PolEval 2018 NER task
Determine nested Named Entities in NKJP-compatible way, that is provide a series of labels with corresponding token indexes. [ver. 1.0.0]
This is a long list of all submissions, if you want to see only the best, click leaderboard.
| # | submitter | when | ver. | description | dev-0 MultiLabel-F1.0 | test-A MultiLabel-F1.0 | test-B MultiLabel-F1.0 | |
|---|---|---|---|---|---|---|---|---|
| 2 | [anonymized] | 2018-09-18 18:47 | 1.0.0 | LM-LSTM-CRF, as used by Borchmann, Gretkowski and Graliński. python neural-network lm lstm | 0.81687 | 0.86979 | 0.81112 | |
| 3 | [anonymized] | 2018-09-18 14:02 | 1.0.0 | Baseline SEARN-based solution, as used by Borchmann, Gretkowski and Graliński. vowpal-wabbit feature-engineering n-grams lemmatization knowledge-based char-n-grams | 0.78564 | 0.82062 | 0.79223 | |
| 1 | [anonymized] | 2018-09-17 17:53 | 1.0.0 | Parallel LSTM-CRFs, as used by Borchmann, Gretkowski and Graliński (GloVe and Contextual String Embeddigns, Flair framework). python neural-network lm lstm | 0.89654 | 0.88284 | 0.84508 |