Algorithmic Scripts and Textual Corpus for the Topological Modelling of Depressive Discourse via Discrete Network Analysis

<p><strong>Overview of the Dataset and Codebase</strong></p> <p>This repository hosts the supplementary data and Python scripts utilised for the structural examination of patient-generated text regarding depression. The primary aim of these files is to facilitate the reproducibility of the network topology analysis and community detection processes detailed in the associated manuscript.</p> <p><strong>Contents</strong></p> <ul> <li><strong>Raw Text Data:</strong> An anonymised collection of linguistic tokens derived from online health communities.</li> <li><strong>Processing Scripts (Python):</strong> Custom-built algorithms designed for natural language processing (NLP).</li></ul> <p><strong>Methodological Specifications</strong></p> <p>The accompanying code executes a discrete mathematical analysis focusing on the co-occurrence of lexical units. Key computational procedures include:</p> <ol> <li><strong>Pre-processing:</strong> Cleaning of textual noise and removal of domain-specific stop-words to isolate core vocabulary.</li> <li><strong>Network Construction:</strong> Creation of an undirected graph where nodes represent words and edges signify proximal co-occurrence within a specific window size.</li> <li><strong>Community Detection:</strong> Application of the Louvain method to identify modularity and semantic clusters.</li> <li><strong>Triad Analysis:</strong> A specialised extraction of three-node motifs to inspect the interplay between distinct domains: <ul> <li><em>ECB Model:</em> Emotion, Cognition, and Behaviour.</li> <li><em>BOS Model:</em> Body, Organ, and Symptom (Psychosomatic lexicon).</li> </ul> </li></ol> <p><strong>Technical Requirements</strong></p> <p>The scripts are optimised for Python 3 environments and necessitate standard libraries such as NLTK, NetworkX, and Matplotlib for the visualisation of graph metrics.</p>