CNN-BERT for measuring agreement between argument in online discussion

• DOI: https://doi.org/10.1108/IJWIS-12-2021-0141
• Published date: 26 August 2022
• Date: 26 August 2022
• Pages: 356-368
• Subject Matter: Information & knowledge management,Information & communications technology,Information systems,Library & information science,Information behaviour & retrieval,Metadata,Internet
• Author: William Harly,Abba Suganda Girsang

CNN-BERT for measuring

agreement between argument

in online discussion

William Harly and Abba Suganda Girsang

Computer Science Department, BINUS Graduate Program –

Master of Computer Science, Bina Nusantara University, Jakarta, Indonesia

Abstract

Purpose –With the rise of online discussion and argument mining, methods that are able to analyze

arguments become increasingly important. A recent study proposed the usage of agreement between

arguments to represent both stance polarity and intensity, two important aspects in analyzing arguments.

However, this study primarilyfocused on finetuning bidirectional encoder representations from transformer

(BERT) model. The purpose of this paper is to propose convolutional neural network (CNN)-BERT

architectureto improve the previous method.

Design/methodology/approach –The used CNN-BERT architecture in this paper directly uses the

generated hidden representation from BERT.This allows for better use of the pretrained BERT model and

makes finetuning the pretrained BERT model optional. The authors then compared the CNN-BERT

architecturewith the method proposed in the previous study (BERT and Siamese-BERT).

Findings –Experiment results demonstrate that the proposed CNN-BERT is able to achieve a 71.87%

accuracy in measuring agreement between arguments. Compared to the previous study that achieve an

accuracy of 68.58%, the CNN-BERT architecture was able to increase the accuracy by 3.29%. The CNN-

BERT architectureis also able to achieve a similar resulteven without further pretraining the BERTmodel.

Originality/value –The principal originality of this paper is the proposition of using CNN-BERT to better use

the pretrained BERT model for measuring agreement between arguments. The proposed method is able to improve

performance and also able to achieve a similar result without further training the BERT model. This allows

separation of the BERT model from the CNN classifier, which significantlyreduces the model size and allows the

usage of the same pretrained BERT model for other problems that also did not need to finetunetheir BERT model.

Keywords Online discussion, Stance polarity and intensity, Natural Language Processing,

Applications of Web mining and searching

Paper type Research paper

1. Introduction

There are many discussions that happen every day onlinedue to the ease of internet access

and the anonymity that it brings (Onyemaet al., 2019). These discussions cover a wide range

of topics and understanding the content of these discussions will give insight into the

current opinions of the general populace. Thus, a method that can analyze an online

discussion is becoming increasingly important to extract information. One of this

information is the relationship between arguments in the discussion. The relationship

between arguments in online discussion describes how arguments support or oppose an

existing stance. This relationshipis represented in the form of stance polarity (support and

oppose) and the intensity of the relationship that describes how strong does the argument

support or opposes the existing stance.

Although many studies have been carried out to determine the relationship between

arguments through classifying stance polarity,often the intensity of the relationship is not

IJWIS

18,5/6

356

Received14 December 2021

Revised19 April 2022

Accepted5 August 2022

InternationalJournal of Web

InformationSystems

Vol.18 No. 5/6, 2022

pp. 356-368

© Emerald Publishing Limited

1744-0084

DOI 10.1108/IJWIS-12-2021-0141

The current issue and full text archive of this journal is available on Emerald Insight at:

https://www.emerald.com/insight/1744-0084.htm