Anchor

A joint research team from KAIST and Germany conducted a comprehensive investigation of over 100 million comments on Naver News, revealing that there are tens of thousands of accounts suspected of manipulating public opinion linked to foreign entities. Evidence of inciting social conflict was also detected, regardless of whether the commentators were progressive or conservative.

TJB reporter Cho Hyung-jun reports.

Reporter

Naver, Korea's largest news portal, is estimated to receive millions of comments daily.

An international joint research team from KAIST and the Max Planck Institute in Germany conducted a comprehensive analysis of approximately 110 million comments written over the 20-year period from 2006 to last year.

The purpose was to identify activities by specific countries or groups influencing public opinion through online news comments.

The research team used an AI model they developed in-house, and the analysis method is as follows:

First, 70 foreign-linked accounts identified by the Korea Institute for National Security Strategy served as the starting point.

The research team used AI to track accounts that were directly linked to these individuals or that repeatedly posted identical or similar comments on the same articles.

The primary targets were comments that could fuel social discord, such as those using awkward Korean speech patterns or those containing blind criticism or praise of specific countries or figures.

As a result, out of 4 million Naver News users, approximately 24,000 accounts were identified as being suspected of engaging in influence activities linked to foreign entities.

[Lee Won-jae / Professor, Graduate School of Culture Technology, KAIST: We applied 24 very conservative criteria, such as whether the same number of comments were posted at the same time or whether content was copied and pasted...]

The characteristics of the comments written by these accounts were also revealed; first, there was a much higher proportion of "criticism" than "praise."

Furthermore, it was confirmed that these "criticism" comments were directed broadly at various political forces, regardless of whether they were progressive or conservative.

Analysis also revealed that during election periods, such as presidential elections where political conflicts are intense, the number of newly active suspected malicious accounts increased by approximately 51% compared to normal times.

[Lee Won-jae / Professor, Graduate School of Culture Technology, KAIST: We were able to observe a phenomenon where peaks appear very consistently during periods such as national or local elections.]

The research team expects that the results of this data analysis will be helpful in detecting organized and sophisticated online activities during times of heightened social conflict.

(Video coverage: Choi Woon-ki, TJB; CG: Park Bo-young, TJB; Footage provided by: KAIST)

TJB Jo Hyung-jun

#SBSNews #News #Realtime #Naver #Comments #PublicOpinion #Manipulation #Criticism #Conflict #Politics #AI #ArtificialIntelligence

Copyright Ⓒ SBS. All rights reserved. Unauthorized reproduction, redistribution, and use for AI learning are prohibited.