topic_modelling_pepe / ctfidf_config.json
guibvieira's picture
Add BERTopic model
d560fd3
{
"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": 0.9,
"max_features": null,
"min_df": 0.05,
"ngram_range": [
1,
1
],
"stop_words": "english",
"strip_accents": "ascii",
"token_pattern": "(?u)\\b\\w\\w+\\b",
"vocabulary": null
},
"vocab": {
"people": 580,
"understand": 840,
"coin": 138,
"gave": 317,
"going": 331,
"pump": 650,
"years": 908,
"forget": 295,
"look": 474,
"miss": 531,
"pepearmy": 582,
"buy": 99,
"later": 435,
"guys": 348,
"list": 459,
"dev": 194,
"supply": 773,
"burned": 97,
"price": 633,
"days": 184,
"smart": 741,
"money": 535,
"memes": 513,
"new": 549,
"world": 900,
"cryptocurrency": 173,
"experienced": 257,
"significant": 734,
"past": 578,
"floki": 287,
"game": 314,
"reward": 688,
"distribution": 207,
"date": 182,
"like": 455,
"saying": 709,
"don": 215,
"funds": 309,
"legit": 448,
"usdt": 849,
"yes": 909,
"usd": 848,
"according": 2,
"different": 202,
"admin": 13,
"times": 806,
"does": 208,
"dollars": 214,
"receive": 671,
"won": 891,
"just": 420,
"wrong": 904,
"team": 788,
"know": 426,
"difference": 201,
"face": 263,
"problem": 637,
"talk": 782,
"support": 774,
"send": 719,
"instead": 402,
"guess": 346,
"spot": 752,
"trading": 820,
"campaign": 105,
"distributed": 206,
"kucoin": 429,
"says": 710,
"working": 898,
"join": 415,
"market": 496,
"analysis": 27,
"trending": 824,
"memecoins": 512,
"aidoge": 18,
"utc": 855,
"trade": 818,
"tell": 791,
"token": 808,
"win": 886,
"feel": 279,
"plan": 611,
"making": 492,
"big": 63,
"project": 642,
"need": 546,
"sure": 776,
"hold": 370,
"success": 770,
"app": 34,
"kyc": 430,
"btc": 89,
"eth": 245,
"bnb": 74,
"good": 334,
"pumped": 651,
"hard": 357,
"dumping": 225,
"available": 45,
"earn": 229,
"kind": 424,
"pepecoins": 587,
"cap": 106,
"lets": 450,
"sold": 744,
"right": 692,
"yeah": 906,
"ago": 15,
"pepecoin": 585,
"starting": 759,
"read": 666,
"lol": 471,
"guy": 347,
"dont": 216,
"think": 799,
"probably": 636,
"really": 669,
"rich": 690,
"eyes": 262,
"attention": 43,
"volume": 863,
"frog": 302,
"dex": 197,
"link": 456,
"easily": 230,
"swap": 777,
"sent": 721,
"bunch": 95,
"pepedex": 588,
"provide": 646,
"fees": 280,
"fake": 269,
"welcome": 882,
"real": 668,
"let": 449,
"getting": 322,
"started": 758,
"chart": 125,
"unless": 843,
"want": 869,
"holders": 372,
"holding": 373,
"strong": 766,
"did": 199,
"dump": 223,
"run": 699,
"looking": 475,
"higher": 367,
"remember": 681,
"pepepal": 595,
"liquidity": 458,
"literally": 464,
"members": 509,
"based": 53,
"love": 482,
"work": 896,
"offer": 560,
"major": 489,
"projects": 643,
"raid": 660,
"reply": 683,
"mins": 529,
"comment": 147,
"share": 723,
"rare": 664,
"frogs": 303,
"told": 811,
"happy": 356,
"connect": 155,
"gon": 332,
"leave": 446,
"original": 570,
"memetic": 514,
"use": 850,
"certain": 117,
"features": 277,
"chain": 119,
"close": 134,
"fam": 271,
"dude": 221,
"shit": 728,
"memecoin": 511,
"follow": 290,
"better": 62,
"long": 472,
"term": 793,
"launched": 438,
"earlier": 227,
"month": 537,
"thats": 796,
"scam": 711,
"bridge": 83,
"meme": 510,
"till": 804,
"locked": 469,
"holder": 371,
"address": 12,
"stake": 754,
"talking": 783,
"mining": 528,
"coming": 146,
"soon": 746,
"time": 805,
"way": 875,
"works": 899,
"pay": 579,
"decentralized": 188,
"power": 627,
"transaction": 821,
"fee": 278,
"process": 638,
"high": 366,
"marketcap": 497,
"doesn": 209,
"product": 639,
"devs": 196,
"help": 364,
"hope": 376,
"got": 335,
"post": 622,
"day": 183,
"brain": 81,
"announce": 28,
"change": 121,
"sell": 717,
"wallet": 866,
"coins": 142,
"inu": 406,
"gem": 318,
"needs": 547,
"build": 90,
"family": 272,
"little": 465,
"wait": 864,
"hear": 361,
"details": 193,
"used": 851,
"thought": 802,
"developers": 195,
"said": 703,
"platform": 613,
"space": 748,
"short": 730,
"profits": 641,
"selling": 718,
"bag": 51,
"hey": 365,
"blockchain": 73,
"crypto": 172,
"called": 102,
"info": 396,
"network": 548,
"protocol": 645,
"growth": 344,
"building": 91,
"version": 859,
"exciting": 253,
"track": 817,
"experience": 256,
"worked": 897,
"tech": 789,
"million": 522,
"investing": 409,
"early": 228,
"important": 390,
"binance": 68,
"group": 340,
"investment": 410,
"interested": 403,
"potential": 625,
"investors": 411,
"wan": 868,
"friends": 301,
"mind": 525,
"risk": 695,
"zero": 912,
"doing": 212,
"make": 490,
"saudi": 705,
"imagine": 388,
"rug": 697,
"shilling": 727,
"community": 150,
"spaces": 749,
"super": 772,
"week": 878,
"truly": 827,
"start": 757,
"trying": 831,
"connected": 156,
"simple": 736,
"mean": 506,
"contract": 161,
"man": 493,
"live": 466,
"hype": 384,
"tried": 825,
"nice": 554,
"hit": 369,
"max": 502,
"uniswap": 842,
"gas": 316,
"pretty": 631,
"cool": 162,
"months": 538,
"telegram": 790,
"saw": 707,
"today": 807,
"listed": 460,
"asking": 40,
"paid": 573,
"influencers": 395,
"beginning": 56,
"able": 0,
"create": 169,
"longer": 473,
"story": 764,
"stop": 763,
"bit": 69,
"try": 830,
"drop": 220,
"chat": 127,
"created": 170,
"account": 3,
"picture": 606,
"dumb": 222,
"question": 656,
"worth": 902,
"invest": 407,
"lot": 480,
"taking": 781,
"thank": 794,
"believe": 57,
"idea": 386,
"example": 250,
"exchange": 251,
"say": 708,
"utility": 857,
"definitely": 191,
"tokens": 810,
"answer": 31,
"enjoy": 239,
"actually": 8,
"frens": 299,
"end": 236,
"doge": 210,
"mcap": 504,
"announcement": 30,
"pair": 574,
"website": 877,
"tweet": 835,
"bags": 52,
"thing": 797,
"pushing": 655,
"spread": 753,
"word": 894,
"moon": 539,
"makes": 491,
"pepecoineth": 586,
"bought": 80,
"doubt": 218,
"sense": 720,
"case": 110,
"fine": 285,
"damn": 179,
"crazy": 168,
"bro": 86,
"awesome": 48,
"best": 60,
"seeing": 715,
"current": 176,
"lost": 479,
"shitcoin": 729,
"haha": 349,
"add": 9,
"future": 311,
"position": 620,
"means": 507,
"correct": 164,
"entire": 241,
"green": 339,
"dollar": 213,
"massive": 499,
"bull": 93,
"fuck": 304,
"stuff": 767,
"running": 700,
"nft": 552,
"morning": 540,
"red": 674,
"come": 144,
"main": 487,
"socials": 743,
"wonder": 892,
"rise": 693,
"stupid": 768,
"video": 861,
"hello": 363,
"pls": 615,
"choose": 130,
"interesting": 404,
"similar": 735,
"access": 1,
"data": 181,
"code": 137,
"sign": 733,
"refer": 675,
"cat": 112,
"initial": 398,
"bonus": 76,
"submit": 769,
"hours": 381,
"meta": 517,
"farm": 274,
"rewards": 689,
"fast": 275,
"enter": 240,
"pool": 618,
"cash": 111,
"daily": 178,
"referral": 676,
"click": 133,
"instant": 401,
"airdrop": 20,
"save": 706,
"king": 425,
"private": 634,
"key": 423,
"ideas": 387,
"issue": 413,
"financial": 283,
"effect": 233,
"entry": 242,
"target": 784,
"loss": 478,
"user": 852,
"complete": 152,
"bsc": 88,
"update": 845,
"finally": 282,
"released": 680,
"ongoing": 564,
"random": 663,
"winners": 887,
"huobi": 383,
"updates": 847,
"things": 798,
"defi": 190,
"twitter": 837,
"pepes": 597,
"note": 557,
"small": 740,
"place": 610,
"trust": 829,
"transactions": 822,
"using": 854,
"opportunity": 567,
"unique": 841,
"stay": 760,
"updated": 846,
"fairlaunch": 267,
"catch": 113,
"following": 292,
"wallets": 867,
"message": 516,
"puppy": 653,
"reason": 670,
"retweet": 686,
"chance": 120,
"billion": 66,
"popular": 619,
"giveaway": 323,
"ends": 238,
"june": 419,
"sale": 704,
"digital": 203,
"assets": 41,
"links": 457,
"channel": 122,
"youtube": 911,
"dao": 180,
"discord": 205,
"coinbase": 139,
"arbitrum": 36,
"lead": 443,
"marketing": 498,
"groups": 341,
"users": 853,
"focus": 289,
"bad": 50,
"calling": 103,
"value": 858,
"news": 550,
"added": 10,
"skrimp": 739,
"gains": 313,
"wojak": 890,
"tho": 801,
"giving": 326,
"benefits": 59,
"maybe": 503,
"black": 72,
"matter": 501,
"launching": 440,
"profit": 640,
"cause": 114,
"alot": 23,
"revolution": 687,
"presale": 629,
"set": 722,
"free": 298,
"ethereum": 246,
"built": 92,
"renounced": 682,
"driven": 219,
"incredible": 394,
"nfts": 553,
"gain": 312,
"innovative": 399,
"staking": 755,
"easy": 231,
"ride": 691,
"life": 454,
"person": 604,
"bitmart": 71,
"expect": 254,
"took": 814,
"announced": 29,
"official": 561,
"listing": 462,
"friend": 300,
"pepeki": 592,
"previous": 632,
"bet": 61,
"invested": 408,
"currency": 175,
"true": 826,
"low": 483,
"knows": 428,
"number": 559,
"shib": 724,
"looks": 476,
"stealth": 761,
"launch": 437,
"buying": 101,
"consider": 157,
"dogecoin": 211,
"shiba": 725,
"milady": 521,
"ladys": 432,
"bitcoin": 70,
"media": 508,
"bob": 75,
"fun": 307,
"audit": 44,
"tax": 786,
"billionaire": 67,
"bear": 55,
"pepega": 591,
"total": 816,
"sir": 738,
"pinksale": 608,
"contact": 158,
"bots": 79,
"fairsale": 268,
"ending": 237,
"pancakeswap": 575,
"lock": 468,
"dyor": 226,
"trump": 828,
"comes": 145,
"great": 338,
"pepeverse": 599,
"alert": 21,
"mong": 536,
"away": 47,
"pepelon": 593,
"went": 883,
"ath": 42,
"weeks": 881,
"aped": 33,
"ceo": 116,
"check": 128,
"whale": 884,
"events": 248,
"missed": 532,
"benefit": 58,
"season": 713,
"coinmarketcap": 141,
"burn": 96,
"cex": 118,
"exchanges": 252,
"bot": 78,
"coingecko": 140,
"basically": 54,
"point": 616,
"heard": 362,
"asked": 39,
"tomorrow": 812,
"polygon": 617,
"ready": 667,
"lady": 431,
"thinking": 800,
"luck": 484,
"joining": 417,
"taken": 780,
"continue": 160,
"dip": 204,
"quite": 659,
"trend": 823,
"recent": 672,
"seen": 716,
"activity": 6,
"whales": 885,
"left": 447,
"tuned": 832,
"happening": 354,
"floor": 288,
"sui": 771,
"arb": 35,
"chinese": 129,
"isn": 412,
"ppl": 628,
"year": 907,
"launches": 439,
"listings": 463,
"amazing": 26,
"grab": 337,
"rest": 685,
"break": 82,
"fair": 266,
"hands": 351,
"creating": 171,
"happen": 352,
"came": 104,
"minimum": 527,
"dead": 185,
"old": 562,
"comments": 148,
"safe": 701,
"learn": 445,
"lose": 477,
"opinion": 566,
"exactly": 249,
"having": 359,
"ecosystem": 232,
"reach": 665,
"level": 451,
"logo": 470,
"social": 742,
"favorite": 276,
"club": 135,
"competition": 151,
"prize": 635,
"joined": 416,
"hyped": 385,
"search": 712,
"open": 565,
"ordi": 569,
"head": 360,
"waiting": 865,
"turn": 834,
"hour": 380,
"bullish": 94,
"shows": 732,
"form": 296,
"increase": 393,
"minutes": 530,
"aipepe": 19,
"confirm": 154,
"fund": 308,
"pepemon": 594,
"tokenomics": 809,
"hand": 350,
"mexc": 519,
"weekend": 879,
"cmc": 136,
"pepepay": 596,
"words": 895,
"rugged": 698,
"claim": 131,
"withdraw": 889,
"recently": 673,
"goal": 328,
"gpt": 336,
"forever": 294,
"public": 647,
"launchpad": 441,
"night": 555,
"advantage": 14,
"wants": 871,
"turbo": 833,
"information": 397,
"push": 654,
"pulse": 649,
"hopefully": 377,
"eye": 261,
"related": 678,
"yesterday": 910,
"pharaoh": 605,
"research": 684,
"mpepe": 542,
"tasks": 785,
"spent": 751,
"filled": 281,
"far": 273,
"contest": 159,
"hot": 379,
"internet": 405,
"grow": 342,
"flip": 286,
"currently": 177,
"upcoming": 844,
"raised": 661,
"army": 37,
"goes": 330,
"pepebnb": 583,
"event": 247,
"charts": 126,
"pepefather": 589,
"burning": 98,
"biggest": 65,
"millions": 523,
"multiple": 543,
"safu": 702,
"huge": 382,
"tier": 803,
"dextools": 198,
"fade": 265,
"kek": 422,
"rekt": 677,
"buyers": 100,
"speak": 750,
"pizza": 609,
"wish": 888,
"fact": 264,
"participate": 576,
"communities": 149,
"pumping": 652,
"ones": 563,
"gives": 325,
"lfg": 453,
"didn": 200,
"fud": 306,
"growing": 343,
"showing": 731,
"aware": 46,
"noticed": 558,
"tweets": 836,
"gets": 321,
"listen": 461,
"wondering": 893,
"decided": 189,
"wave": 874,
"ahead": 17,
"pepeai": 581,
"happened": 353,
"order": 568,
"yea": 905,
"vision": 862,
"posting": 624,
"expected": 255,
"agree": 16,
"active": 5,
"zone": 913,
"explain": 258,
"second": 714,
"lbank": 442,
"wanted": 870,
"quick": 658,
"type": 838,
"hate": 358,
"elon": 235,
"general": 320,
"topic": 815,
"lots": 481,
"shill": 726,
"musk": 544,
"ran": 662,
"course": 166,
"posted": 623,
"including": 392,
"exposure": 259,
"global": 327,
"card": 107,
"moment": 534,
"bigger": 64,
"happens": 355,
"copy": 163,
"named": 545,
"traded": 819,
"keeps": 421,
"hoping": 378,
"history": 368,
"coz": 167,
"action": 4,
"fucking": 305,
"derivative": 192,
"mention": 515,
"promote": 644,
"wow": 903,
"deal": 186,
"imo": 389,
"given": 324,
"bring": 84,
"gone": 333,
"worry": 901,
"late": 433,
"clear": 132,
"plans": 612,
"thanks": 795,
"period": 603,
"ask": 38,
"utilities": 856,
"watching": 873,
"play": 614,
"jefe": 414,
"pepefriends": 590,
"adding": 11,
"ultimate": 839,
"ownership": 571,
"perform": 602,
"symbol": 778,
"insane": 400,
"possible": 621,
"rising": 694,
"mate": 500,
"questions": 657,
"moving": 541,
"alpha": 24,
"solid": 745,
"mission": 533,
"pull": 648,
"baby": 49,
"lately": 434,
"latest": 436,
"partnership": 577,
"actual": 7,
"cointiger": 143,
"bringing": 85,
"dear": 187,
"ama": 25,
"brought": 87,
"taxes": 787,
"known": 427,
"release": 679,
"fomo": 293,
"mil": 520,
"falling": 270,
"jump": 418,
"include": 391,
"potentially": 626,
"mania": 494,
"algorand": 22,
"perfect": 601,
"watch": 872,
"ape": 32,
"double": 217,
"epic": 243,
"care": 108,
"vibes": 860,
"page": 572,
"sort": 747,
"funny": 310,
"pink": 607,
"pepechain": 584,
"min": 524,
"lmao": 467,
"completed": 153,
"weekly": 880,
"lucky": 485,
"magic": 486,
"rocket": 696,
"couple": 165,
"especially": 244,
"god": 329,
"caused": 115,
"dumped": 224,
"non": 556,
"forward": 297,
"culture": 174,
"stand": 756,
"games": 315,
"tonight": 813,
"single": 737,
"gems": 319,
"telling": 792,
"mainnet": 488,
"cares": 109,
"levels": 452,
"tag": 779,
"followed": 291,
"pepeswap": 598,
"mark": 495,
"honest": 374,
"nfa": 551,
"leader": 444,
"eligible": 234,
"stole": 762,
"channels": 123,
"extra": 260,
"finding": 284,
"mini": 526,
"ways": 876,
"honestly": 375,
"bored": 77,
"supposed": 775,
"press": 630,
"character": 124,
"strategy": 765,
"guccipepe": 345,
"pepezilla": 600,
"mcpepe": 505,
"metaverse": 518
}
}
}