Training:
-
Extract dataset directory
(dataset/bbc-fulltext.zip). The directory should be namedbbc. -
Run
compile.pyin the/datasetdirectory from your terminal to compile the text files into single dataset file.$ cd dataset$ python compile.py -
Run
engine.py.$ python engine.py -
If there is no pickle files, the system will automatically train the dataset and generate pickle files. This could take some time depends on your hardware.
-
If you want to retrain your system simply delete the
picklesdirectory.
Classification:
-
Run the
app.pyfile.$ python app.py -
Using Postman (or similar tools), send
POSTrequest tohttp://127.0.0.1:5050/classifywith these form data:post : your_article