A Scalable Unsupervised Framework for multi-aspect labeling of Multilingual and Multi-Domain Review Data

TL;DR

We address the challenge of scalable, multilingual, cross-domain ABSA by introducing MUSCAD, an unsupervised framework that first generates initial aspect candidates via K-means on Word2Vec embeddings, then learns context-aware, aspect-aligned sentence representations through multi-head attention and a Max-Margin Loss with negative sampling. Aspect categories are finally refined with domain knowledge and GPT-based prompts to produce interpretable labels, enabling large-scale automatic labeling across Korean and English reviews in hotel, food, and beauty domains. Across extensive experiments, MUSCAD achieves near state-of-the-art downstream performance, consistently outperforms LLM-based labeling in consistency and cost, and yields coherent, domain-reflective aspect representations in a fully unsupervised setting, validated by human evaluators. The framework thus provides a scalable, multilingual foundation for ABSA and related sentiment analysis tasks, with promising avenues for AI-agent driven labeling and aspect-based summarization.

Abstract

Effectively analyzing online review data is essential across industries. However, many existing studies are limited to specific domains and languages or depend on supervised learning approaches that require large-scale labeled datasets. To address these limitations, we propose a multilingual, scalable, and unsupervised framework for cross-domain aspect detection. This framework is designed for multi-aspect labeling of multilingual and multi-domain review data. In this study, we apply automatic labeling to Korean and English review datasets spanning various domains and assess the quality of the generated labels through extensive experiments. Aspect category candidates are first extracted through clustering, and each review is then represented as an aspect-aware embedding vector using negative sampling. To evaluate the framework, we conduct multi-aspect labeling and fine-tune several pretrained language models to measure the effectiveness of the automatically generated labels. Results show that these models achieve high performance, demonstrating that the labels are suitable for training. Furthermore, comparisons with publicly available large language models highlight the framework's superior consistency and scalability when processing large-scale data. A human evaluation also confirms that the quality of the automatic labels is comparable to those created manually. This study demonstrates the potential of a robust multi-aspect labeling approach that overcomes limitations of supervised methods and is adaptable to multilingual, multi-domain environments. Future research will explore automatic review summarization and the integration of artificial intelligence agents to further improve the efficiency and depth of review analysis.

Figures (10)

• Figure 1: MUSCAD Framework
• Figure 2: Hotel Aspect Categorization Prompts
• Figure 3: Pipeline of MUSCAD : Aspect Term Extraction and Category Labeling from Reviews
• Figure 4: Human Evaluation Criteria for Qualitative Assessment
• Figure E.4: Few-Shot Examples for the Hotel & Food Domains. This figure demonstrates how examples of multi-aspect category labels assigned to reviews in the Hotel and Food domains are provided to the model in a few-shot setting. By referring to the aspect categories assigned to each review, the model is guided to classify new sentences more accurately.