A simple advertisement image recommendation system based on user profiles
Enhui Yu
Liaoning University of Science and Technology
DOI: https://doi.org/10.59429/esta.v12i1.9679
Keywords: User profiling; Z-NFFM model; LightGBM-LR model; Click-through rate prediction; Advertisement cold start
Abstract
The purpose of this paper is to design and implement a simple advertisement image recommendation system based on user profiles. The system constructs a user profile by analyzing the user’s historical behavior, interests and preferences, and makes use of advanced machine learning models for personalized recommendation of advertising images. This paper firstly introduces the theoretical and technical background of the recommendation system, and then elaborates the requirements, outline design and detailed design of the system. In order to improve the recommendation effect, this paper also explores the click rate prediction method based on Z-NFFM (a kind of improved neural network factorization machine) model and the advertisement cold start strategy based on LightGBM-LR (Light Gradient Boosting Machine combined with logistic regression) model. The experimental results show that the system can effectively improve the click-through rate of advertisements, optimize the user experience, and bring higher conversion efficiency for advertisers.
References
[1] Li Xianwei . Research on Spark-based recommender system [D]. Hangzhou : Zhejiang University of Technology ,2017:14.
[2] Jiao Jian . Research on collaborative filtering recommendation algorithm based on Spark [J]. Computer Programming Skills and Maintenance ,2020(3):40-41.
[3] Liu Shanshan . Dynamic user personalized recommendation based on hybrid collaborative filtering in bigdata [J]. Software Engineering ,2019,22(3):16-19.
[4] Shi Aiwu , Li Denggui . Design and Implementation of Movie Recommendation System Based on Spark and Microservice Architecture [J]. Computer Knowledge and Technology ,2021(5):78-80.
[5] Shin Jin-Xiang , Bao Mei-Ying . Optimized recommendation algorithm gate based on user clustering and item segmentation [J]. Computer System Applications ,2019,28(6):159-164.
[6] Q. Tong , Q. Liu , S. Xu , et al. Research on Intelligent Recommendation System for E-commerce Based on Related Items [J]. Enterprise Technology and Development ,2019(12):79-80.
[7] Wang Tengyu . Analysis and design of e-commerce personalized recommendation system in the era of big data [J]. Think Tank Times ,2020(8):132-133.