Query using keywords is one of the user friendly and widely used forms of querying in recent days. The existing systems focused on fetching all or a fixed number of top tuples, which may directly contain the portion of keywords or interconnected to other related tuples via foreign key. These systems will not give any valuable information more than individual interconnected tuple structures.. In this paper, the authors focus on compute structural statistics for keyword by extending the group & aggregate framework. In RDBMS, rows with same values on specific attributes can be grouped together and aggregate functions can be used to aggregate the values of the same group.