Automating App Review Response Generation Based on Contextual Knowledge

User experience of mobile apps is an essential ingredient that can influence the user base and app revenue. To ensure good assist development, several prior studies resort to analysis reviews, a type repository directly reflects opinions about apps. Accurately responding reviews one ways relieve concerns thus improve experience. However, response quality existing method relies on pre-extracted features from other tools, including manually labelled keywords predicted review sentiment, which may hinder generalizability flexibility method. In this article, we propose novel neural network approach, named CoRe, with contextual knowledge naturally incorporated without involving external tools. Specifically, CoRe integrates two types in training corpus, official descriptions store responses retrieved semantically similar for enhancing relevance accuracy generated responses. Experiments practical data show outperform state-of-the-art by 12.36% terms BLEU-4, metric widely used evaluate text generation systems.