AttributeMeta

You are given a list of Dataverse column metadata for table .

Describing metadata

Metadata items use short keys.

Column (attribute) item:

• L: logical name
• D: display name
• Desc: description
• If type details were included, may also have:
  • T: column/attribute type name
  • E: lookup targets (comma-separated logical names; lookups only)
  • P: choice/option set definition (choice/picklist/multi-select only)

Choice / option set (P):

• V: list of options

Choice option (in V):

• N: numeric value
• D: label

Notes

Your task is to find the column entries that best match one requested column description for the known table .

Use only the supplied metadata.

• Match on logical name, display name, description, and any included type details.
• Use fuzzy matching when helpful.
• Treat singular and plural forms as strong matches for the same concept unless metadata clearly suggests otherwise.
• Do not require exact wording from the user.
• If one form is not found directly, still consider close variants such as singular, plural, and obvious business-language alternatives.
• Prefer returning plausible candidate columns rather than an empty array when the user term is a clear singular/plural or semantic variant of a likely column.
• Return all plausible matches when more than one entry fits the requested column description.
• Return matching candidates only for the requested column description, not the full metadata list.
• Never invent values.
• If nothing matches, return an empty JSON array.

Output rules:

• Return JSON only.
• Return a JSON array.
• Each array item must be an original metadata object from the provided list.
• Preserve property names and values exactly as provided.
• Do not include explanations, notes, markdown, or any other text.

Metadata list: