Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

TM-RUGPULL: A Temporary Sound, Multimodal Dataset for Early Detection of RUG Pulls Across the Tokenized Ecosystem

Fatemeh Shoaei, Mohammad Pishdar, Mozafar Bag-Mohammadi, Mojtaba Karami

TL;DR

TM-RugPull is presented, a rigorously curated, leakage-resistant dataset of 1,028 token projects spanning DeFi, meme coins, NFTs, and celebrity-themed tokens that enables causally valid, multimodal analysis of rug-pull dynamics and establishes a new benchmark for reproducible fraud detection research.

Abstract

Rug-pull attacks pose a systemic threat across the blockchain ecosystem, yet research into early detection is hindered by the lack of scientific-grade datasets. Existing resources often suffer from temporal data leakage, narrow modality, and ambiguous labeling, particularly outside DeFi contexts. To address these limitations, we present TM-RugPull, a rigorously curated, leakage-resistant dataset of 1,028 token projects spanning DeFi, meme coins, NFTs, and celebrity-themed tokens. RugPull enforces strict temporal hygiene by extracting all features on chain behavior, smart contract metadata, and OSINT signals strictly from the first half of each project's lifespan. Labels are grounded in forensic reports and longevity criteria, verified through multi-expert consensus. This dataset enables causally valid, multimodal analysis of rug-pull dynamics and establishes a new benchmark for reproducible fraud detection research.

TM-RUGPULL: A Temporary Sound, Multimodal Dataset for Early Detection of RUG Pulls Across the Tokenized Ecosystem

TL;DR

TM-RugPull is presented, a rigorously curated, leakage-resistant dataset of 1,028 token projects spanning DeFi, meme coins, NFTs, and celebrity-themed tokens that enables causally valid, multimodal analysis of rug-pull dynamics and establishes a new benchmark for reproducible fraud detection research.

Abstract

Rug-pull attacks pose a systemic threat across the blockchain ecosystem, yet research into early detection is hindered by the lack of scientific-grade datasets. Existing resources often suffer from temporal data leakage, narrow modality, and ambiguous labeling, particularly outside DeFi contexts. To address these limitations, we present TM-RugPull, a rigorously curated, leakage-resistant dataset of 1,028 token projects spanning DeFi, meme coins, NFTs, and celebrity-themed tokens. RugPull enforces strict temporal hygiene by extracting all features on chain behavior, smart contract metadata, and OSINT signals strictly from the first half of each project's lifespan. Labels are grounded in forensic reports and longevity criteria, verified through multi-expert consensus. This dataset enables causally valid, multimodal analysis of rug-pull dynamics and establishes a new benchmark for reproducible fraud detection research.
Paper Structure (15 sections, 6 figures, 2 tables)

This paper contains 15 sections, 6 figures, 2 tables.

Table of Contents

  1. Introduction
  2. Background
  3. Related Work
  4. Dataset Construction
  5. Source and Scope
  6. Labeling Protocol
  7. Temporal Boundary
  8. Manual Curation and Quality Control
  9. Data Collection Pipeline
  10. Dataset Schema
  11. Visual Overview of TM-RugPull
  12. Illustrative Benchmarking
  13. Design Rationale and Integrity Guarantees
  14. Data Availability
  15. Conclusion

Figures (6)

  • Figure 1: Data collection and processing pipeline for TM-RugPull . Information is aggregated from blockchain platforms, open-source intelligence sources, and security reports, then temporally aligned and verified to construct a leakage-resistant dataset for early-stage rug-pull analysis.
  • Figure 2: Category distribution of projects in TM-RugPull . The dataset encompasses a diverse set of token types, including DeFi protocols, meme coins, NFT-based games, and celebrity-themed tokens, reflecting the heterogeneous nature of rug-pull threats beyond narrow DeFi contexts.
  • Figure 3: Comparison of key on-chain metrics between scam and legitimate projects in TM-RugPull . Scam tokens exhibit significantly higher token concentration and holder variance among top 1% holders, providing robust structural signals for early detection.
  • Figure 4: Temporal alignment of OSINT and on-chain signals for a sample scam project. Social media activity (Twitter volume) and search interest peak well before the project midpoint (dashed line), while price remains stable—validating that predictive signals exist early and are captured in TM-RugPull .
  • Figure 5: Distribution of predicted scam probabilities under identical classification pipelines. The full benchmark exhibits clearer separation between scam and legitimate projects compared to the DeFi-only subset, indicating improved probability calibration and class separability induced by cross-domain and multimodal data coverage.
  • ...and 1 more figures