Results: candidates, filters and charts
Choose what to inspect
Select Documentary, Organizational, Human mediation, or All categories. Each table shows the relevant unit: document endpoints; RG/WG and process clues; or an observed person/address with a linked base event or an exploratory document pair without a base. Consolidated review and My review are separate: one positive review does not confirm a case.
Select Inspect to see the hypothesis, discovery signals, gaps, paths, participants and located evidence. Technical payloads and provenance remain collapsed. The overview shows up to 50 automatic evidence records and eight distinct paths; complete support remains in technical details. Missing passages or dates are explicit. Reviews and annotations cannot be saved while the run is processing.
Start with How candidates are discovered for the flowchart and category rules. Use the review workflow for scientific decisions.
Filter once, inspect and export the same selection
The matching counter, review counters, candidate list, charts and CSV use the same category and filters. Filters include search, family, RG/WG, candidate topology, consolidated state and own-review progress. More filters adds certainty, signal topology, reason and automatic/manual origin.
- Candidate topology partitions cases into Direct only, Chain only, Direct and chain, or No topology recorded. Each case appears once, considering all its recorded signals.
- Family + signal topology + certainty must match the same signal. Family bars can overlap because a candidate may have multiple signals.
- RG/WG uses saved indexed metadata or explicit participants, never name prefixes alone. Unknown associations remain selectable; group association does not establish historical adoption.
- Base status, for Human mediation, distinguishes confirmed T1–T5, pending, disagreement, rejected, uncertain and missing eligible bases. Confirmed base means independent consolidated confirmation, not confirmed T6. Base corrections update the dependency. Missing bases are valid for exploratory leads and prevent T6 confirmation.
Category changes preserve selected filters. An incompatible retained filter can yield zero results; remove its chip or use Clear filters, which keeps the category.
Organizational context
Open Organizational → Evidence between groups for documentary support, charter mentions, group histories and acquisition gaps. Its own group filters and JSON export describe screening context, not reviewed candidate totals. Links open the associated organizational hypotheses.
Filter family ORG_AGENDA_ADOPTION for possible agenda adoption. Inspect the separate research-attribution and work-mandate excerpts. The detail page separates revision observation, revision publication and charter approval. Pending dates remain explicit; supported charter dates do not date research adoption or establish institutional succession.
An organizational inspection shows the preserved input snapshot that generated that hypothesis. The latest group view may be newer. Neither view confirms T5. See group evidence operations.
Independent author intersections
Select Open exploratory queue to view Human mediation with family HUM_AUTHOR_INTERSECTION and base No eligible base linked. The table shows the observed name/address and document endpoints. Exploratory: contribution not established distinguishes these leads from candidates with a linked event. One row is one address/document-pair lead; multiple addresses can belong to one person and multiple rows can describe the same pair.
The owner can select Discover independent author intersections on an idle run. This separate action indexes saved submissions in bounded batches and then searches dated affiliation-compatible document pairs. If the status is indexing, repeat the action until the saved pages have been processed. It requires a current analytical document snapshot and a verified relationship index; unavailable input is reported rather than treated as zero matches. No downloads are performed.
The latest coverage summary distinguishes document pairs from address-based leads. Pairs with documentary candidates or recorded direct relationships are separated from the independent queue. The absence of a recorded direct relationship does not rule out indirect paths or evidence missing from collection.
Open a lead to inspect authorship and historical affiliation evidence, integrity references and pending checks. Use Link or change the related event to associate a documentary or organizational candidate in this run with a justification. A linked lead leaves the missing base selection but remains available under the same family with that filter removed. Linking does not establish contribution, identity or mediation. See review.
Older leads can be retained outside the latest discovery snapshot; their detail page identifies that condition. Repeated discovery preserves IDs, investigator notes, human-selected links and reviews. Coverage shown for an earlier base-driven discovery execution is historical; the independent action may have advanced the shared authorship index since then.
Charts and dates
Open Exploratory report to load charts; they do not poll continuously. Chart links narrow the current selection. Top endpoints describe the selected category, so organizational endpoints are groups rather than document names.
| Display | Interpretation |
|---|---|
| Direct/chain partition | Unique candidates, mutually exclusive topology groups |
| Family and topology | Candidates carrying each matching signal; bars can overlap |
| Origin RG | Saved group associations; multiple groups can overlap, unknown affiliations remain explicit |
| Publication years | Documentary candidates only; earliest eligible evidence publication date, not validated transfer date |
| Approximate destination dates | Fallback for undated documentary pairs from exact destination RFC metadata, text or posted submission, with matching indexed source hashes |
| Historical snapshots | Saved documentary batch scope, not new discoveries; hidden for nondefault categories or additional filters |
| Batch diagnostics | Historical saved scope, explicitly independent of current filters |
Evidence dates take priority over approximations. Source-document publication, collection timestamps, administrative updates and unpublished submissions are not destination publication proxies. Approximate dates may precede the actual relationship. Different collection coverage can change the curve.
Refresh and export
Refresh charts forces recalculation. Otherwise, results may be reused for up to five minutes for the same user and filters. Relevant committed candidate, review, participant, evidence, indexed metadata and report changes invalidate that cache, including bulk writes by collection workers. Unrelated source downloads do not invalidate PostgreSQL caches. Failed cache access falls back to calculation; a continuously changing dataset is not stored as a stable result.
Download CSV exports the selected chart datasets. Candidate dates CSV exports per-documentary-candidate date, basis, precision and source provenance. Print / PDF uses the browser print dialog. Historical snapshots cannot reconstruct arbitrary historical filtered totals.
The read-only endpoint /runs/RUN_UUID/exploratory/ supports ?format=csv, ?format=dates and ?refresh=1; normal run access rules apply. Reading charts does not collect new sources or run discovery. Cache storage is operational and does not replace preserved research artifacts.
A zero count is not a negative scientific finding. Inspect execution status, input dates and coverage gaps before interpreting absence.