DAME: a platform for Deep learning Approaches for Multilingual tExt processing

Natural language understanding is one of the ultimate goals of ICT as it is needed in any natural communication with machines. In our recent research projects, such as in the EU FP7 FET project MUSE, we have focused our efforts on deep learning models for language understanding. Our demonstrators showed slightly better results compared to methods that employ a traditional pipeline of linguistic processing. This is an extremely important step forward for the exploitation of NLP technology at a European and international scale, as linguistic processing is often a bottleneck when porting models for human language understanding to different languages. DAME will enable us to create a generic platform that will include these innovative algorithms, together with example demonstrators and applications, training data, etc. It will allow us to exploit our work through two routes: (i) bilateral contracts with industry for tailoring this generic platform to their needs; and (ii) acquiring funding through H2020 calls such as ICT-20-2017 for further extending this platform. The second route will allow us to pursue even more bilateral contracts. As such, DAME will allow us to create a self-sustaining and highly innovative platform that will drive our research for the next decade.

Keywords:text processing

Disciplines:Applied mathematics in specific fields, Computer architecture and networks, Distributed computing, Information sciences, Information systems, Programming languages, Scientific computing, Theoretical computer science, Visual computing, Other information and computing sciences