What ties to what? Visualizing large-scale customer text data with bipartite graphs

Which suppliers are most likely to have delivery or quality issues? Does service, product placement, or price make the biggest difference in customer sentiment? Text data from sources like email and social media can give answers. Mark Turner explains how to see the associations between any two variables in text data by combining text analytics and the bipartite graph visualization technique.

Which suppliers are most likely to have delivery or quality issues? Does service, product placement, or price make the biggest difference in customer sentiment? Finding the answers to these questions in structured data is often straightforward, but can we answer them using the unstructured data (free text) in emails, social media, call center transcripts, product reviews, and other sources? Mark Turner explains how to clearly see the associations between any two variables in text data by combining large-scale in-database text analytics and the bipartite graph visualization technique. Mark describes which text analytics methods to use for various operational business questions and how to show the associations clearly in a bipartite graph, offering insight into which associations are strongest and which are weakest. This powerful combination of methods gives operational value to the increasingly huge amounts of text data, in which customers express their likes, dislikes, preferences, and issues.