Exploiting Structural Similarity For Effective Web Information Extraction.