1 Mar 2018
16:00 — 16:30
From ancient Egyptian stone carvings to electronic messages on Facebook timelines – text has always been a main driver of human culture. What is different now? It has never been easier to create text than today. This is why the amount of text is constantly growing - also due to user-generated content (UGC): in company wikis, digitalized customer service calls, user groups, forum posts, tweets, facebook posts, product reviews, .... Users write what they think. While this can sometimes feel like losing control over your own reputation, it can be managed and even turned into a treasurable source for innovation, ideas and service improvements. And this information comes directly from customers! This talk sums up the results of a study on how unsupervised and supervised machine learning can be used to get insights from user-generated content. We find the most exciting topics which were previously hidden, create clusters from user profiles with engagement and sentiment. We follow the dynamics of interacting user groups (e.g. influencers vs. trolls). Network graphs show the central nodes in our analysis.