Module maleo.pos_tag

Classes

class POS

A Part-of-Speech Tagging classifier.

POST, also called grammatical tagging is the process of marking up a word in a text as corresponding to a particular part of speech.

References

https://universaldependencies.org/u/pos/index.html

Expand source code Browse git

class POS:
    """A Part-of-Speech Tagging classifier.
    
    POST, also called grammatical tagging is the process of marking up a word in a text as 
    corresponding to a particular part of speech.
    
    References
    ----------
    https://universaldependencies.org/u/pos/index.html
    """
    def __init__(self):
        self.model = self.load_model()


    def check_model(self, model_filename):
        """Check existence of model, if not exist then will download the model."""
        if not path.exists(model_filename):
            url = 'https://drive.google.com/uc?id=1-C8RRM9c-IaGgN2jdzFtlwY7BuOAMY3W'
            print('Downloading model ...')
            gdown.download(url, model_filename, quiet=False)
            print('DONE ...')


    def load_model(self):
        """Load POST model"""
        model_path = pkg_resources.resource_filename('maleo','pos_tag/pos_model.pt')
        self.check_model(model_path)
        model = SequenceTagger.load(model_path)
        return model


    def predict(self, text:str, output_pair=False):
        """Inference POST model.

        Parameters
        ----------
        text: str
            Input text
        output_pair: boolean
            True -> list of tuples 
            False -> tuple
        Returns:
        -------
        out : tuple or list of tuples
            Inference result with format based on output_pair
        """
        input_text = Sentence(text)
        self.model.predict(input_text)
        output_text = input_text.to_tagged_string()

        item = output_text.split()
        out = [(item[idx], item[idx+1]) for idx in range(0, len(item), 2)]

        if not output_pair:
            sent, pos = zip(*out)
            out = (' '.join(sent), ' '.join(pos))
            return out
        else:
            return out

Methods

def check_model(self, model_filename)

Check existence of model, if not exist then will download the model.

Expand source code Browse git

def check_model(self, model_filename):
    """Check existence of model, if not exist then will download the model."""
    if not path.exists(model_filename):
        url = 'https://drive.google.com/uc?id=1-C8RRM9c-IaGgN2jdzFtlwY7BuOAMY3W'
        print('Downloading model ...')
        gdown.download(url, model_filename, quiet=False)
        print('DONE ...')

def load_model(self)

Load POST model

Expand source code Browse git

def load_model(self):
    """Load POST model"""
    model_path = pkg_resources.resource_filename('maleo','pos_tag/pos_model.pt')
    self.check_model(model_path)
    model = SequenceTagger.load(model_path)
    return model

def predict(self, text: str, output_pair=False)

Inference POST model.

Parameters

text : str

Input text

output_pair : boolean

True -> list of tuples False -> tuple

Returns:

out : tuple or list of tuples Inference result with format based on output_pair

Expand source code Browse git

def predict(self, text:str, output_pair=False):
    """Inference POST model.

    Parameters
    ----------
    text: str
        Input text
    output_pair: boolean
        True -> list of tuples 
        False -> tuple
    Returns:
    -------
    out : tuple or list of tuples
        Inference result with format based on output_pair
    """
    input_text = Sentence(text)
    self.model.predict(input_text)
    output_text = input_text.to_tagged_string()

    item = output_text.split()
    out = [(item[idx], item[idx+1]) for idx in range(0, len(item), 2)]

    if not output_pair:
        sent, pos = zip(*out)
        out = (' '.join(sent), ' '.join(pos))
        return out
    else:
        return out