Exploring Spatio-Temporal Networks: Analyzing Brain Traffic and Construction of Geographic Networks

This thesis integrates the analysis of traffic in brain networks and the construction of geographic networks, offering novel insights into spatio-temporal behavior across both biological and geographic systems. The first part of the thesis examines Traffic Analysis in Human Brain Networks, exploring the pathways through which brain signals flow by combining both structural and dynamic perspectives. A primary question in this study is whether brain signal flow aligns more with a globally optimized System Optimum (SO) model or a decentralized User Equilibrium (UE) model. Findings indicate a closer alignment with the SO model, highlighting an inherent efficiency in signal routing within the brain. Building on this, the research investigates how brain coordination is impacted by declining levels of consciousness, providing insights into neural activity under unconscious states, such as during sleep or anesthesia. Additionally, an alternative model is introduced in which brain signals do not have predefined destinations but instead search for specific resources within the network. By employing Markov Decision Process (MDP)-based algorithms, simulations show that this model more accurately mimics brain signal propagation, challenging the assumption that brain signals always have clear, fixed destinations. The second part of the thesis focuses on the extraction of structured geographic data from unstructured text using conversational large language models (LLMs). The study applies these models to convert unstructured sources into structured data, facilitating the construction of a geographic network. This network provides insights into travel patterns, logistical demands, and key locations, with the potential for future analysis of congestion and destination dynamics. Through these investigations, this thesis contributes to a deeper understanding of complex network behavior in both the biological and geographic domains, offering new methods for analyzing traffic patterns, resource-seeking behaviors, and destination-based analyses across diverse systems.