Who is involved in the SELMA project?
Five partners form the consortium: Deutsche Welle, Avignon University – Laboratoire Informatique d’Avignon (LIA), University of Latvia – Institute of Mathematics and Computer Science (IMCS), Priberam and Fraunhofer-Gesellschaft (FhG).
Deutsche Welle is a public service broadcaster, reporting on events and developments in Germany and abroad and does so in 30 languages.
DW is the project coordinator and responsible for overall project management. DW leads the workpackages on Communication, Evaluation and Ethics and plays a leading role in the work on Data Collection and User Requirements.
Priberam is a Portuguese SME founded in 1989, as a spin-off from Instituto Superior Técnico in Lisbon, which offers cutting-edge semantic search and natural language processing technologies. Priberam licenses its technologies to clients such as Amazon and Microsoft and the biggest media publishers in Portugal, Brazil and Spain. Priberam Labs, the company’s research department, is focused on innovative technologies such as automatic media monitoring, recommendation, and social media analysis, with a strong component on machine learning research for Big Data.
Priberam will lead WP1, which describes the use cases and develops the prototypes. Priberam will be working on Natural Language Understanding tasks in WP2 and summarization leveraging on its experience from previous projects in the fields of machine learning and NLP.
The Fraunhofer-Gesellschaft, headquartered in Germany, is one of the leading applied research organizations. With its focus on developing key technologies for the future, Fraunhofer plays a central role in the innovation process with operating 75 institutes. As part of it, the Fraunhofer Institute for Intelligent Analysis and Information Systems IAIS is engaging in the fields of Artificial Intelligence, Machine Learning, and Big Data. With more than 300 employees, the Fraunhofer IAIS supports companies in the optimization of products, services, processes, and structures as well as in the development of new digital business models.
Fraunhofer IAIS leads WP2 enabling the system to learn automatically from a large-volume live multilingual stream of documents, continuously incorporating knowledge into the language models.
LIA is a public computer science research laboratory created in 1987, affiliated to Avignon Université (University of Avignon), in France. The LIA team that works on speech and language processing regularly participates in international evaluation campaigns, showing its mastering of such technologies by reaching top rankings.
LIA leads WP3 on Joint Multilingual and User-Feedback Transfer Learning and focuses on speech and natural language processing, including tasks such as transcription, punctuation, translation and voice synthesis.
The Institute of Mathematics and Computer Science of the University of Latvia (IMCS UL) was founded on 11 November 1959. AI Lab at IMCS is the leading NLP research lab in Latvia, focusing on natural language understanding (NLU) and generation (NLG) via abstract meaning representations, including deep neural networks.
IMCS leads WP4 Platform Integration and focuses on the architecture of the platform, including performance, scalability and continuous integration of user feedback. IMCS leads the Data Management task.