Graph-based service quality evaluation through mining Web reviews

Abstract

This work tries to find a possible solution to the basic research problem: how to conduct service quality evaluation through mining Web reviews? To address this problem, this work proposes a novel approach to service quality evaluation which has two essential subtasks: 1) finding the most important service aspects, and 2) measuring service quality using ranked service aspects. We propose three graph-based ranking models to rank service aspects and a simple linear method of measuring service quality. Empirical experimental results show all our three methods outperform the approach of Noun Frequency. We also show the effectiveness of our service quality evaluation method by conducting intensive regression experiments.