IndoLEM Resource Collection

Leaderboard


We open any submission for each category listed below. Please follow this instruction 😊


POS and NER
Method POS (Acc) NER-UGM (F1) NER-UI (F1)
IndoBERT (Koto et al., 2020) 96.8 74.9 90.1
MalayBERT 96.8 73.2 87.4
mBERT 96.8 71.6 82.2
BiLSTM (Lample et al., 2016) 95.4 70.9 82.2

Dependency Parsing
Method Indo-GSD Indo-PUD
UAS LAS UAS LAS
IndoBERT (Koto et al., 2020) 87.1 82.3 89.2 83.9
MalayBERT 86.9 81.8 88.9 83.6
mBERT 86.9 81.8 90.6 85.4
Kondratyuk and Straka (2019) 86.5 80.1 77.5 56.9

Semantic
Method Sentiment IndoSum Liputan6 (Sum)
F1 R1 R2 RL R1 R2 RL
IndoBERT (Koto et al., 2020) 84.1 69.9 62.9 69.2 41.1 22.9 38.9
MalayBERT 82.0 68.4 61.4 67.7 -- -- --
mBERT 76.6 68.4 61.7 47.8 39.8 21.8 37.0
BiLSTM 71.6 67.9 61.7 67.2 36.1 19.2 33.6

Discourse Coherency
Method NTP (Acc) Tweet Ordering (ρ)
IndoBERT (Koto et al., 2020) 93.7 0.59
MalayBERT 93.1 0.51
mBERT 92.4 0.53
BiLSTM 73.6 0.45
Content