Parties and hypotheses
The two categories of potential claimants in a case report, how each is produced, and the five signals that rank them.
The question a case report answers is who could sue? BINA answers in two clearly separated categories, because the evidence behind them is of a different kind.
#Category 1 — named in the decision
A category 1 party is named by the authority itself. It must cite a recital and carry a quote that passes the quote gate, and the party's name must occur inside that quote. These are findings about what the decision says — not findings about standing.
They are created either by counsel through POST /v1/radar/cases/:caseNumber/parties or by extraction, whose verified party proposals are promoted to category 1.
#Category 2 — added by BINA
A category 2 party is a group the decision does not name but that could plausibly have been harmed — for example, competing app stores a decision mentions only in general terms. The hypothesis pass (hypothesizeRun) proposes at most three such groups. Each one carries:
- a multi-sentence why write-up;
- a to test line naming the evidence that would confirm or refute it;
- a loss formula with its missing input named;
- the recitals it rests on;
- public-source research attached by the research pass.
checkHypothesis refuses a hypothesis that names a group the decision names, uses a number not in the cited notes, or presents a computed result. Hypotheses are stored with status hypothesis and shown under the heading "hypotheses, not findings", labelled Added by BINA. Each pass replaces the previous batch; hypotheses are never sent for counsel approval.
#The five claimant signals
Every case report lists the same five criteria for each party, whatever data is available:
| Criterion | Question |
|---|---|
presence_on_market | Was the party active on the French market during the infringement? |
declared_impact | Did the decision or the party describe an impact? |
change_in_results | Do the party's results show a change? |
visibility | Is the party publicly visible on the issue? |
size_of_stake | How large is the party's stake? |
Each signal is strong, medium, none or unknown. Presence can be assessed against an exact French registry name match for a sourced period; a French registry hit is at most medium. Criteria with no connected source are shown as unavailable — never guessed, never zero.
#Population estimates
A case report can also record affected-population estimates from official French registers — SIRENE undertakings by NAF/APE code and area, or INSEE residents by commune or department. Each estimate stores its formula, inputs, query URL and raw response hash, and is shown as an estimate: an upper bound on the addressable population, never a claimant finding.