Common-sense and world knowledge from knowledge bases (KBs) and language resources have always been a key ingredient in NLP and play a central role in NLG, supporting ML techniques that require expansion, filtering, disambiguation or user adaptation of the generated content.
One key aspect of research concerns how to integrate existing language resources and knowledge bases into NN models for NLG – e.g. how to process the relevant content, taking into account different data formats-. Analysing, reviewing and comparing methods about the use of structured KBs into LG models is a prerequisite for improving the output of these models, raising awareness about resources available and possible uses of them.
Looking at existing resources with a focus on NLG means also considering how to expand them with multilingual and multi-modal content. An interesting challenge concerns the study of methodologies to make KBs multi-modal through automatic mapping, crowdsourcing or a combination of both.
An expected result of WG4 is to increase the varieties of knowledge resources and language resources used in NLG.
WG4, with a focus on the exploitation of large knowledge bases and graphs, will analyse how to efficiently integrate these (multi-modal) KBs, taking into account theoretical models of semantics and semantic processing that can accommodate linguistic and perceptual information.
If you’d like to join WG4, please email Dr Irene Russo at: irene.russo(@)ilc.cnr.it