Abstract
Now-a-days databases become web accessible; these databases having data units are encoded. To separate data units assign meaningful labels. To assign labels there is an automatic annotation approach which automatically assign labels to data units within the search result records returned from web databases. For that it provides an annotation wrapper for the search site to automatically constructed and annotate the new result pages from the same web databases. There are six basic annotators, for every basic annotator we produce label for the data units within their group. A probability model is selected to determine the most appropriate label for each group. This annotation wrapper generate an annotation rule that describes how to extract the data units from result page .Once the annotation wrapper annotate the data there is unnecessary to perform the alignment and annotation phases again. We construct an annotation wrapper by using data types, data contents, presentation styles and adjacency information. In this paper the data annotation problem is studied and proposed a multiannotator approach to automatically constructing an annotation wrapper for annotating the search result records recover from any given web databases.