The TEA Nets framework combines AI and cognitive network science to model targets, events and actors in text

Researchers introduce TEA Nets (Target-Event-Agent Networks), an open-source AI framework that extracts subjects, verbs, and objects from text to analyze emotional and semantic patterns. Testing across conspiracy narratives and psychotherapy transcripts reveals that highly conspiratorial texts link personal pronouns to actions twice as frequently as low-conspiracy texts, while LLMs express emotions with measurably lower intensity than humans.

TEA Nets represents a meaningful advancement in natural language processing by combining cognitive network science with AI to extract and analyze narrative structure at scale. The framework addresses a critical gap in interpretable AI—moving beyond black-box sentiment analysis to reveal how linguistic patterns correlate with specific phenomena like conspiracy thinking and emotional expression. This matters because understanding the computational markers of conspiracy narratives could inform content moderation, misinformation detection, and psychological research.

The research emerges amid growing interest in explainable AI and network-based language analysis. As large language models proliferate, researchers increasingly scrutinize how these systems differ from human communication patterns. The LOCO conspiracy corpus findings—that high-conspiracy texts emphasize personal pronouns and anger-inducing actions—provide quantifiable linguistic signatures that could enhance automated detection systems. Similarly, the psychotherapy findings revealing that Claude 3 Haiku and GPT-3.5 express sadness with lower emotional intensity than humans have practical implications for AI-assisted mental health training.

For the broader AI and NLP community, TEA Nets as an open-source library democratizes access to cognitive network analysis tools. This could accelerate research in computational linguistics, misinformation studies, and human-AI interaction analysis. The framework's ability to perform interpretable analysis addresses recurring critiques that AI systems lack transparency, potentially building trust in AI applications across sensitive domains like mental health and content moderation.

Looking forward, developers should monitor whether TEA Nets gains adoption in commercial content moderation platforms and whether similar cognitive network approaches emerge for other languages and domains.

• →TEA Nets is an open-source framework combining cognitive network science and AI to extract emotional and semantic patterns from text.
• →Highly conspiratorial narratives link personal pronouns to actions twice as frequently as low-conspiracy texts, offering quantifiable linguistic markers for detection.
• →LLMs express emotions with measurably lower intensity than humans in psychotherapy contexts, raising implications for AI-assisted mental health training.
• →The framework enables interpretable emotion detection and semantic analysis, addressing the transparency gap in AI language processing.
• →The research demonstrates practical applications for misinformation detection, content moderation, and understanding human-AI communication differences.