huwiki.yaml

name: huwiki
label: Hungarian Wikipedia
host: hu.wikipedia.org

external_samples:
  sampled_revisions.40k_2016:
      quarry_url: "http://quarry.wmflabs.org/run/79645/output/0/json-lines?download=true"
  human_labeled_revisions.raw.5k_2016:
      labeling_campaign: "https://labels.wmflabs.org/campaigns/huwiki/12/"

autolabeled_samples:
  trusted_edits: 1000
  trusted_groups:
    - sysop
    - oversight
    - trusted
    - bot
    - rollbacker
    - checkuser
    - abusefilter
    - bureaucrat
    - editor
    - templateeditor
    - interface-editor
  labeled_samples:
    autolabeled_revisions.40k_2016: sampled_revisions.40k_2016

balanced_5k_samples:
  revisions_for_review.5k_2016: autolabeled_revisions.40k_2016

merged_samples:
  labeled_revisions.40k_2016:
    - autolabeled_revisions.40k_2016
    - human_labeled_revisions.5k_2016

extracted_samples:
  labeled_revisions.w_cache.40k_2016:
    sample: labeled_revisions.20k_2016
    features_for: [damaging, goodfaith]
    

models:
  damaging:
    sample: labeled_revisions.w_cache.40k_2016
    label: damaging
    pop_rate_true: 0.01
    tune: {}
    cv_train:
      algorithm: GradientBoosting
      label_weight: 10
      parameters:
        max_depth: 7
        learning_rate: 0.01
        max_features: log2
        n_estimators: 700
  goodfaith: {...}