Part-of-speech-tagging-with-discriminatively-re-ranked-Hidden-Markov-Models
Implemented POS tagging by combining a standard HMM tagger separately with a Maximum Entropy classifier designed to re-rank the k-best tag sequences produced by HMM – achieved better results than VITERBI (decoding algorithm)
Disclaimer:
This programs takes approximately 4 to 5 hours to run:
The soft copy of the report is attached incase you want to see the results
Two programs have to be executed:
The original algorithm: Viterbi
The algorithm implemented in Project: Reranked HMM
To execute:
python hmm_beam_search.py
Dependencies:
python3
nltk
megam
numpy