Participating Organizations

• Faculty of Electrical Engineering and Computer Science (FERI), University of Maribor

Problem Description

The key research activities for the HUMANIPA program are focused on understanding and modeling of colloquial language in human-to-human interactions. In the field of natural language processing and comprehension gesticulation and nonverbal communication play an important role in human-machine interaction, as well as the ability to express information not only through the use words.

It has become clear that the inclusion of such signals is a key guideline in determining the more personal aspect of user inputs and how device responses can be presented to people. In direct interaction nonverbal signals along with spoken content (or even its absence) are key to establishing cohesion in discourse.

Solution

As part of the HUMANIPA research project, we are developing a new concept of understanding conversational intelligence CLU (Conversational language understanding) which is a new and unique approach that further develops the idea that verbal and nonverbal conversational signals are complementary and equally important in colloquial expressions. Moreover, within the project, we are developing new models based on fusion as well as algorithms based on AI, which will be able to create in-depth understanding in cognitive interaction with the communicative purpose as the central core in human interaction. These algorithms will be based on deep learning techniques and will use comprehensive conversational signals, which are also contained in the EVA-Corpus.