siiptuo / kriobot Goto Github PK

• Handle umlaut e.g. göra, gör, gjorde, gjort should be categorized into group 2 instead of group 4
• Add stem to strong verbs e.g. springa, springer, sprang, sprungit would have stem spr-ng
• Add separate group for truly irregular verbs (e.g. vara) instead of placing them in group 4 with strong verbs

Based on Böja svenska verb by Om svenska.

For example lexeme "chest of drawers" combines lexemes "chest", "of" and "drawer".

Something like this:

SELECT ?lexeme ?lemma ?dateModified WHERE {
  ?lexeme dct:language wd:Q1860;
    wikibase:lemma ?lemma;
    schema:dateModified ?dateModified.
  hint:Prior hint:rangeSafe true.
  FILTER(?dateModified > "+2023-01-10T00:00:00Z"^^xsd:dateTime)
}

WMF has a policy that makes it easier for them to find malicious code hitting their endpoints. See opencitationsbot for how to set a user agent with WBI.

In https://www.wikidata.org/w/index.php?title=Lexeme%3AL242808&diff=1495165027&oldid=1407681033 kriobot split förvarning into "förvara" and "-ning", which seems wrong to me, since the noun for förvara is "förvaring".

Compare:
https://sv.wiktionary.org/wiki/f%C3%B6rvaring
https://sv.wiktionary.org/wiki/f%C3%B6rvarning

I changed the item in https://www.wikidata.org/w/index.php?title=Lexeme%3AL242808&diff=1791659518&oldid=1511281980, but wanted to give a heads up on this problem case.

Find candidates with something like this:

SELECT ?lexemeId ?lemma WHERE {
  ?lexemeId dct:language wd:Q1860;
    wikibase:lemma ?lemma;
    wdt:P5238 ?lexemes.
  FILTER(NOT EXISTS { ?lexemeId ontolex:lexicalForm ?form. })
}

Maybe skip idioms and other multi-word lexemes as these can be difficult to inflect correctly.

I suggest simplifying by checking if len() == 1 and return None in all other cases