Predicting Typological Features in WALS using Language Embeddings and Conditional Probabilities: \'UFAL Submission to the SIGTYP 2020 Shared Task
In: Proc. SIGTYP Workshop on Computational Research in Linguistic Typology (2020) 29-35; (2020)
Online
report
Zugriff:
We present our submission to the SIGTYP 2020 Shared Task on the prediction of typological features. We submit a constrained system, predicting typological features only based on the WALS database. We investigate two approaches. The simpler of the two is a system based on estimating correlation of feature values within languages by computing conditional probabilities and mutual information. The second approach is to train a neural predictor operating on precomputed language embeddings based on WALS features. Our submitted system combines the two approaches based on their self-estimated confidence scores. We reach the accuracy of 70.7% on the test data and rank first in the shared task.
Titel: |
Predicting Typological Features in WALS using Language Embeddings and Conditional Probabilities: \'UFAL Submission to the SIGTYP 2020 Shared Task
|
---|---|
Autor/in / Beteiligte Person: | Vastl, Martin ; Zeman, Daniel ; Rosa, Rudolf |
Link: | |
Quelle: | Proc. SIGTYP Workshop on Computational Research in Linguistic Typology (2020) 29-35; (2020) |
Veröffentlichung: | 2020 |
Medientyp: | report |
DOI: | 10.18653/v1/2020.sigtyp-1.4 |
Schlagwort: |
|
Sonstiges: |
|