# Week 5 inputs and outputs

The unchanged official archive is stored at `data/marvel_pages.zip`, downloaded from [the course data page](https://sunelehmann.com/socialgraphs2026-web/data/) on 5 October 2026. It contains 303 plain-text Wikipedia articles from the frozen 26 August 2026 crawl, plus the course's README. Filenames are URL-decoded to join against `data/week1/week1_nodes.tsv`. The archive is read in place and is never modified by the analysis.

Archive SHA-256: `36023ca077053b3df5bd55630754cb514c3f9dbecbd70afcb2283e5bcc0017ee`.

`results.json` saves the fit, descriptive correlations, case studies, all 303 page measurements, per-article hashes, input hashes and software versions. `page_lengths.csv` makes the same per-page measurements available as a table. `text_checks.json` records the four passages and sections inspected for the post, along with the interpretation and its limits.

Attribution: the source Wikipedia articles and their URLs are listed in the original node roster and the output table. The short quotations in the post come from the frozen course archive, rather than today's live articles. The original archive retains the course README and its full source texts. Wikipedia contributors are the authors of those articles; the course provides the snapshot. See [Wikipedia's reuse terms](https://en.wikipedia.org/wiki/Wikipedia:Reusing_Wikipedia_content) for attribution and licensing of the source text. The calculations and interpretive notes are the group's analysis with AI assistance.
