build-text-network

Build a Text Network

Build and tune a word co-occurrence graph whose vocabulary reflects the corpus rather than stopwords or collection artifacts. A co-occurrence edge represents proximity in text, not semantic truth.

Follow the gephi skill's rules throughout; this workflow repeats only the ones it is most likely to break. Read ../gephi/references/text-network-analysis.md before choosing construction parameters and ../gephi/references/layout-guide.md before laying out the result.

Rules this workflow must keep

• Asking. When a step says to ask, ask once. If the person cannot answer, is away, or asked for a finished product, use the default named in that step and remove or overwrite nothing. At the end, give one short line for each choice that changed the result, under "Choices I made"; leave the section out when every step used its obvious default. If there is no input to work on, stop and say what is needed.
• Work in a new workspace. Call gephi_new_workspace before the first build, so the person's open work stays as it was, and rebuild there with clear_existing: true. Never create, open, or save a project over their work.
• Stability. Run gephi_community_stability before naming, captioning, or colouring by groups, and say how stable they are.
• Caption and legend. Every export ships with a copy-ready caption (data, layout and settings, what size and colour encode, what the map does and does not show) and, when colour encodes groups, a legend (gephi_export_legend).

Construction Choices

Use gephi_text_to_network. Record the final values and why they were chosen:

• text: prefer a list when the corpus has natural segments; the window resets per item and avoids cross-document edges;
• window_size: smaller captures tighter pairings, larger captures looser themes;
• extra_stopwords: remove corpus-specific filler, names, and boilerplate;
• pos_filter: use nouns or proper nouns for a concept-focused map when suitable;
• min_word_frequency and min_edge_weight: raise these to remove rare noise;
• merge_phrases: combine useful repeated two-word phrases;
• exclude_self_referential and self_referential_threshold: remove words that occur in nearly every document;
• context_snippets: preserve examples needed for later interpretation;
• clear_existing: false (the default) adds to the current workspace; pass true on every rebuild.