The goals of this issue will change depending on the evolution of the datamodel.
Goals
[x] Build a nounification map that associates to a verb V the nouns N that enable you to represent X V Y by (X,N,Y) (ex: The animal lives in the farm -> lives in ~ residence -> (animal,residence,farm))
[x] Build a nounification map that associates to a verb V the nouns N that enable you to represent X V Y by (Y,N,X) (ex: The animal lives in the farm -> lives in ~ inhabitant -> (farm,inhabitant,animal))
[x] Be able to merge a prep dependency edge with its input node (if it's a verb). Ex: Napoleon rules on France -> rules -prep_on-> France -> rules on -prep-> France (in progress on one of my branches)
[x] Be able to obtain past participle predicates from verbs (into the nounification process). Ex: rules on -> ruled on. Indeed, a lot of properties/aliases in Wikidata are past participle (born on, ...). To perform this task we can use a database of verbs/past participles (i already found one)
[x] Support the new representation of triples (?)
Remarks
We can still use the automatic nounification map -> when the 2 new nounification maps don't contain an entry, take the nouns of the automatic nounification map
Perhaps we could avoid to have 2 different manual nounification maps : a single one with functions getPredicate, getReversePredicate...
Linked to: https://github.com/ProjetPP/PPP-QuestionParsing-Grammatical/pull/106, https://github.com/ProjetPP/PPP-QuestionParsing-Grammatical/issues/110, https://github.com/ProjetPP/Documentation/issues/52, https://github.com/ProjetPP/Documentation/pull/56
Branch: https://github.com/ProjetPP/PPP-QuestionParsing-Grammatical/tree/reverse_predicates
The goals of this issue will change depending on the evolution of the datamodel.
Goals
V
the nounsN
that enable you to representX V Y
by(X,N,Y)
(ex:The animal lives in the farm
->lives in ~ residence
->(animal,residence,farm)
)V
the nounsN
that enable you to representX V Y
by(Y,N,X)
(ex:The animal lives in the farm
->lives in ~ inhabitant
->(farm,inhabitant,animal)
)prep
dependency edge with its input node (if it's a verb). Ex:Napoleon rules on France
->rules -prep_on-> France
->rules on -prep-> France
(in progress on one of my branches)rules on
->ruled on
. Indeed, a lot of properties/aliases in Wikidata are past participle (born on, ...). To perform this task we can use a database of verbs/past participles (i already found one)Remarks
getPredicate
,getReversePredicate
...