Simulation - Deterministic model results in different recall curves when not seeded

The CLI suggests that there are 2 seeds available: one for the model and one for the prior knowledge.

  --seed SEED           Seed for models. Use integer between 0 and 2^32 - 1.
  ...
  --init_seed INIT_SEED
                        Seed for setting the prior indices if the --prior_idx option is not used. If the option --prior_idx is used with one or more index, this option is ignored.

Running a simulation with both seeds will result in identical results.

asreview simulate hall -s result_run_3.h5 --init_seed 1 --seed 1
asreview simulate hall -s result_run_4.h5 --init_seed 1 --seed 1

asreview plot result_run_3.h5 --type 'inclusion' -o plot_run_3.png
asreview plot result_run_4.h5 --type 'inclusion' -o plot_run_4.png

The default model is the deterministic NB model. So I thought I could skip the --seed value. However, this results in different results. Any thoughts?

asreview simulate hall -s result_run_1.h5 --init_seed 1
asreview simulate hall -s result_run_2.h5 --init_seed 1

asreview plot result_run_1.h5 --type 'inclusion' -o plot_run_1.png
asreview plot result_run_2.h5 --type 'inclusion' -o plot_run_2.png

[J535D165]

Oke, makes sense. Clarification of the documentation will do, isn’t it?

  --seed SEED           Seed for the active learning cycle (classifiers, balance strategies, 
                        and query strategies). Use integer between 0 and 2^32 - 1.

[J535D165]

Any one interested in doing a PR on this? (the docs)