What question did this study set out to answer?

This research aims to improve personalized recommendations in e-commerce by enhancing workflows in the user-agent-platform paradigm using Large Language Models (LLMs).

July 12, 2026Journal of theoretical and applied electronic commerce researchOpen Access

Enhancing Personalized E-Commerce Recommendations Under the User–Agent–Platform Paradigm: An LLM-Driven Method

View Full Paper

Ask AI

Bookmark

Authors

JXJunbiao Xu ZZZhicai Zhang CZChong Zhang

Discussion

Member takes

Overview

Randomized trial reveals improved recommendation quality in e-commerce using an LLM-driven workflow for user-agent-platform interactions, suggesting beneficial user mediation.

Key Points

This research aims to improve personalized recommendations in e-commerce by enhancing workflows in the user-agent-platform paradigm using Large Language Models (LLMs).
Developed an LLM-driven workflow named ABP, containing three modules: Adaptive Description Enrichment (ADE), Batch-balanced Sampling Strategy (BSS), and Prompt-driven Workflow Optimization (PWO).
Tested ABP on four real-world datasets: Amazon Book, Amazon Movietv, and Yelp, focusing on balancing sample inputs and enriching item descriptions.
Measured average relative improvements and output stability across multiple runs and datasets.
ABP achieved an average relative improvement of 24.62% over the baseline i2Agent across 16 dataset-metric pairs.
Limited run-to-run dispersion was observed on datasets such as Amazon Book and Movietv, indicating stable performance.
Goodreads showed more variation, but still within acceptable bounds, suggesting ongoing potential for further improvements.

Cite This Study

Xu et al. (2026) studied this question.

synapsesocial.com/papers/6a5331ce4f7abc118adedbd6 https://doi.org/10.3390/jtaer21070223

View Full Paper

Ask AI

Bookmark

July 12, 2026Journal of theoretical and applied electronic commerce researchOpen Access

Enhancing Personalized E-Commerce Recommendations Under the User–Agent–Platform Paradigm: An LLM-Driven Method

View Full Paper

Ask AI

Bookmark

Authors

JXJunbiao Xu ZZZhicai Zhang CZChong Zhang

Discussion

Member takes

Overview

Randomized trial reveals improved recommendation quality in e-commerce using an LLM-driven workflow for user-agent-platform interactions, suggesting beneficial user mediation.

Key Points

This research aims to improve personalized recommendations in e-commerce by enhancing workflows in the user-agent-platform paradigm using Large Language Models (LLMs).
Developed an LLM-driven workflow named ABP, containing three modules: Adaptive Description Enrichment (ADE), Batch-balanced Sampling Strategy (BSS), and Prompt-driven Workflow Optimization (PWO).
Tested ABP on four real-world datasets: Amazon Book, Amazon Movietv, and Yelp, focusing on balancing sample inputs and enriching item descriptions.
Measured average relative improvements and output stability across multiple runs and datasets.
ABP achieved an average relative improvement of 24.62% over the baseline i2Agent across 16 dataset-metric pairs.
Limited run-to-run dispersion was observed on datasets such as Amazon Book and Movietv, indicating stable performance.
Goodreads showed more variation, but still within acceptable bounds, suggesting ongoing potential for further improvements.

Cite This Study

Xu et al. (2026) studied this question.

synapsesocial.com/papers/6a5331ce4f7abc118adedbd6 https://doi.org/10.3390/jtaer21070223

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

View Full Paper

Ask AI

Bookmark