Syntactic Multi-view Learning for Open Information Extraction