Near-duplicate matching needs roughly 8 words or more. Shorter snippets are still checked for an exact match.

ℹ️

What this can and cannot find

An exact match is conclusive: the content is byte-for-byte what someone registered. A near match means the content is substantially the same β€” reformatted, lightly edited, or partially reused.

Near matching compares lexical overlap, not meaning. Content that has been thoroughly reworded while keeping its sense will not be found, because at that point it is mathematically indistinguishable from unrelated text. A "no match" result is therefore not evidence that nothing was claimed.

Only claims whose owner opted into near-duplicate detection are searchable this way. An exact-hash check always works.