For the example query “tropical fish aquarium” in Table 4.6, this means we estimate the result set size by multiplying the number of documents containing both “tropical” and “aquarium” by the probability that a document contains “fish” given that it contains “aquarium”, or: