In the second comparison term based approach’s results turned out be less relevant to the query in comparison to phrase based approach. For an example query operating system would fetch file system file as most relevant document while other would fetch operating system concept file as most relevant which is logically correct. This is because former approach uses both word operating and system as distinct term while later treats them as single. But recall in term based approach would be more as there are more terms to find out in the repository, also in case of single word query only term based approach would fetch a result.
Also second comparison may enjoy the benefits of using ontology in both case and hence can improve recall. Although our searching method takes some time to index and then search the query but it reduces time of overall search in comparison to time required to search a document as a whole. By varying threshold of index creation we can vary the no. of words in document descriptive i.e. index table. We have also found that threshold value above a particular limit can eliminate some important words which is not desirable for our search. This limit depends upon the size of the documents we are using in our system.