Skip to main content

Modelling Politeness in Social Signal Processing

European Union Framework 7.

Professor Hastings Donnan, Professor Roddy Cowie (Psychology), Professor Ellen Douglas-Cowie (English), Paul Brunet.

Social Signal Processing Network (SSPNet) is an EU-funded project which aims to develop automatic detection and interpretation of signals that humans use naturally in social interactions.

This interdisciplinary and collaborative project includes Queen’s University Belfast, Imperial College London, CNRS, Idiap Research Institute (Switzerland), the German Research Centre for Artificial Intelligence (DFKI), Twente University, Delft University of Technology, University of Edinburgh, University of Geneva and University of Roma Tre.  The team at Queen’s aims to provide a broad conceptual framework for the project as a whole and an analysis of politeness as a case study.

The exploration of how humans react to the world and interact with it and each other remains one of the greatest scientific challenges.  The range of abilities humans employ to negotiate their social and cultural worlds includes the capacity to express and recognise social signals such as agreement, politeness and empathy.  Despite recent advances in the machine synthesis of some social signals (e.g. gaze exchange, blinks and smiles), machine analysis and synthesis of human social signals such as empathy, politeness and flirting remain tentative and fragmented.

The goal of SSPNet is to integrate knowledge and available resources in Social Signal Processing research domains including conceptual modelling and machine detection, interpretation and synthesis of social behaviour.  SSPNet uses a multimodal approach aimed at: (i) interpreting information for better understanding of human social signals and implicit intentions; and (ii) generating socially proficient behaviour of embodied conversational agents. How can we model, represent and employ human social signals and behaviours to design autonomous systems able to understand and respond to human communicative signals and behaviour?