Read the original document by opening this link in a new tab.
Table of Contents
1. Introduction
2. Background
2.1 Conversational AI System
2.2 Social Bias in Conversational AI Systems
3. Approach and Implementation
3.1 Social Bias Dataset Construction
3.2 BiasAsker Framework
4. Biased Answer Collection
5. Bias Measurement
Summary
The document discusses the importance of measuring bias in conversational AI systems and introduces BiasAsker, a framework designed to identify and measure social bias. It explains the construction of a social bias dataset containing social groups and biased properties. BiasAsker generates diverse questions and conducts measurements to quantify bias in conversational AI systems.