The CIMeC Language, Interaction and Computation group (CLIC) has been developing computational models of lexical semantics based on the distribution of words in parsed and unparsed natural language corpora. A team partly funded by the ERC grant "COMPOSES" has extended these computer techniques to model the meanings of derivationally complex words, phrases and sentence, recognize (non-)sensical word pairs, develop multi-modal (text+images) meaning representations, map meanings across languages and develop large datasets to test meaning similarities at the sentence level. Moreover, we are investigating concept acquisition from large amounts of multimodal data (e.g. language and vision) and on the computational modeling of emotions and their impact on communication and decision making. We also develop computational models of the semantic interpretation process, in particular of anaphora resolution, that is, of how speakers can keep track of which elements of a sentence refer to previously mentioned things.
PIs working in this area
- Language and Vision Group - LaVi | P.I.Raffaella Bernardi
- Reasoning and Decision Group - RAD | P.I.Katya Tentori
- Meaning and Computational Group | P.I. Jakub Szymanik
- DEep Neural ArtificiaL Intelligence group (DENALI) | P.I. Paolo Rota
- Human-Aware Machine Intelligence - HAMI | P.I. Stefano Teso