FastHMM 0.1.2

A python package for HMM model with fast train and decoding implementation

中文版本的 README

FastHMM

A python package for HMM (Hidden Markov Model) model with fast train and decoding implementation

Python version

test by using Python3

Install

pip

pip install FastHMM

source

pip install git+https://github.com/312shan/FastHMM.git

Usage

from FastHMM.hmm import HMMModel

# test model training and predict
hmm_model = HMMModel()
hmm_model.train_one_line([("我", "r"), ("爱", "v"), ("北京", "ns"), ("天安门", "ns")])
hmm_model.train_one_line([("你", "r"), ("去", "v"), ("深圳", "ns")])
result = hmm_model.predict(["俺", "爱", "广州"])
print(result)

# test save and load model
hmm_model.save_model()
hmm_model = HMMModel().load_model()
result = hmm_model.predict(["我们", "爱", "深圳"])
print(result)

Output:

[('俺', 'r'), ('爱', 'v'), ('广州', 'ns')]
[('我们', 'r'), ('爱', 'v'), ('深圳', 'ns')]

Performance:

test on dataset 人民日报

python .test/test_postagging.py

Output:

train size 18484 ,test_size 1000
finish training
eval result: 
predict 57929 tags, 54228 correct,  accuracy 0.9361114467710473
runtime : 370.1029086 seconds

Most of time the consuming is on the decoding stage, I tried many ways to implement viterbi algorithm, The implementation I currently use is the fastest If you have suggestions for improving this decoding algorithm, please let me know, thank you very much.

Reference

MicroHMM
Hidden Markov model
Viterbi algorithm