nlp - Extracting (subject,predicate,object) from dependency tree -
nlp - Extracting (subject,predicate,object) from dependency tree -
i'm interesting in extracting triples (subject,predicate,object) questions.
for example, transform next question :
who wife of president of usa?
in :
(x,iswifeof,y) ∧ (y,ispresidentof,usa)
x , y unknows have find in order reply question (/\ denotes conjunction).
i have read lot of papers topic , perform task using existing parsers such stanford parser. know parsers output 2 types of info :
parse construction tree (constituency relations) dependency tree (dependency relations)some papers seek build triples parse construction tree (e.g., triple extraction sentences), approach seems weak deal complicated questions.
on other hand, dependency trees contain lot of relevant info perform triple extraction. lot of papers claim that, didn't find of them gives explicitely detailed procedure or algorithm. of time, authors analyze dependencies produce triples according rules didn't give.
does knows paper more info extracting (subject,predicate,object) dependency tree of question?
nlp rdf stanford-nlp question-answering
Comments
Post a Comment