SUMMA Monitoring Platform Prototype: Cross-Lingual Overview of DW Content in English

Deutsche Welle certainly offers a lot of interesting content in 30 languages. Too bad we can't understand most of it. Lucky there's project SUMMA which has made major progress in the last couple of months and now offers a tool to digest all the news in one lingua franca.

Prototype of integrated SUMMA platform operational

SUMMA stands for Scalable Understanding of Multilingual Media and develops a cross-lingual extensible media monitoring platform. The H2020 EU-funded research and innovation project, realized in close collaboration with the BBC and seven other partners, is now halfway. Latvian News Agency LETA has developed an integrated platform, ingesting primarily Deutsche Welle content and external live feeds collected by BBC Monitoring. The prototype offers an automated workflow to ingest data via API, transcribe audio, translate into English, analyze, cluster, and summarize.

Currently, it processes five source languages into one target language, i.e., English. At the time of writing, Arabic, German, Latvian, Russian, and Spanish are covered. Farsi, Ukrainian and Portuguese will be added in the coming year. This includes video and audio on demand, text articles and live feeds. The platform is still in a development phase, so output is far from perfect at this stage, but the workflow is in place and progress is continuous. User testing is ongoing at Deutsche Welle and BBC and leads to further recommendations and enhancement.

Cross-lingual clustering

SUMMA is all about media monitoring. For the Deutsche Welle internal use case this means monitoring our own content in different languages. The system clusters related stories (from various languages) and provides an overview of trending stories. This allows us to see at a glance what's trending and what has been covered by different language departments. From this overview, we can get summaries of the stories, and go down to the item level for a summary, transcription and translation of the item. All in English–or in the original language.

User Day at the BBC


Earlier this month, SUMMA had its first User Day, held at BBC Monitoring. We demonstrated the system to a group of about 30 visitors from BBC, VRT, Al Jazeera, Ericsson, and other interested companies and organizations. A panel discussion on potential exploitation of the system and poster sessions going deeper into the technologies showed the challenges of this promising project. Deutsche Welle will be hosting a User Day in 2018.

For more information on Summa, news and status updates please check the SUMMA website, SUMMA on Twitter or get in touch with our colleague Peggy van der Kreeft who is in charge of the project here at DW.

Logo Deutsche Welle
DW Innovation