Emotion Twenty Questions Dialog System for Lexical Emotional Intelligence

TL;DR

This paper introduces a web-based demo of EMO20Q, a dialog game for studying emotion description and developing AI agents capable of engaging in open-ended emotional dialogue, with recent advances in transformer models enhancing the agent's response capabilities.

Contribution

It presents recent developments in the question-answering role of EMO20Q using transformer-based models and details the system design for collecting pilot data.

Findings

Enhanced transformer-based agent for open-ended responses

System design for data collection and future research

Demo availability for pilot data collection

Abstract

This paper presents a web-based demonstration of Emotion Twenty Questions (EMO20Q), a dialog game whose purpose is to study how people describe emotions. EMO20Q can also be used to develop artificially intelligent dialog agents that can play the game. In previous work, an EMO20Q agent used a sequential Bayesian machine learning model and could play the question-asking role. Newer transformer-based neural machine learning models have made it possible to develop an agent for the question-answering role. This demo paper describes the recent developments in the question-answering role of the EMO20Q game, which requires the agent to respond to more open-ended inputs. Furthermore, we also describe the design of the system, including the web-based front-end, agent architecture and programming, and updates to earlier software used. The demo system will be available to collect pilot data during the ACII conference and this data will be used to inform future experiments and system design.