A Quantitative Evaluation Approach for Online Review Quality Based on the AHP-Fuzzy Comprehensive Evaluation
Mingwei Tang, Yihan Qin, Lu Zhang, Hongru Lu, Tingyu Zhang · 2024
Online reviews, being rich in user experience, offer valuable insights for consumers. Despite their high reference value, assessing their quality remains challenging due to their unstructured nature. The paper employs fuzzy comprehensive evaluation with AHP to evaluate the quality of online reviews in quantity. A structured framework with 5 dimensions and 13 features is proposed to capture different aspects of review quality through detailed questionnaires based on the literature reviews. Statistical analysis and natural language processing (NLP) methods are employed to study the quantitative calculation methods for the 13 features, as well as the calculation method for the quantitative values of each dimension based on the weighted average of feature values. On this basis, the fuzzy comprehensive evaluation method was applied to integrate the values of each dimension to obtain a quantitative representation of the online review quality. Finally, the accuracy of the proposed approach is validated by adopting human evaluation as the benchmark and comparing the results with that of ChatGPT 4.0. The study's findings indicate that the proposed approach outperforms ChatGPT 4.0, achieving a 65.8% accuracy rate. This approach could provide objective assessments for comprehensive and detailed reviews, aiding consumers in efficiently evaluating online review quality.