If we do have information about term occurrences in the relevant and non-relevant sets, it can be summarized in a contingency table, shown in Table 7.1. This information could be obtained through relevance feedback, where users identify relevant documents in initial rankings. In this table, ri is the number of relevant
documents containing term i, ni is the number of documents containing term i,
N is the total number of documents in the collection, and R is the number of
relevant documents for this query.