Abstract
This Keyword queries provide fluent access to data over big databases, but there is problem of low and poor ranking quality or priority problem of obtaining results after querying .To satisfy the user it is necessary to identify the queries that have low ranking quality. In this paper, we are creating a framework to calculate the ratio of degree of difficulty of keyword query on the big databases by observing the properties of hard queries, in consideration with both unstructured and the content of the database and the results of query. We are giving the ranking to predicted query results as per user requirement for the database. We create our keyword prediction architecture is made against two algorithms popular for keyword search ranking methods and these methods will work on unstructured big databases. Our unique results show that our methods or algorithms predict the hard queries with high accuracy for unstructured database. We going to reduce the difficulty of keyword prediction over the unstructured big databases and also we trying reducing noise occurred because of ranking mechanism of unstructured database. Further, we present methods to minimize the incurred time overhead.