Information fusion in consumer complaints: A LLM-Guided topic model for product quality improvement

Ling Yan

doi:10.1016/j.ijpe.2026.110046

Information fusion in consumer complaints: A LLM-Guided topic model for product quality improvement

2026-05-04

0

PRE

AI

L

Ling Yan

C

Chenghao Fan

K

Kening Liu

L

Linhan Ouyang

Y

Yuanyuan Gao *

DOI：10.1016/j.ijpe.2026.110046

Original

Original request for help

Save

Abstract

En 中文

• A novel LLM-guided rating-aware neural topic model (RANTM) is proposed that fuses online reviews and ratings to enhance product quality analysis. • Topic coherence and interpretability are enhanced by integrating large language model guidance with a self-attention mechanism to capture multi-rooted hierarchical topic structures. • The effectiveness of LLM-Guided RANTM is demonstrated through a case study on China’s new energy vehicle market, providing actionable insights for product improvement and stakeholder decision-making.

Keywords:

LLM-guided topic model

product quality improvement

online reviews

rating-aware neural topic model

information fusion

Information fusion in consumer complaints: A LLM-Guided topic model for product quality improvement

Abstract

Journal

Organization

Cited Papers

Citing Papers