Journal of Engineering, Project, and Production Management, 2026, 16(4), 2025-291
An Improved CF Algorithm Personalized Recommendation Model for E-Commerce Cold-Start Problems
Associate Professor, School of Business Management, Hangzhou Polytechnic, Hangzhou, 311402, China, E-mail: chenwenwei1979@163.com
Project Management
Received November 26, 2025; revised February 2, 2026; accepted February 9, 2026
Available online May 29, 2026
Abstract: Given the low recommendation accuracy of traditional algorithms in cold-start scenarios in e-commerce, this study proposes a new personalized recommendation algorithm. First, the independent meta-learning model is used to improve the collaborative filtering algorithm, enabling fast parameter adaptation for new users. Multimodal content encoding is combined to introduce product-related multimodal information to alleviate semantic sparsity. Second, a population-society dual regularization constraint is introduced to address the vector offset problem during fine-tuning in Model-Agnostic Meta-Learning (MAML). Ultimately, a personalized recommendation model is constructed for cold-start scenarios. The model was validated on the Amazon Electronics Dataset (AED) and AlicCP datasets. In the experiment, Hit Rate at 5 (HR@5), which measures how often the correct item appears among the top five recommendations, improved by 62.99% on AED when compared to the baseline. In the AlicCP dataset, its HR@5 and Novelty at 5 (Novelty@5) increased by an average of 26.02% and 15.23% compared to other methods. The research model can effectively achieve accurate recommendations in cold-start scenarios, bring users a better shopping experience, and improve the platform’s first purchase conversion rate and next day retention rate.
Keywords: Collaborative filtering, e-commerce, cold-start recommendation, multimodal content encoding, meta learning.
Copyright © Journal of Engineering, Project, and Production Management (EPPM-Journal).
This work is licensed under a Creative Commons Attribution-NonCommercial-NoDerivs 3.0 Unported License.
Requests for reprints and permissions at eppm.journal@gmail.com.
Citation: Chen, W. (2026). An Improved CF Algorithm Personalized Recommendation Model for E-Commerce Cold-Start Problems. Journal of Engineering, Project, and Production Management, 16(4), 2025-291.
