Fake News Detection on Politifact
96.9AccuracyDEBERTA
Evaluation results
| Method | Links | |||||
|---|---|---|---|---|---|---|
| DEBERTAevidence-enhanced=true, cross-validation=ten-fold2024.10 | 96.9 | — | — | — | — | |
| DEBERTAevidence-enhanced=false, cross-validation=ten-fold2024.10 | 96.6 | — | — | — | — | |
| TEGRAGraph Extraction=KGI, Triple Selection=False2026.02 | 96.52 | — | — | 96.39 | — | |
| AVOIDCategory=Our Method2026.01 | 96 | 95.6 | 96 | 95.8 | — | |
| TEGRAGraph Extraction=OpenIE6, Triple Selection=True2026.02 | 95.83 | — | — | 95.52 | — | |
| TEGGraph Extraction=OpenIE62026.02 | 95 | — | — | 94.68 | — | |
| TEGRAGraph Extraction=OpenIE6, Triple Selection=False2026.02 | 95 | — | — | 94.58 | — | |
| BERTevidence-enhanced=true, cross-validation=ten-fold2024.10 | 94.7 | — | — | — | — | |
| GETevidence-enhanced=true, cross-validation=ten-fold2024.10 | 94.2 | — | — | — | — | |
| TEGRAGraph Extraction=KGI, Triple Selection=True2026.02 | 94.2 | — | — | 94.69 | — | |
| BERTevidence-enhanced=false, cross-validation=ten-fold2024.10 | 94 | — | — | — | — | |
| TEGGraph Extraction=KGI2026.02 | 93.62 | — | — | 94.08 | — | |
| PSS-TLBackbone GNN=GCN2026.04 | 93.23 | — | — | — | 92.47 | |
| PSS-TLBackbone GNN=GAT2026.04 | 93.05 | — | — | — | 92.94 | |
| RoBERTaConfiguration=Baselines2026.02 | 92.97 | — | — | 91.86 | — | |
| PSS-TLBackbone GNN=GraphSAGE2026.04 | 92.83 | — | — | — | 92.86 | |
| ARGCategory=LLM-Enhanced2026.01 | 92.6 | 92.4 | 92.8 | 92.4 | — | |
| BREAKCategory=Content-Based2026.01 | 92 | 92.1 | 92 | 92.1 | — | |
| CoALFake2026.04 | 92 | 91 | 88 | 93 | — | |
| ZoFia (DeepSeek-v3)Category=Zero-shot, LLM Usage=LLM-based2025.11 | 91.52 | — | — | 87.88 | — | |
| UPSR2026.04 | 91.46 | — | — | — | 90.86 | |
| SheepDogCategory=LLM-Enhanced2026.01 | 91.3 | 91.2 | 90.9 | 91.1 | — | |
| DECORBackbone GNN=GAT2026.04 | 91.29 | — | — | — | 91.03 | |
| BERTConfiguration=Baselines2026.02 | 91 | — | — | 91.89 | — | |
| SLFEND2026.04 | 91 | 87 | 84 | 93 | — | |
| CSFNDCategory=Content-Based2026.01 | 90.7 | 90.7 | 90.7 | 90.7 | — | |
| neuro-fuzzy model2026.05 | 90.3 | — | — | 90.1 | — | |
| CoALFakeAblation=(-) Domain-shared2026.04 | 90 | 89 | 86 | 91 | — | |
| HYPHENevidence-enhanced=true, cross-validation=ten-fold2024.10 | 89.9 | — | — | — | — | |
| GenFENDCategory=LLM-Enhanced2026.01 | 89.9 | 89.7 | 90 | 89.8 | — | |
| AdstyleCategory=LLM-Enhanced2026.01 | 89.9 | 90.4 | 89.9 | 90.1 | — | |
| EBGCN2026.04 | 89.87 | — | — | — | 89.29 | |
| Auto-CoTCategory=Zero-shot, LLM Usage=LLM-based2025.11 | 89.65 | — | — | 73.67 | — | |
| HMCANCategory=Content-Based2026.01 | 89.4 | 89.8 | 89.2 | 89.6 | — | |
| L-DefenseCategory=LLM-Enhanced2026.01 | 89.3 | 90.2 | 89.9 | 89.6 | — | |
| FuDFEND2026.04 | 89 | 88 | 84 | 90 | — | |
| DECORBackbone GNN=GCN2026.04 | 88.71 | — | — | — | 88.47 | |
| ALGMCategory=Content-Based2026.01 | 88.7 | 89.5 | 88.1 | 88.6 | — | |
| DECORBackbone GNN=GraphSAGE2026.04 | 88.39 | — | — | — | 88.12 | |
| MIMOE-FNDCategory=Content-Based2026.01 | 87.8 | 89.2 | 88.3 | 88.6 | — | |
| CoALFakeAblation=(-) Domain-specific2026.04 | 87 | 87 | 83 | 88 | — | |
| DKFNDCategory=Few-shot, LLM Usage=LLM-assisted2025.11 | 87 | — | — | 82.43 | — | |
| Tsetlin machineConfiguration=Baselines2026.02 | 86 | — | — | 87.04 | — | |
| UPFDBackbone GNN=GraphSAGE2026.04 | 85.81 | — | — | — | 85.6 | |
| HANCategory=Content-Based2026.01 | 85.7 | 85.9 | 85.4 | 85.6 | — | |
| MDFEND2026.04 | 85 | 87 | 84 | 85 | — | |
| DITFEND2026.04 | 85 | 82 | 80 | 85 | — | |
| XLNetModality=Text-Only2026.06 | 84.7 | 90.5 | 70.4 | 79.2 | — | |
| BiGCN2026.04 | 84.62 | — | — | — | 82.98 | |
| MCANAuthors=Yu et al. 20192026.06 | 84.6 | 85.1 | 82.9 | 84 | — | |
| DEFENDevidence-enhanced=true, cross-validation=ten-fold2024.10 | 84.3 | — | — | — | — | |
| BERTModality=Text-Only2026.06 | 84.3 | 90.4 | 69.6 | 78.6 | — | |
| ZoFia (Qwen3-32B)Category=Zero-shot, LLM Usage=LLM-based2025.11 | 84.06 | — | — | 81.76 | — | |
| EDDFN2026.04 | 84 | 83 | 83 | 83 | — | |
| TELLERLLM=Llama2 (7B)2024.02 | 83.82 | — | — | — | 83.81 | |
| TELLER w/ InterventionLLM=Llama2 (7B)2024.02 | 83.82 | — | — | — | 83.82 | |
| UPFDBackbone GNN=GCN2026.04 | 82.99 | — | — | — | 82.69 | |
| UPFDBackbone GNN=GAT2026.04 | 82.99 | — | — | — | 82.9 | |
| GraphSAGE2026.04 | 82.81 | — | — | — | 82.68 | |
| GAT2026.04 | 82.72 | — | — | — | 82.61 | |
| dEFENDCategory=Content-Based2026.01 | 82.5 | 83 | 82.1 | 82.3 | — | |
| GCN2026.04 | 81.63 | — | — | — | 81.46 | |
| MCAN-AAuthors=Yu et al. 20192026.06 | 80.9 | 82.7 | 75.8 | 79.2 | — | |
| FactAgentCategory=Zero-shot, LLM Usage=LLM-based2025.11 | 80.59 | — | — | 70.06 | — | |
| BERTCategory=Content-Based2026.01 | 80.1 | 79.9 | 80.1 | 79.8 | — | |
| TELLERLLM=Llama2 (13B)2024.02 | 79.41 | — | — | — | 79.41 | |
| SAFE2026.04 | 79 | 78 | 77 | 77 | — | |
| DeepSeek-v3Category=Zero-shot, LLM Usage=Only-LLM2025.11 | 78.99 | — | — | 40.24 | — | |
| TELLER w/ InterventionLLM=Llama2 (13B)2024.02 | 78.68 | — | — | — | 78.65 | |
| LLMConfiguration=Baselines, Shots=0-shot2026.02 | 78 | — | — | 78.85 | — | |
| AE2026.04 | 78 | 78 | 77 | 77 | — | |
| CoALFakeAblation=(-) Cross-Attention2026.04 | 78 | 77 | 61 | 85 | — | |
| DirectLLM=Llama2 (13B)2024.02 | 77.94 | — | — | — | 77.1 | |
| Web Retrieval AgentsCategory=Zero-shot, LLM Usage=LLM-based2025.11 | 77.83 | — | — | 64.88 | — | |
| CNN (Text)Modality=Text-Only2026.06 | 77.6 | 85.7 | 61.4 | 71.5 | — | |
| CAFECategory=Content-Based2026.01 | 77.3 | 75.4 | 78 | 75.9 | — | |
| DirectLLM=Llama2 (7B)2024.02 | 77.21 | — | — | — | 77 | |
| CAPE-FNDCategory=Zero-shot, LLM Usage=LLM-based2025.11 | 76.81 | — | — | 62.43 | — | |
| Qwen3-32BCategory=Zero-shot, LLM Usage=Only-LLM2025.11 | 76.57 | — | — | 43.35 | — | |
| ZoFia (GPT-4o-mini)Category=Zero-shot, LLM Usage=LLM-based2025.11 | 75.28 | — | — | 75.19 | — | |
| ReAct (Search API)Category=Zero-shot, LLM Usage=LLM-based2025.11 | 74.73 | — | — | 67.64 | — | |
| ARGCategory=Few-shot, LLM Usage=LLM-assisted2025.11 | 74 | — | — | 67.16 | — | |
| GPT-4o-miniCategory=Zero-shot, LLM Usage=Only-LLM2025.11 | 73.58 | — | — | 41.66 | — | |
| GPT-3.5 turboMode=Prompting2026.04 | 73 | 73 | 73 | 73 | — | |
| LLMConfiguration=Baselines, Shots=3-shots2026.02 | 70 | — | — | 71.15 | — | |
| HPNF + LIWCFeatures=LIWC2026.04 | 70 | 72 | 70 | 71 | — | |
| PSMCategory=Few-shot, LLM Usage=Non-LLM2025.11 | 70 | — | — | 49.15 | — | |
| CNN (Image)Modality=Image-Only2026.06 | 69.2 | 82.1 | 73.3 | 77.4 | — | |
| TELLER w/ InterventionLLM=FLAN-T5-xl (3B)2024.02 | 69.12 | — | — | — | 68.79 | |
| TELLERLLM=FLAN-T5-xxl (11B)2024.02 | 69.12 | — | — | — | 68.57 | |
| TELLER w/ InterventionLLM=FLAN-T5-xxl (11B)2024.02 | 69.12 | — | — | — | 68.95 | |
| HPNF2026.04 | 69 | 69 | 68 | 68 | — | |
| TELLERLLM=FLAN-T5-xl (3B)2024.02 | 68.82 | — | — | — | 64.68 | |
| TELLER w/ InterventionLLM=FLAN-T5-large (780M)2024.02 | 68.38 | — | — | — | 68.29 | |
| TELLERLLM=FLAN-T5-large (780M)2024.02 | 67.65 | — | — | — | 67.65 | |
| MDFENDCategory=Few-shot, LLM Usage=Non-LLM2025.11 | 65.5 | — | — | 62.3 | — | |
| HiSSCategory=Zero-shot, LLM Usage=LLM-based2025.11 | 64.82 | — | — | 56.8 | — | |
| CHATGPTevidence-enhanced=true, cross-validation=ten-fold2024.10 | 62.2 | — | — | — | — | |
| MISTRALprompting-style=Vanilla (VaN), cross-validation=ten-fold2024.10 | 61.2 | — | — | — | — | |
| KPLCategory=Few-shot, LLM Usage=Non-LLM2025.11 | 58.33 | — | — | 60.4 | — |