Semantic Categorization of Social Knowledge for Commonsense Question Answering

TL;DR

This paper introduces a method to categorize social knowledge semantics for commonsense question answering, demonstrating that incorporating these categories improves model performance with simpler models.

Contribution

It proposes a novel categorization of social knowledge semantics and integrates this into neural QA models, enhancing performance efficiently.

Findings

Models with social knowledge categories achieve comparable results to complex models.

Semantic categorization enables simpler models to perform well on social QA tasks.

Incorporating relation information from knowledge bases improves understanding.

Abstract

Large pre-trained language models (PLMs) have led to great success on various commonsense question answering (QA) tasks in an end-to-end fashion. However, little attention has been paid to what commonsense knowledge is needed to deeply characterize these QA tasks. In this work, we proposed to categorize the semantics needed for these tasks using the SocialIQA as an example. Building upon our labeled social knowledge categories dataset on top of SocialIQA, we further train neural QA models to incorporate such social knowledge categories and relation information from a knowledge base. Unlike previous work, we observe our models with semantic categorizations of social knowledge can achieve comparable performance with a relatively simple model and smaller size compared to other complex approaches.