Defective products cannot meet the needs of consumers, and they also bring security risks. Prior product weakness detection methods resort to time and money consuming questionnaire, and the accuracy could not be guaranteed. As a result, this paper proposes a novel opinion-aware approach, PRODWeakFinder, which extracts product weaknesses from online reviews through sentiment analysis. PRODWeakFinder detects product weakness by considering both comparative and non-comparative evaluations in online reviews. For comparative evaluation, we build a feature comparison network, where the authority score of each node is assessed. For non-comparative evaluation, sentiment score is calculated through sentiment analysis. The composite score is calculated by combing both comparative and non-comparative evaluations. The features with lower score represent potential product weaknesses. The experiment result shows that PRODWeakFinder outperforms baseline methods in terms of both accuracy and reducing of operating costs.
Wang, Hongwei; Wang, Wei; and Yin, Pei, "PRODWEAKFINDER: AN INFORMATION EXTRACTION SYSTEM FOR DETECTING PRODUCT WEAKNESSES IN ONLINE REVIEWS BASED ON SENTIMENT ANALYSIS" (2014). PACIS 2014 Proceedings. 101.