Towards Bridging Review Sparsity in Recommendation with Textual Edge Graph Representation

2025-08-01

AI Summary

In real-world scenarios, user reviews are extremely sparse, severely limiting the accuracy and interpretability of recommendation models. To address this, we propose TWISTER—a novel framework that pioneers the integration of Textual Edge Graphs (TEGs) with graph-aware large language model (LLM) aggregators. TEGs explicitly encode semantic relationships on user-item interaction edges, while line-graph transformation maps the original interaction graph into an edge-level topology. The graph-aware LLM aggregator then jointly fuses structural dependencies and natural language representations within local neighborhoods to generate context-aware missing reviews. Evaluated on Amazon and Goodreads datasets, TWISTER significantly outperforms numerical, GNN-based, and LLM-based baselines across three dimensions: recommendation accuracy, and the authenticity, specificity, and interpretability of generated reviews—achieving state-of-the-art performance in all.

Technology Category

• Data Mining & Knowledge Management: Conversational Systems for Recommendation & Retrieval
• Natural Language Processing: Sentence-level Semantics, Textual Inference, etc.
• Machine Learning: Large Multimodal Models (LMMs)

Application Category

• User Modeling, Personalization and Recommendation: Large Language Models (LLM) for user modeling and recommendation
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

Textual reviews enrich recommender systems with fine-grained preference signals and enhanced explainability. However, in real-world scenarios, users rarely leave reviews, resulting in severe sparsity that undermines the effectiveness of existing models. A natural solution is to impute or generate missing reviews to enrich the data. However, conventional imputation techniques -- such as matrix completion and LLM-based augmentation -- either lose contextualized semantics by embedding texts into vectors, or overlook structural dependencies among user-item interactions. To address these shortcomings, we propose TWISTER (ToWards Imputation on Sparsity with Textual Edge Graph Representation), a unified framework that imputes missing reviews by jointly modeling semantic and structural signals. Specifically, we represent user-item interactions as a Textual-Edge Graph (TEG), treating reviews as edge attributes. To capture relational context, we construct line-graph views and employ a large language model as a graph-aware aggregator. For each interaction lacking a textual review, our model aggregates the neighborhood's natural-language representations to generate a coherent and personalized review. Experiments on the Amazon and Goodreads datasets show that TWISTER consistently outperforms traditional numeric, graph-based, and LLM baselines, delivering higher-quality imputed reviews and, more importantly, enhanced recommendation performance. In summary, TWISTER generates reviews that are more helpful, authentic, and specific, while smoothing structural signals for improved recommendations.

Problem

• Addressing review sparsity in recommender systems
• Imputing missing reviews with semantic and structural signals
• Enhancing recommendation performance via textual edge graphs

Innovation

• Represents user-item interactions as Textual-Edge Graph
• Uses LLM as graph-aware aggregator for review generation
• Combines semantic and structural signals for imputation