DSSQEM-IHK: Dynamic SERVQUAL Service Quality Evaluation Method Integrating Hypergraph Knowledge

Abstract

Service quality evaluation has long faced challenges such as quantification difficulties, cumbersome processes, and low efficiency. This study proposes an innovative hypergraph deep learning method that integrates the SERVQUAL model with topic modeling and sentiment analysis techniques to dynamically and continuously obtain customer satisfaction information from massive online reviews, thereby constructing a comprehensive real-time service quality feedback mechanism. The method extracts review topics and keywords based on pre-trained language models and maps them onto the five SERVQUAL dimensions. Subsequently, the reviews are constructed into a semantically rich hypergraph representation. Then, a hypergraph attention network is employed to extract customers’ perceived service quality values. Finally, the perceived quality is compared with the expected quality to estimate the overall service quality level. The results demonstrate that this method can effectively evaluate the service quality embedded in online reviews. For the Arboricultural Services (AS) dataset containing 7210 reviews, the hypergraph network performs well in extracting perceived quality, achieving an accuracy of 85.2% and an F1-score of 82.6%, laying the foundation for precise service quality computation. The network's performance surpasses FastText by 5.4% and 4.8%, outperforms KG-GCN by 2.5% and 1.5%, and exceeds the Transformer and EGJO-LSTM models by 0.4%, 0.8%, and 1.2%, 1.0%, respectively. This study offers businesses an effective solution for service quality evaluation that does not rely on survey methods, thereby improving efficiency, reducing costs, and possessing significant theoretical value and practical prospects.