Hm, if we allow the RelationLearner to fill in AttributeSemanticNodes as possible slot fillers as a backup when other ObjectSemanticNodes aren't available I think the relationLearner might be able to handle this. However, this type of slot filling probably has ripple effects on our composite learners.
Hm, if we allow the
RelationLearnerto fill inAttributeSemanticNodes as possible slot fillers as a backup when otherObjectSemanticNodes aren't available I think the relationLearner might be able to handle this. However, this type of slot filling probably has ripple effects on our composite learners.