QiBERT -- Classifying Online Conversations Messages with BERT as a Feature

TL;DR

This paper presents QiBERT, a BERT-based model that classifies short online conversation messages in Portuguese schools with high accuracy, aiding social scientists in understanding communication and engagement.

Contribution

It introduces a novel application of SBERT embeddings with supervised learning for classifying online debate messages in Portuguese, achieving over 95% accuracy.

Findings

Achieved over 0.95 accuracy in classifying messages

Demonstrated effectiveness of SBERT embeddings in short text classification

Enabled better analysis of online student engagement

Abstract

Recent developments in online communication and their usage in everyday life have caused an explosion in the amount of a new genre of text data, short text. Thus, the need to classify this type of text based on its content has a significant implication in many areas. Online debates are no exception, once these provide access to information about opinions, positions and preferences of its users. This paper aims to use data obtained from online social conversations in Portuguese schools (short text) to observe behavioural trends and to see if students remain engaged in the discussion when stimulated. This project used the state of the art (SoA) Machine Learning (ML) algorithms and methods, through BERT based models to classify if utterances are in or out of the debate subject. Using SBERT embeddings as a feature, with supervised learning, the proposed model achieved results above 0.95 average accuracy for classifying online messages. Such improvements can help social scientists better understand human communication, behaviour, discussion and persuasion.