Sentiment based approval prediction for enhancement reports

摘要

The maintenance and evolution of the software application is a continuous phase in the industry. Users are frequently proposing enhancement requests for further functionalities. However, although only a small part of these requests are finally adopted, developers have to go through all of such requests manually, which is tedious and time consuming. To this end, in this paper we propose a sentiment based approach to predict how likely enhancement reports would be approved or rejected so that developers can first handle likely-to-be-approved requests. This could help the software applications to compete in the industry by upgrading their features in time as per user's requirements. First, we preprocess enhancement reports using natural language preprocessing techniques. Second, we identify the words having positive and negative sentiments in the summary attribute of the enhancements reports and calculate the sentiment of each enhancement report. Finally, with the history data of real software application, we train a machine learning based classifier to predict whether a given enhancement report would be approved. The proposed approach has been evaluated with the history data from real software applications. The cross-application validation suggests that the proposed approach outperforms the state-of-the-art. The evaluation results suggest that the proposed approach increases the accuracy from 70.94% to 77.90% and improves the F-measure significantly from 48.50% to 74.53%. (C) 2019 Elsevier Inc. All rights reserved.