Exploring the usability and user experience of social media apps through a text mining approach

Abstract

Abstract This study aims to evaluate the applicability of a text mining approach for extracting UUX-related issues from a dataset of user comments and not to evaluate the Instagram (IG) app. This study analyses textual data mined from reviews in English written by IG mobile application users. The article’s authors used text mining (based on the LDA algorithm) to identify the main UUX-related topics. Next, they mapped the identified topics with known theoretical constructs to place them in their nomological network relevant to the usability (the 5Es framework by Quesenbery) and UX (the Honeycomb model by Morville). Finally, to expand the study with an emotional diagnosis, sentiment analysis was performed on two levels: (i) for each recognised topic, and (ii) for the full dataset to uncover general insights into users’ emotions within all reviews. The case study of the IG app confirms the usefulness of user feedback data for software development and points out that the review data have the potential for the early detection of frustration and negative feelings introduced during the use of the application. Conducting conventional UUX evaluations with users is problematic since they are remotely located, and the user-generated content of a social app undergoes continuous and frequent changes. Thus, the consecutive stages of the proposed methodology, based on text mining algorithms, constitute a proposed framework for examining the user-perceived quality projection of applications from user feedback, and they are the main contribution of this article. The used approach can be valuable for helping developers, designers and researchers to reveal user problems and fulfil user satisfaction regarding UUX aspects for specific software features.