The spec currently defines meta on an alignment record as covering the link as a whole. A common review tool use case requires flagging potential discrepancies between the source and target sides — for example, where a source token's status and a target token's status appear inconsistent.
This is correctly modeled as a property of the link, not of either token in isolation: the source and target systems remain authoritative for their own data, but the alignment record is the right place to record a reviewer's assessment of the relationship between them.
Since meta is already extensible, no structural change is needed. We believe per-side metadata within meta was part of the original design intent, but there are currently no examples in the spec illustrating it. We propose adding examples using source and target keys:
{
"source": ["n40001001001"],
"target": ["40001001001"],
"meta": {
"origin": "manual",
"status": "created",
"source": { "flag": "unverified" },
"target": { "flag": "possible-omission" }
}
}A review tool that does not assume source or target status is always correct can write its findings here without modifying either referenced system.
The spec currently defines
metaon an alignment record as covering the link as a whole. A common review tool use case requires flagging potential discrepancies between the source and target sides — for example, where a source token's status and a target token's status appear inconsistent.This is correctly modeled as a property of the link, not of either token in isolation: the source and target systems remain authoritative for their own data, but the alignment record is the right place to record a reviewer's assessment of the relationship between them.
Since
metais already extensible, no structural change is needed. We believe per-side metadata withinmetawas part of the original design intent, but there are currently no examples in the spec illustrating it. We propose adding examples usingsourceandtargetkeys:{ "source": ["n40001001001"], "target": ["40001001001"], "meta": { "origin": "manual", "status": "created", "source": { "flag": "unverified" }, "target": { "flag": "possible-omission" } } }A review tool that does not assume source or target status is always correct can write its findings here without modifying either referenced system.