TagLID

A word-level Language Identification (LID) tool for Tagalog-English (Taglish) text

About

TagLID is a library that labels each word in a Taglish (Tagalog-English mix) text by language. It gives either a simple tag (tgl or eng) or detailed frequency info with flags indicating how the word was identified. It is a rule-based and opinionated system that mostly uses dictionary lookups. It also handles cases like skipping numbers, names, and interjections, and includes logic for dealing with slang, abbreviations, contractions, stemming or lemmatizing inflected words, intrawords, and correcting misspellings.

Installation

pip install git+https://github.com/andrianllmm/taglid.git@main

Usage

TagLID can act as a standalone library that can be imported via import taglid or as a CLI application via python -m taglid.

Library Mode

Textual data

Use the lid module for textual data.

Use lang_identify to identify each word in a text. This takes any string and returns a list of words and their corresponding English and Tagalog values, flag, and correction.

from taglid.lid import lang_identify

labeled_text = lang_identify("hello, mundo")
print(labeled_text)

Output:

[{'Word': 'hello', 'eng': 1.0, 'tgl': 0.0, 'Flag': 'DICT', 'Correction': None}, {'Word': 'mundo', 'eng': 0.0, 'tgl': 1.0, 'Flag': 'DICT', 'Correction': None}]

Use tabulate to view output in tabular format.

from tabulate import tabulate

print(tabulate(labeled_text, headers="keys"))

Output:

word      eng    tgl  flag    correction
------  -----  -----  ------  ------------
hello       1      0  DICT
mundo       0      1  DICT

Use simplify to only show the words and their language. This takes the return value of lang_identify and returns a list of tuples containing the word and its language.

from taglid.lid import simplify

simplified_text = simplify(labeled_text)
print(simplified_text)

Output:

[('hello', 'eng'), ('mundo', 'tgl')]

Datasets

Use the lid_dataset module for datasets.

Use lang_identify_df to label each word in each cell in a pandas DataFrame. This takes a DataFrame of multiple rows and columns with each cell containing textual data and returns a labeled DataFrame where each token is a row labeled by its original row, original column, and token index.

import pandas as pd
from taglid.lid_dataset import lang_identify_df

data = [['hello po', 'ano?'], ['mag-aask lang po', 'what?']]

df = pd.DataFrame(data)

labeled_df = lang_identify_df(df)
print(labeled_df)

Output:

     col  token_index      word  eng  tgl  flag correction
row
0      0            1     hello  1.0  0.0  DICT       None
0      0            2        po  0.0  1.0  DICT       None
0      1            1       ano  0.0  1.0  FREQ       None
1      0            1  mag-aask  0.5  0.5  INTW       None
1      0            2      lang  0.0  1.0  FREQ       None
1      0            3        po  0.0  1.0  DICT       None
1      1            1      what  1.0  0.0  DICT       None

CLI Mode

Run TagLID from the terminal.

python -m taglid.lid

Then type a sentence when prompted.

text: hello, mundo

Output:

word      eng    tgl  flag    correction
------  -----  -----  ------  ------------
hello       1      0  DICT
mundo       0      1  DICT

Add --simplify to only show the words and their language.

python -m taglid.lid --simplify --text hello, mundo

Output:

-----  ---
hello  eng
mundo  tgl
-----  ---

Use lid_dataset with Excel files to directly label spreadsheets.

python -m taglid.lid_dataset in_path out_path

Accuracy

The accuracy hasn't been tested yet.

Contributing

Contributions are welcome! To get started:

1. Fork the project
2. Create your feature branch (git checkout -b feature/AmazingFeature)
3. Commit your changes (git commit -m 'Add some AmazingFeature')
4. Push to the branch (git push origin feature/AmazingFeature)
5. Open a pull request

Issues

Found a bug or issue? Report it on the issues page.