Is there a way to continue training?
I saw that the part where the model is saved was commented in your code, so I uncommented it, saved a model a loaded it with the load_model( ) method, but when I call the method train( ), it starts to train from the beginning.
Am I missing something? Or is the feature not implemented?
Thanks a lot!

Hey Ali, congratulations for your work!

I'm making some experiences with your solution with the data I'm working on, with the requirements described at README.md, but I think that I'm not getting the desired results...

I runned the main.py (with my own files) and I'm getting this error while fitting TfidfVectorizer on get_topics_of_agents method from Cordinator.py line 390:

empty vocabulary; perhaps the documents only contain stop words

If I comment the section where get_topics_of_agents is called to save topics, the logs on console show that the number of agents becomes zero, and the saved clusters (which are only one, or two in some runs) only appear to have one text:

Init_agents done : Number of agents : 2
2020-01-02 20:58:00 : Communicating -> Number of agents : 1
2020-01-02 20:58:00 : Save Model and Outputs -> Number of agents : 1
2020-01-02 20:59:00 : Communicating -> Number of agents : 0

Any idea what might be causing this behaviour?

Thank you so much in advance :)

Does ComStream become gradually slow as data stream comes?