HMM based tagger

This is a tagger for Natural Languages based on Hidden Markov Model.

Usage

All programs in the package are compiled by issuing the make command in the main directory. Note that the main script does calls make by itself.

The main script is tagger. It should be used as follows:

tagger <body1> [<body2>]

where parameters denote catalogs with language corpuses in the IPI PAN format. It has two modes:

It can be run in two modes:

one-parameter - then it performs a 10-fold cross-validation in order to test the HMM tagger and compare it with a simple unigram tagger
two-parameter - then it trains the model on the first given body and tags the second body using the created model.

After launching, the program prints all information to the standard output including a progress bar.

Name		Name	Last commit message	Last commit date
Latest commit History 1 Commit
README.md		README.md
divide.cpp		divide.cpp
extract_dict		extract_dict
extract_ngrams.cpp		extract_ngrams.cpp
extract_tagset		extract_tagset
find_gammas.cpp		find_gammas.cpp
generate_probs.cpp		generate_probs.cpp
generic_tagger.h		generic_tagger.h
get_prob.cpp		get_prob.cpp
get_prob.h		get_prob.h
gt_smoothen.cpp		gt_smoothen.cpp
hmm_tag.cpp		hmm_tag.cpp
interpret_scores.cpp		interpret_scores.cpp
make_backup		make_backup
makefile		makefile
model_dumper.cpp		model_dumper.cpp
model_dumper.h		model_dumper.h
ngram_tools.cpp		ngram_tools.cpp
ngram_tools.h		ngram_tools.h
score.cpp		score.cpp
tagger		tagger
tagger_wrapper.cpp		tagger_wrapper.cpp
tokenizer.cpp		tokenizer.cpp
tokenizer.h		tokenizer.h
unigram_tag.cpp		unigram_tag.cpp
untag.cpp		untag.cpp