A Knowledge-Based User Feedback Classification Approach for Software Support
摘要
The analysis of the textual content of user opinions on social networks about software applications in use can provide valuable information to the development and support teams, in terms of errors, dissatisfactions, new functional requirements, among others. The paper presents a solution based on intelligent technologies to automatically classify whether or not the content of a review is relevant to a software support team. This solution combines machine learning algorithms, with the use of a domain-specific glossary for feature selection, in predicting the relevance of reviews. The proposed solution was evaluated experimentally with three datasets, specifically Facebook, Tapfish and SwiftKey, and the results obtained were very promising.