Actionable Advice from Reviews via Mixture of LoRA Experts: A Two-LLM Pipeline for Issue Extraction and Business Recommendations

2026-01-18

AI Summary

This work addresses the challenge of transforming unstructured user feedback from reviews into actionable business recommendations. The authors propose a two-stage large language model architecture: an Issue model first identifies and categorizes key problems in user comments, and an Advice model then generates targeted suggestions based on the extracted issue representations. A novel hybrid LoRA-of-experts mechanism is introduced, dynamically fusing multi-domain knowledge at the token level during inference to avoid full-parameter fine-tuning, thereby significantly enhancing the professionalism and adaptability of the generated advice. Trained on synthetic triplets constructed from Yelp reviews, the approach substantially outperforms prompt engineering and single-adapter baselines in both airline and restaurant domains, achieving superior performance across eight dimensions—including actionability and specificity—while maintaining high efficiency and output quality.

Technology Category

• Natural Language Processing: (Large) Language Models
• Machine Learning: Large Multimodal Models (LMMs)
• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions
• Search and Retrieval-Augmented AI: Large language models for search

Abstract

Customer reviews contain detailed, domain specific signals about service failures and user expectations, but converting this unstructured feedback into actionable business decisions remains difficult. We study review-to-action generation: producing concrete, implementable recommendations grounded in review text. We propose a modular two-LLM framework in which an Issue model extracts salient issues and assigns coarse themes, and an Advice model generates targeted operational fixes conditioned on the extracted issue representation. To enable specialization without expensive full fine-tuning, we adapt the Advice model using a mixture of LoRA experts strategy: multiple low-rank adapters are trained and a lightweight gating mechanism performs token-level expert mixing at inference, combining complementary expertise across issue types. We construct synthetic review-issue-advice triples from Yelp reviews (airlines and restaurants) to supervise training, and evaluate recommendations using an eight dimension operational rubric spanning actionability, specificity, feasibility, expected impact, novelty, non-redundancy, bias, and clarity. Across both domains, our approach consistently outperforms prompting-only and single-adapter baselines, yielding higher actionability and specificity while retaining favorable efficiency-quality trade-offs.

Problem

• actionable advice
• customer reviews
• review-to-action generation
• business recommendations
• issue extraction

Innovation

• Mixture of LoRA Experts
• Two-LLM Pipeline
• Actionable Recommendation Generation
• Issue Extraction
• Low-Rank Adaptation