predicts the author of chat or chat-line, a model trainable from whatsapp exportable chat texts
-
Open project directory.
-
Export whatsapp chats from your phone. Go through link for help.
-
Use personal chats and not groupchats preferably. (optional)
-
Create a directory named dataset and put the exported chats inside that 'dataset' directory.
-
Open 'myname.json' and modify "name" value and by replacing "Sayan Dey" with
-
Automated: Change mode of the file train to executable, run it and wait.
In linux terminal:
chmod +x train
./train
Manual: Run the three python script in order: WhatsApp.py, process_dataframe.py, author_classify.py
Alternative: Run the .ipynb files in jupyter or with IPython in order: WhatsApp.ipynb, process_dataframe.ipynb, author_classify.ipynb
- Open 'test.ipynb' with jupyter or ipython. The sentence variable is a list of strings.
Modify the strings to the text you think someone often says or texts and run it. It might match with your person in imagination.