Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

REDDIX-NET: A Novel Dataset and Benchmark for Moderating Online Explicit Services

MSVPJ Sathvik, Manan Roy Choudhury, Rishita Agarwal, Sathwik Narkedimilli, Vivek Gupta

TL;DR

ReddiX-NET introduces a novel, multilingual dataset and benchmark for moderating online sexual services on Reddit, addressing limitations of traditional NSFW filters through a six-category taxonomy. It combines data collection, careful annotation with substantial inter-annotator agreement, and rigorous NLP analyses using state-of-the-art LLMs and PLMs to classify users, analyze expressions, and examine comments and time-based engagement. The study reveals strengths and challenges of current AI moderation approaches, demonstrates the utility of ensemble methods, and provides insights into psychological and social implications as well as temporal dynamics for intervention. Ethical safeguards, privacy-preserving protocols, and a restricted data release plan accompany the benchmark to promote responsible research and practical deployment in platform moderation and policy contexts.

Abstract

The rise of online platforms has enabled covert illicit activities, including online prostitution, to pose challenges for detection and regulation. In this study, we introduce REDDIX-NET, a novel benchmark dataset specifically designed for moderating online sexual services and going beyond traditional NSFW filters. The dataset is derived from thousands of web-scraped NSFW posts on Reddit and categorizes users into six behavioral classes reflecting different service offerings and user intentions. We evaluate the classification performance of state-of-the-art large language models (GPT-4, LlaMA 3.3-70B-Instruct, Gemini 1.5 Flash, Mistral 8x7B, Qwen 2.5 Turbo, Claude 3.5 Haiku) using advanced quantitative metrics, finding promising results with models like GPT-4 and Gemini 1.5 Flash. Beyond classification, we conduct sentiment and comment analysis, leveraging LLM and PLM-based approaches and metadata extraction to uncover behavioral and temporal patterns. These analyses reveal peak engagement times and distinct user interaction styles across categories. Our findings provide critical insights into AI-driven moderation and enforcement, offering a scalable framework for platforms to combat online prostitution and associated harms.

REDDIX-NET: A Novel Dataset and Benchmark for Moderating Online Explicit Services

TL;DR

ReddiX-NET introduces a novel, multilingual dataset and benchmark for moderating online sexual services on Reddit, addressing limitations of traditional NSFW filters through a six-category taxonomy. It combines data collection, careful annotation with substantial inter-annotator agreement, and rigorous NLP analyses using state-of-the-art LLMs and PLMs to classify users, analyze expressions, and examine comments and time-based engagement. The study reveals strengths and challenges of current AI moderation approaches, demonstrates the utility of ensemble methods, and provides insights into psychological and social implications as well as temporal dynamics for intervention. Ethical safeguards, privacy-preserving protocols, and a restricted data release plan accompany the benchmark to promote responsible research and practical deployment in platform moderation and policy contexts.

Abstract

The rise of online platforms has enabled covert illicit activities, including online prostitution, to pose challenges for detection and regulation. In this study, we introduce REDDIX-NET, a novel benchmark dataset specifically designed for moderating online sexual services and going beyond traditional NSFW filters. The dataset is derived from thousands of web-scraped NSFW posts on Reddit and categorizes users into six behavioral classes reflecting different service offerings and user intentions. We evaluate the classification performance of state-of-the-art large language models (GPT-4, LlaMA 3.3-70B-Instruct, Gemini 1.5 Flash, Mistral 8x7B, Qwen 2.5 Turbo, Claude 3.5 Haiku) using advanced quantitative metrics, finding promising results with models like GPT-4 and Gemini 1.5 Flash. Beyond classification, we conduct sentiment and comment analysis, leveraging LLM and PLM-based approaches and metadata extraction to uncover behavioral and temporal patterns. These analyses reveal peak engagement times and distinct user interaction styles across categories. Our findings provide critical insights into AI-driven moderation and enforcement, offering a scalable framework for platforms to combat online prostitution and associated harms.

Paper Structure

This paper contains 18 sections, 29 figures, 14 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. ReddiX-NET Construction
  4. Data Collection
  5. Data Annotation
  6. Statistical Analysis
  7. NLP-Driven Analysis of ReddiX-NET
  8. ReddiX-NET User Classification
  9. ReddiX-NET Expression Analysis
  10. ReddiX-NET Comments Classification
  11. Time-based Analysis on ReddiX-NET
  12. Results and Analysis
  13. ReddiX-NET User Classification
  14. ReddiX-NET Expression Analysis
  15. Psychological and Social Implications:
  16. ...and 3 more sections

Figures (29)

  • Figure 1: Structure of the proposed dataset, categorizing online sexual services across six distinct categories and their further subdivisions.
  • Figure 2: Distribution of sentiment classifications across six different service categories (VS, PS, MF, Ex, CGI, CCS).
  • Figure 3: Category-wise impact proportion highlighting the proportion of emotional dependency, exploitation, mental health concerns, neutral perceptions, and positive experiences across different online prostitution categories. This provides insights into the psychological and socio-emotional consequences associated with various engagement types
  • Figure 4: Illustration of comment activity trends over different hourly ranges in a day, highlighting peak engagement times for Channels A, B, and C. The x-axis categorizes the day into six time periods, while the y-axis measures the proportion of total posts within each window.
  • Figure 5: Visualization of temporal posts activity patterns across distinct hourly intervals, emphasizing peak engagement periods for Channels A, B, and C.
  • ...and 24 more figures