An Information Extraction Method from Different Structural Web Sites by Word Distances between a User Instantiated Label and Similar Entity.