Library of Alexandria
A private research corpus for studying mythology, folklore, religion, and cross-cultural primary texts. The real work is provenance and curation, not accumulating the largest folder.
The question
Can a local research system reason from primary and carefully edited sources instead of shallow summaries? The project began as a collection problem and became a data-quality problem.
Curation before scale
Early versions contained duplicates, mislabeled material, weak metadata, and sources that did not belong. The important milestone was not making the corpus larger. It was removing untrustworthy material and recording enough provenance to explain what remained.
Responsible sourcing
Collection respects source terms and explicit opt-outs. Material is checked for identity and provenance before it becomes part of the research set. A download succeeding is not evidence that the content is correct or appropriate to retain.
What it supports
The corpus supports private retrieval and model experiments where answers need to remain connected to source material. The public proof is the curation method and the lessons learned. The corpus contents, collection manifests, storage layout, and training recipes remain private.