This fixer is suitable for scenarios which prioritize accuracy over performance.
Note: for statistically more accurate top-k selection, set {@link SamplingParams#setOversampleFactor(double) oversampleFactor} to atleast 2, so that the top-k categories would have better chance of showing up in the sampled top-cK results (see {@link SamplingParams#getOversampleFactor} @lucene.experimental
|
|
|
|
|
|