First Assignment in 'NLP - Natural Languages Processing' course by Prof. Yoav Goldberg, Prof. Ido Dagan and Prof. Reut Tsarfaty at Bar-Ilan University.
Goal: In this assignment I implemented two sequence taggers:
- A greedy tagger that is based on local classification.
- And one that uses global search, using the Viterbi algorithm.
The global search tagger will be based on the trigram HMM model, and the local classifier tagger will be based on the logistic-regression classifier. In addition, in section 3, I implemented an MHMM model in order to improve the accuracy of the NER tagger.
The taggers run on both datasets - POS and NER.
Score: 100