tang, L., Kim, E., & Wang, X. (2020). Finding the reviews on yelp that actually matter to me: Innovative approach of improving recommender systems, International Journal of Hospitality Management, 91.
McAuley, J., Leskovec, J. (2013). Hidden factors and hidden topics: Understanding ratings dimensions with review text. 7th ACM Conference on Recommender Systems, 165-172.
Mikolov, T., Chen, K., Corrado, G., & Dean, J. (2013). Efficient estimation of word representations in vector space. Computation and Language, arXiv preprint arXiv:1301.3781.
Mnih, A. & Salakhutdinov, R. R. (2008). Probabilistic matrix factorization. Advances in Neural Information Processing Systems, 1257-1264.
Patel, K., & Patel, H. (2020). A state-of-the-art survey on recommendation system and prospective extensions. Computers and Electronics in Agriculture, 178.
Pennington, J. Socher, R., & Manning, C. D. (2014). GloVe: global vectors for word representation. computer science department, Conference on Empirical Methods in Natural Language Processing, 1532-1543.
Portugal, I., Alencar, P., & Cowan, D. (2018). The use of machine learning algorithms in recommender systems: A systematic review. Expert System Application., 97, 205-227.
Ricci, F., Rokach, L., & Shapira, B. (2015). Recommender systems: introduction and challenges. In Recommender Systems Handbook, 1-34
Robertson, S. (2004). Understanding inverse document frequency: on theoretical arguments for IDF, Journal of Documentation, 60(5), 503-520.
Salakhutdinov, R., Mnih, A., & Hinton, G. (2007). Restricted Boltzmann machines for collaborative filtering, 24th International Conference on Machine learning, 791-798.
Seo, Y. D., Kim, Y. G., Lee, E., & Kim, H. (2021). Group recommender system based on genre preference focusing on reducing the clustering cost. Expert Systems with Applications, 183.
Wang, C., & Blei, D. M. (2011). Collaborative topic modeling for recommending scientific articles, The 17th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 448-456.
Wang, D., Yih, Y., & Ventresca, M. (2020) Improving neighbor-dased collaborative filtering by using a hybrid similarity measurement. Expert Systems with Applications, 160.
Wang, H., Wang, N., & Yeung, D. (2015). Collaborative deep learning for recommender systems, The 21th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, 1235-1244.
Wang, Z., Xia, H., Chen, S., & Chun, G. (2021). Joint representation learning with ratings and reviews for recommendation, Neurocomputing, 425, 181-190.
Wu, L., Quan, C., Li, C., Wang, Q., Zheng, B., & Luo, X. (2019). A context-aware user-item representation learning for item recommendation. ACM Transactions on Information Systems, 37, 1-29.
Zhang, Y., Liu, Z., & Sang, C. (2021). Unifying paragraph embeddings and neural collaborative filtering for hybrid recommendation. Applied Soft Computing, 106.
Zheng, L., Noroozi, V., & Yu, P. S. (2017). Joint deep modeling of users and items using reviews for recommendation. 10th ACM International Conference on Web Search and Data Mining, 425-434.