pipeline/segment-classifier-model.json

{
  "classifier": "ProtocolSegmentClassifier",
  "description": "Per-word logistic regression for protocol segment detection",
  "features": [
    "is_strong_protocol",
    "is_weak_protocol",
    "is_expanded_symbol",
    "has_syntax_chars",
    "word_length_norm",
    "is_short_word",
    "context_strong_density",
    "context_any_density",
    "left_is_strong",
    "right_is_strong",
    "is_number_like",
    "strong_neighbor_count",
    "is_all_lower",
    "position_ratio"
  ],
  "weights": [
    0.33411720958159186,
    -0.01883618036579912,
    0.0,
    0.0,
    -0.09512784613770699,
    -0.27082908357760427,
    0.2582600934920609,
    0.24189819020570574,
    0.3569088147419212,
    0.33751703709094505,
    0.09194955109384083,
    1.0515119624353424,
    -0.17715608130670168,
    -0.11667765899752854
  ],
  "bias": -1.126425236495759,
  "threshold": 0.5
}