models/BERTopic/ctfidf_config.json

{
  "ctfidf_model": {
    "bm25_weighting": false,
    "reduce_frequent_words": false
  },
  "vectorizer_model": {
    "params": {
      "analyzer": "word",
      "binary": false,
      "decode_error": "strict",
      "encoding": "utf-8",
      "input": "content",
      "lowercase": true,
      "max_df": 1.0,
      "max_features": null,
      "min_df": 2,
      "ngram_range": [
        1,
        2
      ],
      "stop_words": "english",
      "strip_accents": null,
      "token_pattern": "(?u)\\b\\w\\w+\\b",
      "vocabulary": null
    },
    "vocab": {
      "request": 285,
      "large": 197,
      "collection": 63,
      "posted": 261,
      "parking": 245,
      "sign": 312,
      "violation": 365,
      "working": 376,
      "contrary": 83,
      "stop": 328,
      "work": 374,
      "order": 241,
      "dirty": 110,
      "sidewalk": 311,
      "access": 3,
      "receipt": 273,
      "site": 316,
      "street": 331,
      "condition": 76,
      "ice": 169,
      "non": 232,
      "missed": 224,
      "area": 20,
      "license": 203,
      "engine": 126,
      "idling": 170,
      "cond": 75,
      "lead": 200,
      "residential": 290,
      "sewer": 309,
      "use": 358,
      "comments": 66,
      "water": 370,
      "meter": 221,
      "broken": 34,
      "leaking": 202,
      "private": 263,
      "residence": 288,
      "refund": 277,
      "return": 295,
      "permit": 249,
      "improper": 175,
      "certificate": 50,
      "occupancy": 236,
      "illegal": 171,
      "plumbing": 256,
      "pedestrian": 248,
      "signal": 313,
      "defective": 101,
      "inadequate": 179,
      "heat": 159,
      "new": 230,
      "bus": 36,
      "placement": 254,
      "repair": 284,
      "building": 35,
      "damaged": 93,
      "cracked": 88,
      "bicycle": 28,
      "flooding": 140,
      "overnight": 242,
      "commercial": 67,
      "storage": 329,
      "surveillance": 337,
      "waste": 369,
      "blocked": 30,
      "construction": 79,
      "school": 306,
      "property": 270,
      "cover": 86,
      "noise": 231,
      "gas": 147,
      "problem": 265,
      "delivery": 103,
      "goods": 150,
      "curb": 90,
      "hitting": 162,
      "phone": 252,
      "c1": 40,
      "trees": 351,
      "rent": 283,
      "unauthorized": 354,
      "chronic": 55,
      "hanging": 157,
      "accident": 4,
      "cleaning": 58,
      "asp": 22,
      "establishment": 129,
      "public": 272,
      "space": 320,
      "dispute": 111,
      "home": 163,
      "electronics": 123,
      "chemical": 53,
      "chained": 51,
      "smoking": 319,
      "car": 42,
      "general": 149,
      "maintenance": 214,
      "asbestos": 21,
      "open": 239,
      "missing": 226,
      "emergency": 124,
      "odor": 237,
      "catch": 48,
      "basin": 26,
      "tax": 342,
      "temporary": 345,
      "failure": 134,
      "debris": 98,
      "falling": 136,
      "danger": 95,
      "air": 9,
      "defect": 100,
      "metal": 219,
      "protruding": 271,
      "information": 183,
      "cut": 91,
      "vacant": 360,
      "lot": 211,
      "resident": 289,
      "pipe": 253,
      "toilet": 349,
      "button": 38,
      "wiring": 373,
      "buzzer": 39,
      "vehicle": 361,
      "carbon": 43,
      "monoxide": 227,
      "smoke": 318,
      "audit": 24,
      "damp": 94,
      "leak": 201,
      "st": 324,
      "facility": 133,
      "law": 198,
      "cigarette": 56,
      "sale": 303,
      "minor": 222,
      "pool": 258,
      "graffiti": 151,
      "speed": 322,
      "scale": 305,
      "hours": 165,
      "safety": 301,
      "equipment": 128,
      "signs": 314,
      "notice": 233,
      "box": 31,
      "weeds": 372,
      "grating": 152,
      "removal": 280,
      "requested": 286,
      "controller": 85,
      "flasher": 139,
      "loose": 210,
      "time": 347,
      "switch": 340,
      "stump": 335,
      "sampling": 304,
      "required": 287,
      "head": 158,
      "card": 45,
      "stuck": 333,
      "commission": 68,
      "lack": 195,
      "litter": 207,
      "comm": 65,
      "bldg": 29,
      "basket": 27,
      "fallen": 135,
      "bridge": 33,
      "warning": 367,
      "prohibited": 269,
      "inspection": 187,
      "roof": 298,
      "illness": 174,
      "injury": 185,
      "ticket": 346,
      "clear": 59,
      "insects": 186,
      "highway": 161,
      "multiple": 229,
      "devices": 107,
      "animal": 13,
      "lane": 196,
      "control": 84,
      "dirt": 108,
      "clothing": 61,
      "high": 160,
      "pressure": 262,
      "debt": 99,
      "materials": 218,
      "agency": 8,
      "application": 17,
      "station": 325,
      "unguarded": 355,
      "driveway": 117,
      "gallons": 144,
      "device": 106,
      "service": 307,
      "swimming": 338,
      "coin": 62,
      "tobacco": 348,
      "taste": 341,
      "filing": 138,
      "technical": 343,
      "issues": 192,
      "rights": 297,
      "miscellaneous": 223,
      "color": 64,
      "division": 112,
      "retaining": 293,
      "zoning": 378,
      "lawn": 199,
      "status": 326,
      "enforcement": 125,
      "excessive": 131,
      "contractor": 82,
      "dry": 118,
      "complaince": 71,
      "electrical": 121,
      "amusement": 11,
      "ride": 296,
      "incident": 180,
      "received": 274,
      "program": 268,
      "nypd": 235,
      "issue": 191,
      "electronic": 122,
      "transfer": 350,
      "eft": 120,
      "address": 6,
      "incorrect": 181,
      "wrong": 377,
      "paper": 244,
      "list": 206,
      "passenger": 247,
      "guide": 154,
      "assistance": 23,
      "exemption": 132,
      "sro": 323,
      "truck": 352,
      "driver": 115,
      "city": 57,
      "tunnel": 353,
      "licensed": 204,
      "improvement": 177,
      "sticker": 327,
      "animals": 14,
      "company": 69,
      "waterway": 371,
      "abcs": 1,
      "housing": 167,
      "haitian": 155,
      "creole": 89,
      "apartment": 16,
      "unit": 356,
      "retail": 292,
      "store": 330,
      "initial": 184,
      "fdny": 137,
      "approved": 18,
      "multi": 228,
      "business": 37,
      "annual": 15,
      "related": 278,
      "dep": 104,
      "internal": 189,
      "vending": 362,
      "machine": 213,
      "marine": 217,
      "dumpster": 119,
      "damage": 92,
      "cable": 41,
      "missed collection": 225,
      "street cond": 332,
      "use comments": 359,
      "private residence": 264,
      "improper use": 176,
      "residential building": 291,
      "plumbing work": 257,
      "work illegal": 375,
      "construction site": 80,
      "cover missing": 87,
      "odor sewer": 238,
      "sewer catch": 310,
      "catch basin": 49,
      "danger falling": 96,
      "metal protruding": 220,
      "defective street": 102,
      "carbon monoxide": 44,
      "safety equipment": 302,
      "permit license": 250,
      "grating missing": 153,
      "card stuck": 46,
      "stuck meter": 334,
      "warning signal": 368,
      "clear water": 60,
      "dirt litter": 109,
      "litter debris": 208,
      "open unguarded": 240,
      "swimming pool": 339,
      "amusement ride": 12,
      "address incorrect": 7,
      "incorrect status": 182,
      "driver license": 116,
      "home improvement": 164,
      "improvement contractor": 178,
      "company license": 70,
      "abcs housing": 2,
      "haitian creole": 156,
      "dep internal": 105,
      "internal use": 190,
      "vending machine": 363,
      "unknown": 357,
      "line": 205,
      "knocked": 193,
      "post": 260,
      "wall": 366,
      "excavation": 130,
      "support": 336,
      "foreign": 142,
      "dead": 97,
      "contact": 81,
      "installation": 188,
      "break": 32,
      "house": 166,
      "change": 52,
      "management": 215,
      "conditioning": 77,
      "condo": 78,
      "foundation": 143,
      "referral": 275,
      "route": 299,
      "concrete": 74,
      "panel": 243,
      "complaint": 72,
      "basement": 25,
      "garage": 145,
      "sink": 315,
      "reflected": 276,
      "chinese": 54,
      "spanish": 321,
      "arabic": 19,
      "hqs": 168,
      "english": 127,
      "russian": 300,
      "portable": 259,
      "korean": 194,
      "10": 0,
      "television": 344,
      "retaining wall": 294,
      "parking lot": 246,
      "air conditioning": 10,
      "location": 209,
      "manufacturing": 216,
      "care": 47,
      "activity": 5,
      "low": 212,
      "food": 141,
      "number": 234,
      "remove": 281,
      "pet": 251,
      "compressed": 73,
      "illegal use": 173,
      "illegal improper": 172,
      "sewage": 308,
      "drinking": 113,
      "garbage": 146,
      "small": 317,
      "removing": 282,
      "plants": 255,
      "problem use": 266,
      "drinking water": 114,
      "gas sewer": 148,
      "ventilation": 364,
      "problems": 267,
      "related problems": 279
    }
  }
}