Explore vs Research
Explore is the default mode. It routes each query dynamically across the full catalog, selects the best matching source, executes, and returns results. It is discovery-first and open-ended — you do not need to know which pack answers your question.
Research mode narrows the scope to a corpus you loaded. The system reads from that bounded set only. If the data you need is not in the corpus, Research names the gap instead of routing around it.
| Explore | Research | |
|---|---|---|
| Routing | Full catalog, dynamic | Corpus only, bounded |
| State | Stateless per query | Corpus manifest loaded |
| Missing data | Routes to best available source | Names the gap explicitly |
| Reproducibility | Varies with catalog updates | Corpus-scoped, stable |
| Setup | None | Build corpus on account page |
Corpus setup
Corpus management lives on the account page, not inside the map shell. Select published packs, name the corpus, and save it. Load the corpus in the app and Research mode activates over that specific data. You can save multiple named corpora and switch between them.
What corpus-bound means in practice
Research answers are scoped to the manifest. Cross-domain queries work as long as both sources are in the corpus. A query that requires data outside the corpus gets an explicit gap response — not a silent fallback to something adjacent.
This makes Research mode better suited for systematic analysis, academic work, and reproducible reporting where you need to control exactly what data is in scope.
Export
Query results export to CSV and Parquet. Both formats are compatible with R, Python, Stata, and QGIS. Pack metadata, source attribution, and coverage scope travel with the export for citations and reproducibility.
Bring your own data
The pack schema is public. Your own datasets normalize to the same
loc_id schema and work alongside maintained packs in the
same Research workspace. See the
pack schema guide for implementation details.