Paper Info
Title
User Feedback from Tweets vs App Store Reviews: An Exploratory Study of Frequency, Timing and Content
Abstract
Context: User feedback on apps is essential for gauging market needs and maintaining a competitive edge in the mobile apps development industry. App Store Reviews have been a primary resource for this feedback, however, recent studies have observed that Twitter is another potentially valuable source for this information. Objective: The objective of this study is to assess user feedback from Twitter in terms of timing as well as content and compare with the App Store reviews. Method: This study employs various text analysis and Natural Language Processing methods such as semantic analysis and Latent Dirichlet Allocation (LDA) to analyze tweets and App Store Reviews. Additionally, supervised learning classifiers are used to classify them as semantically similar tweet and App Store reviews. Results: In spite of a difference in the magnitude between tweets and App Store Review counts, frequency analysis shows that bug report and feature request are discussed mostly on Twitter first as the number of Tweets during the reporting time reached the peak a few days earlier. Likewise, timing analysis on a set of 426 tweets and 2,383 reviews (which are bug reports and feature requests) show that approximately 15% appear on Twitter first. Of these 15% tweets, 72% are related to functional or behavioural aspects of the mobile app. Content analysis shows that user feedback in tweets mostly focuses on critical issues related to the feature failure and improper functionality. Conclusion: The results of this investigation show that the Twitter is not only a strong contender for useful information but also a faster source of information for mobile app improvement.
Year
DOI
Venue
2018
10.1109/AIRE.2018.00008
2018 5th International Workshop on Artificial Intelligence for Requirements Engineering (AIRE)
Keywords
Field
DocType
social media, user feedback, mobile apps, machine learning, text analysis, natural language processing, mobile application improvement.
Latent Dirichlet allocation,Content analysis,App store,Information retrieval,Computer science,Software bug,Supervised learning,Exploratory research,Semantics,Market research
Conference
ISBN
Citations 
PageRank 
978-1-5386-8405-4
1
0.37
References 
Authors
0
2
Name
Order
Citations
PageRank
Gouri Deshpande110.37
Jon G. Rokne226345.63