Article: User purchase intention prediction algorithm for e-commerce precise operation needs Journal: International Journal of Information and Communication Technology (IJICT) 2026 Vol.27 No.87 pp.71 - 95 Abstract: To predict users' purchase intention, in terms of feature extraction method, an improved graph sampling aggregation algorithm that introduces time series and location information is designed. Next, the study proposes a user purchase intention prediction algorithm that combines graph sampling aggregation algorithm and higher-order graph neural network. The experimental results based on the public dataset of mainstream e-commerce platforms from 2022 to 2024 show that, the feature extraction capability of the improved graph sampling aggregation algorithm was significantly better than the traditional method. The performance of the willingness prediction algorithm was even more outstanding. In experiments comparing it with high-order graph neural networks, lightweight gradient boosting machines, graph convolutional network logistic regression combination models, etc., its accuracy was above 92%, the maximum F1 value was 0.967. The performance of the feature extraction method and prediction algorithm is better than that of the comparison model. Inderscience Publishers - linking academia, business and industry through research

Title: User purchase intention prediction algorithm for e-commerce precise operation needs

Authors: Peipei Wang

Addresses: School of Business Administration, Henan University of Animal Husbandry and Economy, Zhengzhou, 450044, China

Abstract: To predict users' purchase intention, in terms of feature extraction method, an improved graph sampling aggregation algorithm that introduces time series and location information is designed. Next, the study proposes a user purchase intention prediction algorithm that combines graph sampling aggregation algorithm and higher-order graph neural network. The experimental results based on the public dataset of mainstream e-commerce platforms from 2022 to 2024 show that, the feature extraction capability of the improved graph sampling aggregation algorithm was significantly better than the traditional method. The performance of the willingness prediction algorithm was even more outstanding. In experiments comparing it with high-order graph neural networks, lightweight gradient boosting machines, graph convolutional network logistic regression combination models, etc., its accuracy was above 92%, the maximum F1 value was 0.967. The performance of the feature extraction method and prediction algorithm is better than that of the comparison model.

Keywords: purchase intention; prediction; GraphSAGE; HOGNN; electric commerce; e-commerce.

DOI: 10.1504/IJICT.2026.155640

International Journal of Information and Communication Technology, 2026 Vol.27 No.87, pp.71 - 95

Received: 26 Jan 2026
Accepted: 08 May 2026
Published online: 07 Aug 2026 *

Title: User purchase intention prediction algorithm for e-commerce precise operation needs

Keep up-to-date