Corpus-based analysis of the Behavioral Cues and Social Attitudes related to Persuasion

I2M, Aix-Marseille Université

Date(s) : 07/04/2022   iCal
16 h 00 min - 17 h 00 min

The perceived persuasion depends on specific behavioral signals such as the acoustic descriptors (fundamental frequency, pauses,…), head movements (position, speed and acceleration), or the facial expressions (variation of each point in facial image) but also on other perceived social attitudes, such as the dominance, confidence or expertise.
We propose in our project to explore persuasion by analysis the social attitudes and the behavioral cues related to persuasion usingmachine learning methods.
First, we consider two corpora : POM and POTUS to train, validate and test the learned prediction models.
Then, we explore different classifiers and clustering strategies to investigate the predictability of the social attitudes relatedto persuasion as well as the vocal and non-verbal cues associated to these attitudes.
Finally, we model the most important signals used to ensure an efficient prediction in order to integrate statistical rules in anartificial agent.

Participer à la réunion Zoom

ID de réunion : 857 5340 5035
Code secret : voir mail



Retour en haut 

Secured By miniOrange