Integration and evaluation of social competences such as humor in an artificial interactive agent

Abstract

In this paper, we present a brief overview of our ongoing work about artificial interactive agents and their adaptation to users. Several possibilities to introduce humorous productions in a spoken dialog system are investigated in order to enhance naturalness during social interactions between the agent and the user. We finally describe our plan on how neuroscience will help to better evaluate the proposed systems, both objectively and subjectively. CCS CONCEPTS • Human-centered computing → Sound-based input / output; • Computing methodologies → Neural networks; Adversarial learning;

Findings from Experiments of On-line Joint Reinforcement Learning of Semantic Parser and Dialogue Manager with real Users

“…It can be implemented with sample-efficient learning algorithms (Daubigney et al 2012) and can involve on-line learning through direct interactions with users Lefèvre 2013bc 2015). More recently, on-line learning has been generalised to the input/output modules, SP and NLG, with protocols to control the cost of such operations during this human-in-the loop system development (as in (Ferreira et al 2016Riou et al 2017)). The work presented here is a first attempt to combine the on-line learning of SP and DM in a single phase of development.…”