Review Analyzer Textual and Visual Data
Processing image and textual review data to provide product improvement suggestions with short- and long-term action plans.
Challenge
Unlock circularity and avoid waste.
Imagify seeks to come up with product improvement suggestions based on customer feedback. Processing huge chunks of visual and textual data is a great challenge. A machine learning system that processes this information and recognizes common problems is vital. Support Imagify with a process that translates qualitative context information around photos into key-take-aways to help engineers and designers even more to minimize their time-efforts and maximize their ability to deduct actionable insights.
Team
Sarah Sykora, Maria Ivanova, Verena Vogt, Samet Berk Oksuz, Zeynep Kocaahmet
Contact: ssykora@calpoly.edu
About the prototype
After going through different types of customer reviews, we came to identify 4 main categories of feedback: fatal defects, normal wear and tear, loved features, and improvement suggestions. Review timing, content, focus are all important elements that need to be analyzer at deeper levels. On the short-term, we came up with first a review analyzer that is made to learn and categorize these 4 types and classify reviews accordingly. Then, a chatbot that provides product improvement suggestion for product developers is built on top of that. The next step of our prototype is a plan to develop a visual analyzer that extracts the location of the features. Then a visual analyzer is put into action to associate location and categories. Hence, the product developer has the useful information from reviews at hand.
Outputs
Example of the prototype
Impressions
