List your company

Entity Disambiguation on ZELDA Benchmark (test)

91.1AIDA-BCL-RECALL
62.39669.84877.384.752Apr 2, 2024Jul 18, 2024Nov 3, 2024Feb 19, 2025Jun 6, 2025Sep 22, 2025Jan 8, 2026

Evaluation results

MethodLinks
CL-RECALLType=Baseline (Recall of candidate list)2024.0491.19498.498.392.498.898.856.773.189.1
Upper boundEvaluation protocol=Baseline2026.0191.19498.498.392.498.898.856.773.189.1
VERBALIZED + iter. trainingTrain data=Z2025.0588.278.992.288.471.590.898.266.365.982.3
VERBALIZEDTrain data=Z, iterative training=true2025.0588.278.992.288.471.590.898.266.365.982.3
VERBALIZED ITEvaluation protocol=Classical (with training data)2026.0188.278.992.288.471.590.898.266.365.982.3
VERBALIZED + iter. predictionTrain data=Z2025.0584.478.489.587.170.491.298.765.367.281.4
BLINK-OG crossTrain data=B2025.0584.282.492.891.277.382.399.264.874.283.2
LELAEvaluation protocol=True Zero-Shot, LLM selection model=Qwen3-30B-A3B2026.0184.279.183.984.664.887.597.348.86777.5
LELAEvaluation protocol=True Zero-Shot, LLM selection model=Qwen3-4B2026.01847884.182.666.385.197.148.26576.7
LELAEvaluation protocol=True Zero-Shot, LLM selection model=Magistral-Small-25092026.0183.379.180.779.566.787.597.248.966.776.6
VERBALIZEDTrain data=Z2025.0582.678.989.286.269.891.498.665.36781
VERBALIZEDTrain data=Z, iterative training=false2025.0582.678.989.286.269.891.498.665.36781
LUKEPCategory=Classification, Mode=Fine-tuned2024.0481.277.981.578.570.376.59822.551.871
LUKE_FTTrain data=Z2025.0581.277.981.578.570.376.59822.551.871
LUKE FTEvaluation protocol=Classical (with training data)2026.0181.277.981.578.570.376.59822.551.871
BLINK-OG biTrain data=B2025.0580.677.390.887.868.279.897.950.157.376.6
FUSIONEDCategory=Generative2024.0480.181.493.992.373.68998.341.557.978.7
FUSIONEDTrain data=Z2025.0580.181.493.992.373.68998.341.557.978.7
FusionEDEvaluation protocol=Classical (with training data)2026.0180.181.493.992.373.68998.341.557.978.7
FEVRYCategory=Classification, Configuration=CL (Using candidate lists)2024.0479.576.98986.570.384.587.631.947.772.7
FEVRY_CLTrain data=Z2025.0579.576.98986.570.384.587.631.947.772.7
FEVRY CLEvaluation protocol=Classical (with training data)2026.0179.576.98986.570.384.587.631.947.772.7
LUKEPCategory=Classification, Mode=Pre-trained2024.0479.373.876.169.966.868.497.720.450.867
LUKE_PRETrain data=Z2025.0579.373.876.169.966.868.497.720.450.867
FEVRYCategory=Classification, Configuration=ALL (Restricted generation process)2024.0479.271.888.584.16884.363.843.453.170.7
FEVRY_ALLTrain data=Z2025.0579.271.888.584.16884.363.843.453.170.7
GENRECategory=Generative, Configuration=CL (Using candidate lists)2024.0478.680.192.891.573.688.499.637.352.877.2
GENRE_CLTrain data=Z2025.0578.680.192.891.573.688.499.637.352.877.2
GENRE CLEvaluation protocol=Classical (with training data)2026.0178.680.192.891.573.688.499.637.352.877.2
LELAEvaluation protocol=True Zero-Shot, LLM selection strategy=None (w/o LLM selection)2026.0176.475.17979.96576.29743.660.672.5
GENRECategory=Generative, Configuration=ALL (Without candidate lists)2024.0472.475.988.883.966.585.295.338.743.572.2
GENRE_ALLTrain data=Z2025.0572.475.988.883.966.585.295.338.743.572.2
BLINK-ZTrain data=Z2025.0565.572.183.179.158.173.196.341.842.668
BLINK biEvaluation protocol=Classical (with training data)2026.0165.572.183.179.158.173.196.341.842.668
MFS baselineEvaluation protocol=Baseline2026.0163.572.383.48161.265.199.414.941.364.7