The databases on web are accessible through HTML based search engines. The information mined from these web servers mostly in unstructured format. Whenever, a web user submitted query, the web database shows multiple Search Result Records (SRRs). There is no any provision of the semantic labels of data units in result pages. Early applications require tremendous human efforts to annotate data units manually, which severely limit their scalability. To reduce the human efforts, the automatic annotator is proposed to extract set of SRRs from result page returned from the web databases in structured format.