Abstract

Over the last 20 years, the increasing digitalization of societies gave rise to an unprecedented data abundance. News portals, social media, e-commerce – these are exemplary indicators for a major shift of everyday life to the world wide web. This development provides many important opportunities for both researchers and businesses. For instance, online media allow insights into private consumer activities that used to be notoriously hard to investigate via classical surveys in the past.

However, the analysis of online media data is methodologically challenging. The data is known to be noisy, e.g., due to people providing false information on social media, news portals distributing fake news, or botnets distorting the public discourse. Moreover, well-established inference theory from classical statistics is often not applicable, as online media data sets are rarely the result of a random sampling process.

Against this background, the presentation will address how online media data is collected, processed, and analysed at CURE Intelligence. CURE is a Luxembourg-based IT company founded in 2009. We assess more than 180 million data sources in over 150 languages worldwide in real-time. For this, we create adaptive information systems based on contextual combinations of natural language processing, machine learning algorithms, statistical weighting, as well as robust optimization theory. These systems are subsequently deployed to provide B2B solutions in fields like business intelligence, supply chain analytics, as well as data-driven marketing.