search

ag-sc / matoll

M-ATOLL: A Framework for the Lexicalization of Ontologies in Multiple Languages
Other
10 stars 5 forks source link

each Sense needs own vector for prediction (only each entry is not sufficient enaugh) #57

Closed swalter2 closed 9 years ago

swalter2 commented 9 years ago

also then add, how often a label, e.g. wife occurs in other properties etc.

swalter2 commented 9 years ago

Each sense does not need a FeatureVector, as all we need is already stored in the provenance. When doing #50 convert Provenance into FeatureVector and call training/prediction.