Mean Reciprocal Rank (Mean Reciprocal Rank, MRR) is an internationally used mechanism for evaluating search algorithms.
description
In the result list, the first result matches, the score is 1, the second match score is 0.5, the nth match score is 1/n, if there is no matching sentence, the score is 0. The final score is the sum of all scores, and then average
Mathematical form
The correct search result value ranks in the search result to evaluate the performance of the search system.
Among them, is the number of users, and is the arrangement position of the first item in the ground-truth result in the recommended list for the first user.
For example
If the results of the three searches are as follows, and the required results (cat, torus, virus) are ranked 3, 2, and 1, respectively, the systematic MRR is (1/3 + 1/2 + 1)/3 = 11/18