[12:22:41] All the best for your talk. I'll be listening from Sydney (for 30 min only unfortunately - I'll catch up on the recording later). (re @Dnshitobu: Thank you this is much better and would be a good highlight for our session. See you all today at 2:30pm at the Classe Numerice...) [14:05:54] what was the etherpad for mahirs talk? [14:06:50] https://etherpad.wikimedia.org/p/wikimania2026-mahir256 [14:07:05] thank you! [14:30:21] 3829 [14:41:02] Now I should have added an additional heuristic that prefers lexemes with a lemma that is closer (via Levenshtein distance) to the item label. Do you have in mind any test for this? (re @u99of9: @dvd_ccc27919 is this something you've dealt with? If both L19560 and L19297 were linked to Q11500, I'd like to pick the one mos...) [14:45:55] It is only used as the last chance, when no other heuristic can determine which lexeme is better (re @dvd_ccc27919: Now I should have added an additional heuristic that prefers lexemes with a lemma that is closer (via Levenshtein distance) to t...) [14:59:15] Unfortunately I want to be able to pass in the label I want to match. In the first case I needed, the string will come from the label of the property that is linked to the item: (re @dvd_ccc27919: Now I should have added an additional heuristic that prefers lexemes with a lemma that is closer (via Levenshtein distance) to t...) [14:59:38] Details here. (re @u99of9: @dvd_ccc27919 is this something you've dealt with? If both L19560 and L19297 were linked to Q11500, I'd like to pick the one mos...) [15:00:44] This will be good for more common cases. (re @dvd_ccc27919: Now I should have added an additional heuristic that prefers lexemes with a lemma that is closer (via Levenshtein distance) to t...) [15:01:02] That seems fair to me. (re @dvd_ccc27919: It is only used as the last chance, when no other heuristic can determine which lexeme is better) [15:53:04] Now Z38197 should work