Research Field: Computational Linguistics
Content of Research:
The system is based on a full-fledged NLP pipeline that goes from tokens to the underlying semantics. Current technologies are using machine learning to create a chat-bot automatically but in fact their success rate is very low and not at all guaranteed due to the inability of machine learning to cope with fine-grained semantic analysis. Features like negation, comparatives and conditionals are out of the scope of any current pre-trained language model. Our system needs only to upgrade the basic vocabulary related to the domain. The grammar feeding the semantics, currently implemented only covers two languages, English and Italian, but more languages could be added if needed while the semantics will remain the same.
It is a tool for Customer Relation Management based on deep semantics and text understanding and is greatly beneficial for large and small industries, business companies, improving their Customer Service. It can work at the same time with multiple customers (multitasking) and at any time of the day.