KANDINSKY : visualization of social comments on mobile devices

Abstract

Used by half of the world’s population, social media allows users to create and share content in various formats which can collectively be termed as anchor posts. Such anchor posts may attract hundreds of comments from many social users, initiating online conversations, which could generate meaningful insights for both individuals and organizations. Given the large volume of conversations happening, generating insights by simply browsing the comments is tedious, time-consuming and ineffective. Thus, there is a need for a user-friendly and effective tool to facilitate the in-depth analysis and understanding of the social discussions landscape of an anchor post. Moreover, given that mobile devices are one of the fastest growing platforms with which people access social media, it is important to realize the said visualizations on mobile devices. This project has developed KANDINSKY Mobile, a mobile-based end-to-end visualization application, for supporting the user on analysing copious and complicated social conversations. The challenges are largely due to the unique characteristics of mobile devices, including the small screen size, limiting computing resources, network constraint, and different ways users navigating mobile interfaces. Inspired by abstract paintings of renowned artist Wassily Kandinsky, KANDINSKY Mobile’s visualization scheme is based on a collection of colorful circles and concentric circles. It allows mobile users to visualize a bird’s-eye view of the conversation landscape, analyze evolution of discussions over time, perform keyword search on the discussion threads, and find comments that are topically similar to a comment-of-interest. Based on usability evaluation, the visualization design and human computer interaction design of KANDINSKY Mobile are both intuitive and satisfying for users. There are several directions of future work for KANDINSKY Mobile, including support for more social media platforms and languages, and incorporation of sentiment analysis and function.