Extracting the necessary information from the stored database is one of the most demanded research areas today, especially in the field of subject-based information extraction. This application is a crucial area of focus for many researchers. The existing technique in this area relies on user observation-based responses. In this approach, each time a user submits a query, they need to select the most relevant content from the query extraction and then rank the items. The search is then refined, and the process continues in the next iteration. This method is time-consuming and requires domain knowledge about the system. To overcome these issues in subject-based information extraction, creating an index of datasets can improve search time and enhance information extraction efficiency. Initially, this process may take some time to train the input datasets, but the extraction will yield the most relevant data. Experimental results and practical applications indicate that the proposed technique performs well overall. The outputs confirm that even if the query is designed by an unprofessional user or lacks specific query points, the technique produces effective results.