Open Rebecca-auque opened 2 years ago
To detect the error in this sentence, we need to disable these two antipatterns:
<antipattern>
<!-- Perhaps better as tags in disambiguation: _GN_... -->
<unify>
<feature id="number"/>
<feature id="gender"/>
<token postag="[NZ].*" postag_regexp="yes"/>
<token postag="J.*|V ppa.*" postag_regexp="yes"/>
</unify>
</antipattern>
<antipattern>
<!-- Perhaps better as tags in disambiguation: _GN_... -->
<unify>
<feature id="number"/>
<feature id="gender"/>
<token postag="D.*" postag_regexp="yes"/>
<!-- <exception regexp="yes">mon|ton|son</exception>-->
<token postag="N.*" postag_regexp="yes"/>
</unify>
</antipattern>
This change causes 9 test failures. We could try to fix all those failures, but perhaps it is easier to write a new rule for this kind of errors.
Okay, thank you for the explaination. For the time being I just created a rule to solve the specific issue 'leurs propres démarcation'.
I think this new rule D_J_N (the rule J_N but with a determiner before the adjective) will generalize well.
A different approach: create a rule for adjective + noun, but only with adjectives usually found before a noun so that we are sure they are adjectives that have to agree with the next noun. Currently, we have this list in disambiguation.xml: propre|sale|vrai|petit|bon|meilleur|mauvais|nouveau|grand|vieux|long|ancien|haut|seul|beau|superbe
@udomai @Rebecca-auque: Could you help me expand this list? It cannot be done with LanguageTool itself because the disambiguation is not good enough. In French, adjectives before noun are less common than in other languages (Spanish). The list doesn't need to be large.
Here is a list of some adjectives: important|gentil|méchant|parfait|grave|rare|énorme|lourd|futur|ancien|cher|majeur|simple|joli|mineur|meme|immense|minsucule|lointain|proche|infime|gigantesque|étrange|hideux|ravissant|magnifique|sombre|timide|sympathique|généreux|fidèle|loyal.
Les chats ont leurs propres démarcation territoriale. → Les chats ont leurs propres démarcations territoriales.