Comparative Visual Analysis of Large Customer Feedback Based on Self-Organizing Sentiment Maps

Textual customer feedback data, e.g., received by surveys or incoming customer email notifications, can be a rich source of information with many applications in Customer Relationship Management (CRM). Nevertheless, to date this valuable source of information is often neglected in practice, as service managers would have to read manually through potentially large amounts of feedback text documents to extract actionable information. As in many cases, a purely manual approach is not feasible, we propose an automatic visualization technique to enable the geospatial-aware visual comparison of customer feedback. Our approach is based on integrating geospatial significance calculations, textual sentiment analysis, and visual clustering and aggregation based on Self-Organzing Maps in an interactive analysis application. Showing significant location dependencies of key concepts and sentiments expressed by the customer feedback, our approach helps to deal with large unstructured customer feedback data. We apply our technique to real-world customer feedback data in a case-study, showing the capabilities of our method by highlighting interesting findings. Keywords—customer relationship management, review analysis, self-organizing maps, sentiment analysis.