Linguistic Sentiment Quality for AI Driven Service Evaluation

Abstract

Sentiment analysis is widely used in artificial intelligence to analyse customer feedback from online reviews. Most approaches focus on classifying reviews into positive, negative, or neutral categories. While this provides a general overview, it often overlooks how meaning is constructed through language, particularly in service contexts. The concept of Linguistic Sentiment Quality (LSQ) emerged from earlier empirical work examining the relationship between linguistic features and perceived service quality in airline customer reviews. Previous findings based on 958 online reviews revealed noticeable shifts in both sentiment and language use following the COVID-19 pandemic, particularly through stronger negative tone and emotionally loaded expressions. Building on these findings, this paper proposes an extended LSQ framework for AI-driven service evaluation. The framework combines AI-driven sentiment analysis, linguistic interpretation, and service quality dimensions to move beyond surface-level classification toward a deeper understanding of the customer experience. A qualitative re-evaluation of 10 complex, misclassified neutral reviews demonstrated that 100% contained definitive underlying sentiment (6 strongly negative and 4 positive), which the LSQ framework successfully resolved and mapped to operational service dimensions. The framework advances explainable AI (XAI) and service operations by providing interpretable linguistic features that clarify how sentiment is constructed.