The EUMSSI Project - Event Understanding through Multimodal Social Stream Interpretation

Journalists, as well as users at home, face increasing amounts of data from a large variety of sources, both in professionally curated media archives and in the form of user-generated-content or social media. This provides a great opportunity at the same time as a great challenge to use all of this information, which EUMSSI approaches by providing semantically rich analysis of multimedia content, together with intuitive visual interfaces to explore the data and gain new insights. The goal of the EUMSSI project is to provide a complete system for large-scale multimedia analysis, including a wide range of analysis components working on different media modalities (video, audio, text). Additionally, we have developed two example applications (demonstrators) that build upon this platform with the goal to showcase the platform’s potential, but also to lead towards a commercial exploitation of the project outcomes. The first is a tool to help journalists when writing an article or preparing a report, while the second is aimed at viewers of TV or streaming content, providing background information and entertainment.They are currently running and publicly accessible online, incorporating hundreds of thousands of videos and news articles, as well as almost 10 million tweets.