22set2026
14:00 Master's Defense room 85 of IC2
Topic on
AI-society relationship as represented in the media: Multidisciplinary construction and analysis of named entity networks identified in news reports.
Student
Pedro Henrique de Carvalho Godoy
Advisor / Teacher
Ruben Interian Kovaliova
Brief summary
Artificial Intelligence (AI) is a cross-cutting technology with a profound social impact, considered a General Purpose Technology (GPT). This research is primarily motivated by the need to investigate the relationship between AI and contemporary society, focusing on the representation of this dynamic in the news. The objective is to create, analyze, and characterize the network structure of named entities present in Brazilian news articles about AI, identifying and examining the most relevant entities and groups in this network, in order to understand how this technology is perceived and publicly discussed in different journalistic media. The analysis focused on companies, governments, people, places, and other entities identified in the headlines and texts of the publications, reflecting the debate about AI in society. We used Named Entity Recognition and Relationship Extraction in tens of thousands of news articles related to AI in Brazil as a representative means of the AI-society relationship. The network formed by these entities and their relationships was examined using graph theory and complex network theory. Additionally, techniques of situational analysis and other sociological concepts were used to analyze structural and conjunctural socio-historical factors, correlations of forces, and trends. In total, 31.995 news articles published between 1994 and 2026 were collected. A directed, weighted, and multilayered knowledge graph composed of 47.027 nodes and 110.624 edges was constructed. The analysis revealed that a few entities concentrate a large volume of connections, while most have a low degree, suggesting a scale-free network and the existence of the cumulative advantage effect. The most frequent relationships showed a strong presence of themes associated with geopolitics, public institutions, large companies, and regulation. The results demonstrate the potential of integrating natural language processing, complex network analysis, and sociological frameworks to understand the structure of the public debate on AI, highlighting how it is perceived and discussed publicly.
Examination Board
Headlines:
| Ruben Interian Kovaliova | IC / UNICAMP |
| André Santanché | IC / UNICAMP |
| Thomas Kauê Dal'Maso Peron | ICMC / USP |
Substitutes:
| Sandra Eliza Fontes de Avila | IC / UNICAMP |
| Renato Moraes Silva | ICMC / USP |