nlp
nlp copied to clipboard
Natural language processing course thought at AGH University of Science and Technology
Change #9-NER to #8-NER in the list of all tasks.
Clarify also whether base forms or inflected forms should be used.
Better api for polish wordnet is available: `http://ws.clarin-pl.eu/lex.shtml`. ``` url = 'http://ws.clarin-pl.eu/lexrest/lex' headers = {'Content-Type': 'application/json'} data = {'task': 'all', 'tool': 'plwordnet', 'lexeme': word} ```
`test()` method provide wrong results, but `test_label()` works just fine. Test label provide p/r/f1 for each label, just like `classification_report(test_y, predictions)` for SVM. ``` classifier = fastText.train_supervised('train-10%.txt') result = classifier.test_label('test-10%.txt')...
task 3: compute counts for not lemmatized forms
compute trigram stats in lab 5