March 3, 2026Open Access

AI models for demographic prediction in e-commerce: age and gender from initial user interactions

Puntos clave

Automatic user profiling reveals high accuracy in gender determination and age estimation from user interactions.
AI models trained on a dataset of 592 volunteers showed effective demographic prediction using interaction patterns.
Evaluation metrics like Area Under the Curve of the Precision-Recall Curve (AUC-PR) demonstrated the models' efficacy.
These findings indicate the potential for enhanced personalized experiences in e-commerce, driving customer engagement.

Resumen

This research investigates the feasibility of inferring key demographic features, specifically gender and age, from analyzing the performance of fundamental interactive tasks by anonymous users of an e-commerce site. A dataset of interaction patterns from 592 volunteers, encompassing tasks such as Point & Click, Drag & Drop, Text Selection, Text Editing , and Menu Item Selection , was collected and analyzed. Various artificial intelligence (AI) models were trained to identify predictive correlations between task execution times and user demographics. Rigorous evaluation using metrics like Area Under the Curve of the Precision-Recall Curve (AUC-PR) demonstrated the models’ efficacy in automatic user profiling, revealing the potential for highly accurate gender determination and age group estimation. These findings highlight the significant potential of AI to create systems for automatically inferring user demographics from initial interactions, enabling dynamic and personalized e-commerce experiences that enhance customer perception and potentially increase revenue.

Leer artículo completoexternamente

Me gusta

Guardar

Ver artículo completo