Contextual Recall and Precision
Recently in one of my university lectures I was teaching the concepts of Recall and Precision to my students. As you might know these concepts are used by IR researchers to assess the search performance of various IR systems. In my lecture I highlighted the fact that regarding the intrinsic complexity in the relevance concept our calculation of Recall and Precision will always be problematic. I think to address this issue we can consider it form a contextual approach. As a result, for any system we might be able to consider new measures which I'd like to call "Contextual Recall and Precision" which not only shows the performance of the system, but also contains the contextual elements which can potentially affect our measurement. The only dilemma here is the way we use to identify the important contextual elements and how we attribute each one's role in the final Recall and Precision.