Research map / Detection and segmentation / Open-set and few-shot detection
Scalable and Generalizable Social Bot Detection through Data Selection
Kai-Cheng Yang, Onur Varol, Pik-Mai Hui, Filippo Menczer
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Select a subset of training datasets rather than train on all candidate data (ablation) | Random-forest bot detector (M196) · midterm-18 | AUC: 0.83 -> 0.99 | Improved |
| Select a subset of training datasets rather than train on all candidate data (ablation) | Random-forest bot detector (M196) · botwiki-verified | AUC: 0.91 -> 0.99 | Improved |
| Select a subset of training datasets rather than train on all candidate data (ablation) | Random-forest bot detector (M196) · 5-fold cross-validation | AUC: 0.95 -> 0.98 | Improved |
| Select a subset of training datasets rather than train on all candidate data (ablation) | Random-forest bot detector · cresci-rtbust | AUC: 0.57 -> 0.60 | Mixed |
| Use selected training datasets instead of all candidate data (ablation) | Random-forest bot detector · gilani-17 | AUC: 0.64 -> 0.68 | Improved |
| Train using selected datasets rather than all candidate datasets (ablation) | Random-forest bot detector · 100,000 random accounts | Spearman's r with Botometer: 0.60 -> 0.60 | No clear effect |
| Use only user metadata and derived profile features instead of rich contextual information | Proposed bot-detection framework · Twitter public stream | processing rate: almost 900M tweets per day | Improved |
| Include account age directly as a model feature (ablation) | Proposed bot-detection framework · Bot detection datasets | accuracy | Hurt |
| Choose a threshold using cross-domain validation rather than cross-validation alone | M196 · Cross-domain testing | F1: F_1=0.94 -> F_1=0.77 | Mixed |
Closest papers by meaning
- BotSim: LLM-Powered Malicious Social Botnet Simulation (AAAI 2025 · similarity 0.85)
- SeGA: Preference-Aware Self-Contrastive Learning with Prompts for Anomalous User Detection on Twitter (AAAI 2024 · similarity 0.83)
- Exploring and Mitigating Adversarial Manipulation of Voting-Based Leaderboards (ICML 2025 · similarity 0.82)
- Adversarial Socialbots Modeling Based on Structural Information Principles (AAAI 2024 · similarity 0.82)
- A Multi-Task Benchmark for Abusive Language Detection in Low-Resource Settings (NeurIPS 2025 · similarity 0.80)
- IOHunter: Graph Foundation Model to Uncover Online Information Operations (AAAI 2025 · similarity 0.80)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
