Banca de DEFESA: MARINA TATIANE RIBEIRO DA SILVA

Uma banca de DEFESA de DOUTORADO foi cadastrada pelo programa.
STUDENT : MARINA TATIANE RIBEIRO DA SILVA
DATE: 17/08/2026
TIME: 08:30
LOCAL: ICe, Sala 1
TITLE:

Computational Analysis of Language and Cognition: Complementary Approaches from Graph Theory to Explainable Artificial Intelligence

KEY WORDS:

Natural Language Processing; Graph Theory; Explainable Artificial Intelligence; Transformer-based Language Models; Language-based Biomarkers; Human Cognition; Spontaneous Language; Reading; Psychosis; Alzheimer's Disease.

PAGES: 120
BIG AREA: Ciências Biológicas
AREA: Biologia Geral
SUMMARY:

Spontaneous language provides a rich and multidimensional source of information about human cognition, reflecting the interplay of linguistic, memory, executive, emotional, and social processes. Consequently, advances in Natural Language Processing (NLP), Graph Theory, and Explainable Artificial Intelligence (XAI) have enabled the extraction of objective and interpretable language-based biomarkers across educational and clinical contexts. This thesis investigates how complementary computational approaches characterize distinct dimensions of cognition through spontaneous language. Structural representations of discourse, psycholinguistic features, and transformer-based language models are explored, demonstrating how each approach captures complementary information about cognitive development, learning, and neuropsychiatric and neurodegenerative conditions. These approaches are evaluated in studies of language development, reading fluency, psychosis risk, and Alzheimer's disease, as well as their generalizability across different languages and spontaneous speech elicitation tasks. The findings demonstrate that complementary computational representations capture distinct dimensions of cognition expressed through language. By combining interpretable language-based biomarkers with simple, low-cost, and widely accessible spontaneous speech elicitation protocols, this thesis highlights the potential of computational language analysis as a scalable approach to cognitive assessment. The ease of acquiring these data supports their large-scale application, contributing to screening, longitudinal monitoring, and decision support strategies in educational, clinical, and public health settings.

COMMITTEE MEMBERS:
Externa à Instituição - ALINE VILLAVICENCIO - UFE
Interno - ***.705.618-** - ANDRÉ SALLES CUNHA PERES - UC
Presidente - 2276280 - CESAR RENNO COSTA
Externo à Instituição - MARCO IDIART - UFRGS
Externo à Instituição - RODRIGO WILKENS - UFE