N-gram

Simple N-gram model that utilizes dictionaries to achieve its purpose. This N-gram model provides a modular approach for gernerating and reading N-grams. It basically uses nested dictionaries in this way: (for bigrams)

'word1': {
.  'word2': 0.5
.  'Wordn': 0.5
}
'word2': {
.   'word3': 1
} ...

usage for the python library

Download the ngram.py file and use the library by typing

import ngram

into your working space.

Generating N-gram

generate_ngram(text, ngramLength, file_write, punctuations)

• text (str) -- The text you want to generate your ngram from.
• ngramLength (int) -- the length of the ngram to generate (2 generates two pairs and so on.) (Default 2)
• file_write (bool) -- Whether you want to write a text file containing another representation of the ngram. (Default True)
• punctuations (str) -- Symbols to remove. (Default !"#$%&'()*+,-./:;<=>?@[]^_`{|}~)

word1 -> word2 : probability

Writing to file

Replacing the entire file

write_to_json(model, file)

• model(dict) - your generated ngram.
• file (str["N-gram.json"]) - Json file to write. (Default N-gram.json)

Adding to the file

add_to_json(model, file)

• model(dict) - your generated ngram.
• file (str["N-gram.json"]) - Json file to write. (Default N-gram.json)

Reading from file

read_from_json(file)

• file (str["N-gram.json"]) - Json file to write. (Default N-gram.json)

Using the ngram for prediction

Single word/ list of words

predict_next_word(model, text, choose)

• model(dict) - your ngram.
• text(str) - text to predict from.
• choose(bool) - whether to return one word or a list of words. (Default True)

More words

predict_more(model, text, word_count)

An extension to predict_next_word.

• model(dict) - your ngram.
• text(str) - text to predict from.
• word_count(int) - the amount of words you want to predict.

Usage example

I made a website using this library which you can checkout here https://ngram.vercel.app.