Skip to content

PoTeC DE

Domain Expertise (PoTeC)

Test

ModelUnseen Reader Balanced AccuracyUnseen Text Balanced AccuracyUnseen Text and Reader Balanced AccuracyAverage Balanced AccuracyUnseen Reader AUROCUnseen Text AUROCUnseen Text and Reader AUROCAverage AUROC
Majority Class / Chance52.5 ± 2.349.9 ± 0.649.9 ± 1.351.4 ± 1.352.5 ± 2.349.9 ± 0.649.9 ± 1.351.4 ± 1.3
Reading Speed59.2 ± 2.259.1 ± 4.057.7 ± 4.759.0 ± 1.060.2 ± 1.461.0 ± 5.556.8 ± 6.860.4 ± 1.7
Text-Only Roberta50.0 ± 0.050.0 ± 0.050.0 ± 0.050.0 ± 0.065.7 ± 4.457.6 ± 5.655.2 ± 2.762.0 ± 4.0
Logistic Regression [meziere2023using]55.3 ± 1.650.6 ± 2.842.5 ± 5.051.6 ± 1.558.8 ± 2.053.1 ± 2.841.6 ± 7.854.0 ± 1.7
SVM [hollenstein2023zuco]53.5 ± 2.257.0 ± 1.649.3 ± 4.454.5 ± 1.553.5 ± 2.257.0 ± 1.649.3 ± 4.454.5 ± 1.5
Random Forest [makowski2024detection]56.9 ± 3.650.7 ± 0.651.7 ± 1.753.6 ± 1.869.2 ± 3.652.7 ± 7.460.2 ± 4.162.3 ± 3.4
AhnRNN [ahn2020towards]50.0 ± 0.050.0 ± 0.050.0 ± 0.050.0 ± 0.050.0 ± 0.149.9 ± 0.150.0 ± 0.050.0 ± 0.1
AhnCNN [ahn2020towards]50.6 ± 0.549.9 ± 0.149.8 ± 0.250.2 ± 0.260.7 ± 2.459.8 ± 7.360.8 ± 6.660.6 ± 3.4
BEyeLSTM [reich_inferring_2022]64.1 ± 4.147.1 ± 3.350.5 ± 5.253.0 ± 2.765.7 ± 3.847.2 ± 6.846.8 ± 12.151.8 ± 3.5
PLM-AS [Yang2023PLMASPL]53.0 ± 2.047.7 ± 1.150.2 ± 1.150.0 ± 0.652.6 ± 2.852.6 ± 2.549.0 ± 9.551.3 ± 2.4
PLM-AS-RM [haller2022eye]55.5 ± 3.949.9 ± 0.150.0 ± 0.052.4 ± 1.764.7 ± 5.865.4 ± 3.660.7 ± 4.164.2 ± 4.0
RoBERTEye-W [Shubi2024Finegrained]50.0 ± 0.050.0 ± 0.050.0 ± 0.050.0 ± 0.065.3 ± 5.362.6 ± 9.861.3 ± 9.462.5 ± 7.3
RoBERTEye-F [Shubi2024Finegrained]50.0 ± 0.050.0 ± 0.050.0 ± 0.050.0 ± 0.071.3 ± 2.152.0 ± 6.566.7 ± 1.864.5 ± 3.1
MAG-Eye [Shubi2024Finegrained]50.8 ± 0.750.0 ± 0.050.0 ± 0.050.4 ± 0.465.2 ± 7.647.4 ± 9.348.9 ± 13.457.6 ± 7.1
PostFusion-Eye [Shubi2024Finegrained]49.9 ± 0.150.0 ± 0.050.0 ± 0.050.0 ± 0.055.0 ± 4.050.7 ± 4.252.3 ± 4.753.6 ± 0.9

Validation

ModelUnseen Reader Balanced AccuracyUnseen Text Balanced AccuracyUnseen Text and Reader Balanced AccuracyAverage Balanced AccuracyUnseen Reader AUROCUnseen Text AUROCUnseen Text and Reader AUROCAverage AUROC
Majority Class / Chance52.9 ± 1.649.8 ± 0.649.8 ± 1.051.0 ± 0.552.9 ± 1.649.8 ± 0.649.8 ± 1.051.0 ± 0.5
Reading Speed60.2 ± 4.756.8 ± 6.557.1 ± 3.958.9 ± 3.762.3 ± 3.356.6 ± 7.656.8 ± 6.860.1 ± 3.0
Text-Only Roberta50.0 ± 0.050.0 ± 0.050.0 ± 0.050.0 ± 0.069.8 ± 3.063.3 ± 9.362.8 ± 5.366.7 ± 3.8
Logistic Regression [meziere2023using]55.9 ± 3.250.2 ± 2.747.0 ± 7.053.2 ± 1.454.6 ± 4.051.1 ± 2.946.0 ± 7.053.3 ± 1.5
SVM [hollenstein2023zuco]56.4 ± 1.559.1 ± 2.559.0 ± 2.357.8 ± 1.856.4 ± 1.559.1 ± 2.559.0 ± 2.357.8 ± 1.8
Random Forest [makowski2024detection]62.8 ± 6.053.3 ± 4.049.2 ± 2.058.4 ± 3.263.6 ± 5.159.4 ± 7.349.2 ± 5.560.9 ± 3.6
AhnRNN [ahn2020towards]50.0 ± 0.050.0 ± 0.050.0 ± 0.050.0 ± 0.050.0 ± 0.250.1 ± 0.150.0 ± 0.050.0 ± 0.1
AhnCNN [ahn2020towards]49.9 ± 0.150.0 ± 0.050.0 ± 0.050.0 ± 0.059.9 ± 5.560.0 ± 9.459.0 ± 5.361.1 ± 2.6
BEyeLSTM [reich_inferring_2022]67.5 ± 1.960.0 ± 5.059.2 ± 4.963.6 ± 1.871.1 ± 1.971.5 ± 5.667.3 ± 3.371.7 ± 2.3
PLM-AS [Yang2023PLMASPL]56.2 ± 3.452.5 ± 1.847.1 ± 0.754.4 ± 2.561.4 ± 6.555.5 ± 2.534.4 ± 6.356.2 ± 4.7
PLM-AS-RM [haller2022eye]61.4 ± 5.950.0 ± 0.050.0 ± 0.057.0 ± 3.870.8 ± 3.848.5 ± 11.242.0 ± 10.857.9 ± 7.7
RoBERTEye-W [Shubi2024Finegrained]50.0 ± 0.050.0 ± 0.050.0 ± 0.050.0 ± 0.070.6 ± 6.367.0 ± 9.473.7 ± 7.369.9 ± 4.3
RoBERTEye-F [Shubi2024Finegrained]50.0 ± 0.050.0 ± 0.050.0 ± 0.050.0 ± 0.073.3 ± 4.870.9 ± 5.957.8 ± 7.571.4 ± 1.6
MAG-Eye [Shubi2024Finegrained]54.7 ± 4.150.0 ± 0.050.0 ± 0.052.5 ± 2.267.8 ± 3.574.4 ± 5.570.9 ± 8.271.6 ± 3.5
PostFusion-Eye [Shubi2024Finegrained]50.6 ± 0.650.0 ± 0.050.0 ± 0.050.4 ± 0.459.5 ± 2.164.5 ± 6.754.8 ± 7.261.1 ± 4.2