config.json

{
  "data": "./data/words.csv",
  "model": "./models/wiki.en.bin",
  "labels": "./data/labels.csv",
  "vectors": "./data/vecs.csv",
  "Z": "./data/Z_hclust.csv",
  "distance": 1,
  "cutoff": 5,
  "stepsize": 0.5,
  "lev_threshold": 3,
  "wg_threshold": 2,
  "ng_threshold": 2.5,
  "trouble": {
    "product":"P",
    "equipment": "E",
    "frozen": "F",
    "goods": "G",
    "pallet":"B",
    "powder": "O",
    "machine":"M",
    "groupage":"R",
    "liquid":"L",
    "solution":"S",
    "box":"X",
    "empty": "MT",
    "chilled": "C",
    "mix":"I",
    "ambient":"A",
    "hazardous":"H",
    "haz": "HZ",
    "series": "SR",
    "agricultural": "AG",
    "appliances": "AP",
    "bales": "BL",
    "bottle": "BT",
    "tinned": "T",
    "carton": "CR",
    "compressed": "CM",
    "concentrate": "CT",
    "drums": "DR",
    "dry": "D",
    "bulk": "BU",
    "kitchen": "K",
    "additive": "AD",
    "recycled": "RC",
    "refrigerated": "RF",
    "supplies": "SP",
    "waste": "W",
    "pellets": "PL",
    "granules": "GR",
    "fittings": "FT",
    "components": "CP",
    "cylinder" : "CY",
    "tank" : "T"
  },
  "stoppers": [
    "entity",
    "credit card",
    "abstraction",
    "artifact",
    "commercial",
    "commercial enterprise",
    "substance",
    "physical entity",
    "reproductive structure",
    "communication",
    "covering",
    "device",
    "discipline",
    "dish",
    "drug",
    "material",
    "matter",
    "message",
    "person",
    "utility",
    "organism",
    "vascular plant",
    "whole",
    "object",
    "body part",
    "administrative unit",
    "application",
    "aquatic vertebrate",
    "activity",
    "events",
    "causal agent",
    "artillery",
    "feeling",
    "chamber",
    "visual property",
    "athlete",
    "constant",
    "control",
    "imaginary being",
    "psychological feature"
  ],
  "decode_output":true,
  "regex": [
    ["[/!#,.`\\[\\]\\(\\)]", " "],
    ["  ", " "],
    ["x?[0-9]x?", ""],
    ["&|\\+", ""],
    [" of ", " "]
  ]
}