moderation-topics

This is a BERTopic model. BERTopic is a flexible and modular topic modeling framework that allows for the generation of easily interpretable topics from large datasets.

Usage

To use this model, please install BERTopic:

pip install -U bertopic

You can use the model as follows:

from bertopic import BERTopic
topic_model = BERTopic.load("jaimevera1107/moderation-topics")

topic_model.get_topic_info()

Topic overview

  • Number of topics: 94
  • Number of training documents: 1403
Click here for an overview of all topics.
Topic ID Topic Keywords Topic Frequency Label
0 suicide - nssi - tendency - recent - self 40 0_suicide_nssi_tendency_recent
1 exposed - minimal - sexualized - possessing - performs 33 1_exposed_minimal_sexualized_possessing
2 drug - reference - purposes - substances - substance 32 2_drug_reference_purposes_substances
3 regulated - consumption - tobacco - relate - associate 31 3_regulated_consumption_tobacco_relate
4 male - region - pubic - exposure - nipple 31 4_male_region_pubic_exposure
5 testing - wildlife - endangered - poaching - hunting 31 5_testing_wildlife_endangered_poaching
6 nudity - fine - implied - documentaries - indigenous 30 6_nudity_fine_implied_documentaries
7 text - language - pickup - textual - texts 28 7_text_language_pickup_textual
8 fighting - incitement - violent - reactive - event 27 8_fighting_incitement_violent_reactive
9 hate - ideology - hateful - based - disability 27 9_hate_ideology_hateful_based
10 sensual - pleasure - demonstration - objectification - dialogue 26 10_sensual_pleasure_demonstration_objectification
11 detailing - stimulation - fetishism - allusion - adults 26 11_detailing_stimulation_fetishism_allusion
12 pornography - vulgarity - website - tapes - softcore 26 12_pornography_vulgarity_website_tapes
13 lead - highly - is - imitable - professionals 25 13_lead_highly_is_imitable
14 brand - code - csam - qr - multiple 25 14_brand_code_csam_qr
15 expressions - dance - performing - performances - express 24 15_expressions_dance_performing_performances
16 intellectual - copyright - copyrighted - stolen - cover 24 16_intellectual_copyright_copyrighted_stolen
17 slur - slurs - designation - remarks - status 24 17_slur_slurs_designation_remarks
18 undressing - striptease - process - panties - voyeuristic 23 18_undressing_striptease_process_panties
19 workplace - peeping - upskirting - tom - coercion 23 19_workplace_peeping_upskirting_tom
20 hostility - degradation - statement - discriminatory - characteristics 23 20_hostility_degradation_statement_discriminatory
21 low - quality - organic - host - grow 22 21_low_quality_organic_host
22 terrorist - terrorism - recruitment - organizations - international 21 22_terrorist_terrorism_recruitment_organizations
23 spam - jump - makeup - scary - scare 20 23_spam_jump_makeup_scary
24 firearms - ammunition - explosive - explosives - weapons 20 24_firearms_ammunition_explosive_explosives
25 culturally - appropriate - wear - protected - not 19 25_culturally_appropriate_wear_protected
26 disturbing - cannibalism - disgusting - coverage - anatomy 18 26_disturbing_cannibalism_disgusting_coverage
27 homicide - mutilated - death - accident - torture 18 27_homicide_mutilated_death_accident
28 privacy - invasion - surveillance - espionage - confidential 18 28_privacy_invasion_surveillance_espionage
29 age - requirement - signals - identifiers - admission 18 29_age_requirement_signals_identifiers
30 framing - gaze - angles - piercings - camera 17 30_framing_gaze_angles_piercings
31 stalking - doxing - lists - encourage - addresses 17 31_stalking_doxing_lists_encourage
32 damage - destruction - property - arson - vandalism 17 32_damage_destruction_property_arson
33 eating - disorders - disorder - eat - loss 16 33_eating_disorders_disorder_eat
34 bullying - statements - cyberbullying - vulnerable - users 16 34_bullying_statements_cyberbullying_vulnerable
35 scams - frauds - scamming - schemes - fraudulent 16 35_scams_frauds_scamming_schemes
36 criminal - crime - criminals - gang - burglary 15 36_criminal_crime_criminals_gang
37 identifiable - data - personally - reveal - others 15 37_identifiable_data_personally_reveal
38 work - sex - prostitution - workers - escort 15 38_work_sex_prostitution_workers
39 conspiracy - theories - disinformation - baseless - current 14 39_conspiracy_theories_disinformation_baseless
40 consensual - recording - blackmail - intention - displaying 14 40_consensual_recording_blackmail_intention
41 child - featuring - pedophilic - defense - intimate 14 41_child_featuring_pedophilic_defense
42 polarization - opposing - social - incite - deepen 14 42_polarization_opposing_social_incite
43 pedophilia - grooming - normalization - predators - normalizing 14 43_pedophilia_grooming_normalization_predators
44 platforms - direction - ads - third - party 14 44_platforms_direction_ads_third
45 products - items - enhancement - grafitication - demonstrations 13 45_products_items_enhancement_grafitication
46 possession - consuming - drinking - tobacco - smoking 13 46_possession_consuming_drinking_tobacco
47 credible - threats - menacing - aggressive - plans 12 47_credible_threats_menacing_aggressive
48 hacking - malware - phishing - ransomware - hacks 12 48_hacking_malware_phishing_ransomware
49 proxy - lgbtq - bully - harassment - trolling 12 49_proxy_lgbtq_bully_harassment
50 going - live - 13 - 18 - u18 12 50_going_live_13_18
51 unintentionally - genitalia - animals - pornographic - bestiality 12 51_unintentionally_genitalia_animals_pornographic
52 artificial - traffic - way - methods - generate 12 52_artificial_traffic_way_methods
53 slaughter - mutilation - humans - dead - animal 12 53_slaughter_mutilation_humans_dead
54 goods - gangs - organized - counterfeit - illicit 11 54_goods_gangs_organized_counterfeit
55 gambling - betting - cheating - game - devices 11 55_gambling_betting_cheating_game
56 trafficking - forced - coerced - traded - function 11 56_trafficking_forced_coerced_traded
57 unsolicited - messages - favors - requests - advances 11 57_unsolicited_messages_favors_requests
58 blood - gore - shock - bloodshed - value 11 58_blood_gore_shock_bloodshed
59 victim - abduction - vehicle - motor - glorification 11 59_victim_abduction_vehicle_motor
60 inappropriate - kiss - sexualizes - objectifies - towards 10 60_inappropriate_kiss_sexualizes_objectifies
61 toddlers - infants - unintentional - touch - abdomen 10 61_toddlers_infants_unintentional_touch
62 traditional - traditions - sacred - cultural - misappropriation 10 62_traditional_traditions_sacred_cultural
63 nuclear - weapon - peaceful - advocating - energy 9 63_nuclear_weapon_peaceful_advocating
64 exploiting - child - marriage - exploitation - labor 9 64_exploiting_child_marriage_exploitation
65 impersonation - famous - figure - slandering - profiles 9 65_impersonation_famous_figure_slandering
66 defamation - someones - defamatory - allegations - businesses 9 66_defamation_someones_defamatory_allegations
67 recipes - creating - may - tools - instructions 9 67_recipes_creating_may_tools
68 election - interference - campaigns - misinformation - political 9 68_election_interference_campaigns_misinformation
69 claims - expertise - apocalypse - authority - media 9 69_claims_expertise_apocalypse_authority
70 featuring - nude - partial - implied - depictions 8 70_featuring_nude_partial_implied
71 operations - police - military - enforcement - law 8 71_operations_police_military_enforcement
72 tax - laundering - crimes - money - ponzi 8 72_tax_laundering_crimes_money
73 cosmetic - surgery - procedures - diy - unlicensed 8 73_cosmetic_surgery_procedures_diy
74 subject - optical - innuendos - illusion - suggestive 8 74_subject_optical_innuendos_illusion
75 bodies - fantasy - lifeless - accident - fictional 8 75_bodies_fantasy_lifeless_accident
76 controversial - constructive - politics - issues - discussion 7 76_controversial_constructive_politics_issues
77 kissing - lip - only - greeting - as 7 77_kissing_lip_only_greeting
78 pirated - plagiarism - incites - glorifies - first 7 78_pirated_plagiarism_incites_glorifies
79 mental - conditions - health - mocks - stigmatization 7 79_mental_conditions_health_mocks
80 daredevil - reckless - precautions - risking - caution 7 80_daredevil_reckless_precautions_risking
81 pranks - intentions - cybersecurity - harmful - targeted 7 81_pranks_intentions_cybersecurity_harmful
82 dark - web - underground - marketplaces - glorifies 6 82_dark_web_underground_marketplaces
83 vax - anti - medical - false - misinformation 6 83_vax_anti_medical_false
84 sports - danger - adventures - stunts - professional 6 84_sports_danger_adventures_stunts
85 environmental - pollution - experiments - ecosystems - natural 6 85_environmental_pollution_experiments_ecosystems
86 incest - incestuous - taboo - themes - discussion 5 86_incest_incestuous_taboo_themes
87 neglect - child - endangerment - abuse - physical 5 87_neglect_child_endangerment_abuse
88 radicalization - extremist - extremism - views - propaganda 5 88_radicalization_extremist_extremism_views
89 waste - bodily - excretion - unsanitary - images 5 89_waste_bodily_excretion_unsanitary
90 emotional - psychological - mind - gaslighting - relationships 5 90_emotional_psychological_mind_gaslighting
91 solicitation - offer - request - prostitution - act 5 91_solicitation_offer_request_prostitution
92 elderly - elders - elder - neglect - against 5 92_elderly_elders_elder_neglect
93 education - terms - term - relating - general 4 93_education_terms_term_relating

Training hyperparameters

  • calculate_probabilities: False
  • language: english
  • low_memory: False
  • min_topic_size: 10
  • n_gram_range: (1, 1)
  • nr_topics: None
  • seed_topic_list: None
  • top_n_words: 10
  • verbose: False

Framework versions

  • Numpy: 1.23.5
  • HDBSCAN: 0.8.33
  • UMAP: 0.5.4
  • Pandas: 1.5.3
  • Scikit-Learn: 1.2.2
  • Sentence-transformers: 2.2.2
  • Transformers: 4.24.0
  • Numba: 0.58.1
  • Plotly: 5.15.0
  • Python: 3.10.12
Downloads last month
1,055
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.