Home AI News AI System Identifies Social Norm Violations Using Zero-shot Text Classification

AI System Identifies Social Norm Violations Using Zero-shot Text Classification

0
AI System Identifies Social Norm Violations Using Zero-shot Text Classification

The APA Dictionary of Psychology provides a comprehensive definition of social norms as socially determined standards that indicate typical and appropriate behaviors within a specific social context. These norms may either be universal, applying broadly across cultures, or contextual, specific to particular cultural settings.

Understanding Social Norm Violations

Social norms may vary across cultures and contexts, but violations of these norms can often be categorized into a few common themes that transcend cultural boundaries.

An AI System for Detecting Social Norm Violations

Researchers at Ben-Gurion University of the Negev have developed an AI system capable of detecting social norm violations. This study aims to combine social sciences and data science to gain deeper insights into human behavior and societal dynamics.

The researchers built this system using Zero-shot text classification, GPT-3, and automatic rule discovery. They trained the system to detect ten social emotions, including competence, politeness, trust, discipline, caring, agreeableness, success, conformity, decency, and loyalty. The system can classify a given text into one of these emotions and further categorize them as positive or negative.

Achieving Impressive Results

The researchers utilized the EmpatheticDialogues dataset to train the models, resulting in a 64% match between the system’s classification and human subjects’ emotions. The system achieved an accuracy of approximately 94% and a precision of about 96% in detecting norm violations.

The researchers believe that this is just the beginning and that their approach can be scaled up to include more social norms.

To learn more about this research, you can read the paper here and the reference article here.


Source link

LEAVE A REPLY

Please enter your comment!
Please enter your name here