Neural-symbolic approaches for NLG systems
AI programming can be used to produce text from data. This is called natural language generation (NLG), which transforms complex data into natural-sounding language – as if it were written by a human. The EU-funded NG-NLG project will explore the neural approaches to NLG, which currently remain confined to experimental use. The reason for this is that, despite the very natural outputs of recent neural systems, the behaviour of neural NLG systems is not transparent or reliable. The project will develop innovative approaches that combine neural approaches with explicit symbolic semantic representations, thus allowing greater control over the outputs and explicit logical inferences over the data. The project will test its approaches on data-to-text generation, summarisation and dialogue response generation.
Fields of science
- HORIZON.1.1 - European Research Council (ERC) Main Programme
Call for proposalSee other projects for this call
Funding SchemeHORIZON-AG - HORIZON Action Grant Budget-Based
116 36 Praha 1
See on map