Is your feature request related to a problem? Please describe.
I would like to be able to report recall scores as part of overall benchmark / comparison between the different databases
Describe the solution you'd like
I would like the output JSON file to include a recall score
Describe alternatives you've considered
N/A
Additional context
Recall is defined as the proportion of returned samples that belong to the set of top-k neighbors.
Evaluating this thus requires a ground truth, which means extra computations; note that we can pre-compute the true top-k sets since we know the dataset & query set ahead of time.
Is your feature request related to a problem? Please describe.
I would like to be able to report recall scores as part of overall benchmark / comparison between the different databases
Describe the solution you'd like
I would like the output JSON file to include a recall score
Describe alternatives you've considered
N/A
Additional context
Recall is defined as the proportion of returned samples that belong to the set of top-k neighbors.
Evaluating this thus requires a ground truth, which means extra computations; note that we can pre-compute the true top-k sets since we know the dataset & query set ahead of time.