Natural Language Processing

1 Lab

HMM realisation PoS-tagger and comparing accuracy of results with CRF algorithm.

Corpus	HMM	CRF
Brown	91.68487501710086	95.71677057478453
af-ud (Afrikaans)	81.99701937406856	92.86636860407353
de-ud (German)	79.13769123783032	90.24611477293341
zh-ud (Chinese)	78.31335331335332	87.4042624042624

2 Lab

Running and testing SyntaxNet (tensorflow model). Analysing Romanian and Latvian corpuses.