一句话：基于4.4万条Reddit帖子，结合LLM、聚类与时间分析，揭示公众对科技公司政治化的主题、情感与演变。
论文关注科技公司日益塑造公共话语并被认为具有政治影响力所引发的公众担忧。为弥补大规模实证分析的不足，作者收集了超过44,000条来自政治类和技术类subreddit的Reddit帖子，综合使用大语言模型、无监督聚类和时间分析，识别与科技公司政治化感知相关的帖子，进行情感分类，提取主要主题类别，并考察讨论随时间的变化。结果显示，用户最常讨论高知名度高管，同时关注审查、政治偏见和影响力等问题；整体情感常为中性，但在审查和LGBTQ+权利等争议社会议题上负面语气尤为明显。论文贡献了对这些讨论的结构化概览，并反思了理解科技行业政治影响归因的启示。
As technology companies increasingly shape public discourse, their perceived political influence has become a growing source of concern. While anecdotal accounts and media coverage suggest that these concerns are widespread, there remains limited large-scale empirical analysis of how the perceived politicization of tech companies is discussed by the public. This work analyzes over 44,000 Reddit posts from politically and technologically oriented subreddits. We use a combination of large language models, unsupervised clustering, and temporal analysis to identify posts relevant to the perceived politicization of tech companies, classify their sentiment, uncover prevailing thematic categories, and examine how these discussions evolved over time. Our findings show that users most frequently discussed high-profile executives, alongside broader concerns about censorship, political bias, and influence. While sentiment was often neutral, negative tone was particularly pronounced in discussions around censorship and contentious social issues, such as LGBTQ+ rights. We contribute a structured overview of these discussions and reflect on their implications for understanding how political influence is attributed to the tech sector.
@inproceedings{paperbot11967,
title = {From Big Tech to Big Politics: Users' Discourse on the Politicization of Technology Companies},
author = {Amelia Kalecińska and Panagiotis Fotopoulos and Emitzá Guzmán},
booktitle = {ICSE},
year = {2026},
doi = {10.1145/3786581.3786929}
}