A CNN based Personalized Product Recommendation Model
AUTHORS
Daniela De Venuto, Department of Computer Science and Engineering, University of Bologna, Bologna, Italy
Francesco Bellotti*, Department of Computer Science and Engineering, University of Bologna, Bologna, Italy
ABSTRACT
With the rapid development of information technology and Internet technology, how to dig out products or services that users are interested in from massive information and display them in a personalized manner according to the score prediction results has become a current research hotspot. Collaborative filtering is currently a technology commonly used in personalized recommendation systems. Its basic idea is to use users with the same interests in the past to choose similar products in the future. This paper proposes a personalized product recommendation model based on deep neural networks and dynamic collaborative filtering. This model uses the BERT model and two-way GRU to replace the traditional word vector text processing method, which can effectively reduce the impact of data sparsity and achieve deep hidden feature extraction. The scoring matrix is predicted by the coupled CNN network, and the product scoring prediction is realized based on the dynamic collaborate filtering of the fusion time series items. The experimental results verify that the model reduces the impact of data sparsity and cold start problems, takes into account the changes in user interest over time, and has obvious advantages in recommendation accuracy.
KEYWORDS
CNN, Recommendation model, Collaborative filtering
REFERENCES
[1] V. Subramaniyaswamy, G. Manogaran, R. Logesh, “An ontology-driven personalized food recommendation in IoT-based healthcare system,” J Supercomput, vol.75, pp.3184-3216, (2019), DOI: 10.1007/s11227-018-2331-8
[2] X. Feng, X. U. Yixiong, Y. Zeng, “Dual channel CNN recommendation algorithm combining user and product reviews,” Modern Electronics Technique, (2019)
[3] P. Zhang, Z. Zhang, T. Tian, “Collaborative filtering recommendation algorithm integrating time windows and rating predictions,” Applied Intelligence, (2019)
[4] B. Cheng, D. O. E-Commer Ce, “Research on e-commerce personalized recommendation algorithm based on collaborative filtering,” Modern Electronics Technique, (2019)
[5] Y. Han, C. Han, and L. Liu, “Collaborative filtering recommendation algorithm based on score matrix filling and user interest,” Computer Engineering, (2016)
[6] F. Yang, Y. Zheng, and C. Zhang, “Hybrid recommendation algorithm based on probability matrix factorization,” Journal of Computer Applications, (2018)
[7] G. Liu, K. Meng, and J. Ding, “An entity-association-based matrix factorization recommendation algorithm,” Computers, Materials and Continua, vol.58, no.1, pp.101-120, (2019)
[8] K. Lee and Suk, “Quantitation of hypoechoic lesions for the prediction and Gleason grading of prostate cancer: A prospective study,” World Journal of Urology, (2018)
[9] J. L. Che, L. W. Tang and S. J. Deng, “Chinese word segmentation based on bi-directional GRU-CRF model,” Fire Control and Command Control, (2019)
[10] W. Wang, Y. Sun, and Q. Qi, “Text sentiment classification model based on BiGRU-attention neural network,” Application Research of Computers, (2019)
[11] F Kokkinos, A. Potamianos, “Structural attention neural networks for improved sentiment analysis,” (2017)
[12] O. Xi, Z. Pan, and H. L. Cheng, “Sentiment analysis using convolutional neural network,” IEEE International Conference on Computer and Information Technology Ubiquitous Computing and Communications Dependable, IEEE, (2015)
[13] M. Frans, M. Siti, and P. Setia, “Sentiment analysis: A comparison of deep learning neural network algorithm with SVM and nave Bayes for Indonesian text,” Journal of Physics Conference, 971:012049, (2018)