A method to improve accuracy in comparing performance of the paired information retrieval system comprising the steps of: deploying test collection at various participating information retrieval systems; identifying relevant documents to a specific topic within various participating information retrieval systems; choosing top retrieved document from each participating information retrieval system by implementing a pre-defined assessment method; counting the frequency of top retrieved documents of each participating information retrieval system by deploying individual run for every topic; ranking the retrieved documents received after deploying by individual runs for every topic; computing the document relevance weight for every pair of systems-topics; conducting statistical significance test of all the retrieved documents between pairs of systems having its document relevance weight higher than a pre-defined range and presenting the same to a user.