Entity Linking on AIDA A
89.8Micro F1base model + att + global
Evaluation Results
| Method | Links | ||
|---|---|---|---|
| base model + att + globalAttention=true, Global modeling=true2018.08 | 89.8 | 87.2 | |
| base modelAttention=false, Global modeling=false2018.08 | 89.5 | 87 | |
| base model + att + globalMatching criteria=strong matching, Model architecture components=base + local attention + global2018.08 | 89.4 | 86.6 | |
| base model + attAttention=true, Global modeling=false2018.08 | 89.3 | 87.1 | |
| base modelMatching criteria=strong matching, Mention detection=end-to-end2018.08 | 89.1 | 86.6 | |
| ed base model + att + globalTraining set=AIDA-train set, Attention mechanism=true, Global context=true, Evaluation platform=Gerbil2018.08 | 89 | 86.6 | |
| base model + attMatching criteria=strong matching, Model architecture components=base + local attention2018.08 | 88.9 | 86.5 | |
| ed base model + attTraining set=AIDA-train set, Attention mechanism=true, Global context=false, Evaluation platform=Gerbil2018.08 | 88.9 | 86 | |
| ed base modelTraining set=AIDA-train set, Attention mechanism=false, Global context=false, Evaluation platform=Gerbil2018.08 | 88.8 | 86.5 | |
| Kolitsas et al. (2018)2019.09 | 86.6 | — | |
| KnowBert-W+WBackbone=BERT-Base, Knowledge Base=Wikipedia + WordNet2019.09 | 82.1 | — | |
| ED base model + att + globalAttention=true, Global modeling=true, Mention detection source=Stanford NER mentions2018.08 | 80.5 | 76 | |
| ED base model + att + global using Stanford NER mentionsMention detection source=Stanford NER, Task setting=Entity Disambiguation (ED)2018.08 | 80.3 | 75.7 | |
| KnowBert-WikiBackbone=BERT-Base, Knowledge Base=Wikipedia2019.09 | 80.2 | — | |
| PBOHEvaluation platform=Gerbil2018.08 | 80.1 | 77.3 | |
| WATEvaluation platform=Gerbil2018.08 | 78.5 | 75.6 | |
| AIDAEvaluation platform=Gerbil2018.08 | 74.7 | 71.4 | |
| WAT2018.08 | 73.3 | 69.7 | |
| Best baseline2018.08 | 73.3 | 69.7 | |
| WAT2018.08 | 72.8 | 69.2 | |
| AIDA2018.08 | 72.6 | 69 | |
| AIDA2018.08 | 72.4 | 68.8 | |
| BabelfyEvaluation platform=Gerbil2018.08 | 71.9 | 66.4 | |
| Hoffart et al. (2011)2019.09 | 68.8 | — | |
| KeaEvaluation platform=Gerbil2018.08 | 63.5 | 62.6 | |
| FOXEvaluation platform=Gerbil2018.08 | 58.9 | 55.5 | |
| DBpedia Spotlight2018.08 | 58.3 | 55.5 | |
| FOX2018.08 | 58.1 | 54.9 | |
| FOX2018.08 | 58 | 54.7 | |
| AGDISTISEvaluation platform=Gerbil2018.08 | 55.6 | 49.2 | |
| DBpedia Spotlight2018.08 | 55.2 | 49.9 | |
| Entityclassifier.euEvaluation platform=Gerbil2018.08 | 53.7 | 53.6 | |
| DBpedia SpotlightEvaluation platform=Gerbil2018.08 | 53.2 | 51.5 | |
| Daiber et al. (2013)2019.09 | 49.9 | — | |
| Babelfy2018.08 | 48.1 | 42.1 | |
| Babelfy2018.08 | 47.2 | 41.2 | |
| Entityclassifier.eu2018.08 | 47 | 44.9 | |
| Entityclassifier.eu2018.08 | 44.7 | 43 | |
| Kea2018.08 | 42.9 | 38.7 | |
| Kea2018.08 | 40.4 | 36.8 | |
| FREME2018.08 | 38.3 | 23.9 | |
| FREME2018.08 | 37.6 | 23.6 | |
| FREMEEvaluation platform=Gerbil2018.08 | 33 | 25.3 |