Incheon National University Research Turns Customer Reviews into Actionable Guidance

INCHEON, South Korea, May 19, 2026 /PRNewswire/ -- Service quality improvement requires a holistic approach to analyze customer reviews and identify relevant, actionable issues. Researchers from South Korea have addressed this by developing a model that integrates service-specific aspects and customer actions, which are typically analyzed separately in text mining studies. They also validated the model using online customer reviews of a metaverse platform, demonstrating its potential to help service managers identify high-priority issues and allocate resources strategically.

Customer reviews on platforms like the Google Play Store and Apple App Store are a goldmine for companies to evaluate and improve their services. Companies specifically use text mining techniques to convert large volumes of unstructured reviews into structured insights. These techniques extract service-related aspects such as app performance, device compatibility, and payment, highlighting factors that customers are satisfied with or not. However, aspects alone do not differentiate top-priority issues from less urgent ones. Customer actions, meaning what users specifically experience or do, should also be considered. For example, in a review like 'The app keeps crashing,' the aspect 'performance issue' is vague. But when coupled with the action 'crash,' it indicates a specific issue, and its urgency increases if 'crash' is frequently mentioned in customer reviews.

Against this backdrop, a research team from South Korea, led by Associate Professor Do-Hyeon Ryu from Incheon National University, developed a model that integrates service aspects and customer actions using text mining and machine learning techniques, and successfully validated it. Dr. Ryu highlights, "Diagnosing the problem and implementing actionable, prioritized improvements necessitates both aspect identification and an understanding of the associated customer actions." Their study was made available online on December 23, 2025, and published in Volume 90 of the Journal of Retailing and Consumer Services on March 1, 2026.

Researchers outlined the model's four-stage approach for identifying top-priority issues. First, they collected online reviews from major platforms and cleaned and tagged the data. Second, they extracted aspects and associated actions using advanced natural language processing techniques. Third, each review sentence was assigned a sentiment score using a sentiment analysis tool to capture customers' emotional tone, and these scores were then averaged across sentences associated with each aspect to obtain aspect-level sentiment scores. Supervised learning models and explainable artificial intelligence (AI) techniques were also used to estimate the importance of each aspect in influencing overall customer ratings. Lastly, the aspect-action pairs were ranked based on their relevance and the emotional intensity of the reviews to identify critical improvement areas.

Researchers tested the model on 231,705 online reviews of Roblox, an online platform where users play games programmed by them or other users. They found that the model effectively identified the core technical issues causing user frustration as well as what users loved about Roblox. "Given these results, managers at Roblox can make targeted decisions, such as urgently investing in their foundational technology, while continuing to innovate on their creative and social content that drives user engagement," says Dr. Ryu.

The model is valuable to service managers across metaverse platforms and other sectors such as hospitality and retail. Dr. Ryu emphasizes, "The model supports faster problem detection, better resource allocation, and customer-centered service management." As digital services become more personalized, interactive, and AI-driven, such approaches could reshape how companies listen and respond to customers.

Reference
Title of original paper: Integrating customer actions into aspect-based service quality evaluation: A text mining framework
Journal: Journal of Retailing and Consumer Services 
DOI: https://doi.org/10.1016/j.jretconser.2025.104692

About Incheon National University
Website: https://www.inu.ac.kr/sites/inuengl/index.do?epTicket=LOG

Media Contact:
Jong Jin Lee
82-32 - 835 - 9325
[email protected]

SOURCE Incheon National University