This project aims to create a model and system for classifying and perdicting categories of persian news. The classifying part is developed in OOP manners which can be used in any project by adding the module into your code.
For this project, it is used programming languages and libraries including:
- Python3
- scikit-learn
- numpy
- pandas
- nltk
- hazm: You can see it here
- requests
- beautifulsoup4
- flask
| Category | precision | recall | f1-score | support |
|---|---|---|---|---|
| 0 | 0.99 | 0.99 | 0.99 | 299 |
| 1 | 0.93 | 0.93 | 0.93 | 295 |
| 2 | 0.93 | 0.94 | 0.93 | 291 |
| 3 | 0.99 | 0.99 | 0.99 | 307 |
| 4 | 0.99 | 0.98 | 0.98 | 308 |
accuracy: 0.9653333333333334
Dataset has been retrieved from ISNA News Agency in five category:
- cinema
- economy
- politic
- sport
- tech and science
The code exports a csv file which includes normalized texts with their labels in each row. It also saves the model as a pickle file.
Scraper's python code is added in scraper directory. If you want to add new texts, you should give a valid url like: url
You can install libararies by running the command pip install -r reqirements.txt in cmd-windows or if you are using linux, use pip3 instead of pip.
You can run the code by executing python3/python app.py.