Visualization Tool for Interpreting User Needs From User-Generated Content via Text Mining and Classification

Abstract

The amount of user-generated content related to consumer products continues to grow as users increasingly take advantage of forums, product review sites, and social media platforms. The content is a promising source of insight into users’ needs and experiences. However, the challenge remains as to how concise and useful insights can be extracted from large quantities of unstructured data. We propose a visualization tool which allows designers to quickly and intuitively sift through large amounts of user-generated content and derive useful insights regarding users’ perceptions of product features. The tool leverages machine learning algorithms to automate labor-intensive portions of the process, and no manual labeling is required by the designer. Language processing techniques are arranged in a novel way to guide the designer in selecting the appropriate inputs, and multidimensional scaling enables presentation of the results in concise 2D plots. To demonstrate the efficacy of the tool, a case study is performed on action cameras. Product reviews from Amazon.com are analyzed as the user-generated content. Results from the case study show that the tool is helpful in condensing large amounts of user-generated content into useful insights, such as the key differentiations that users perceive among similar products.