for any input bot gives output as ('bot', 'No possible replies could be determined.') about chatterbot HOT 14 CLOSED

This will happen if the program hasn't been trained with any pre-existing engrams. You will need to either create these your self or you can download the sample copy. The engram files just need to be placed in the specified log directory.

As the chatbot has conversations with people, the number of possible responses it can give will increase.

from chatterbot.

ok great Thanks , but how can i create more such samples . For example for terminal chatting bot doesnt know what to answer for What is your name ? How can i add this to its library by just chatting . is it possible ?

Type something to begin...
HI
('bot', 'How are you doing?')
What is your Name ?
('bot', 'No possible replies could be determined.')

from chatterbot.

For specific phrases, it may be easiest to manually create a conversation log which contains the desired responses.

This is a known limitation of the engram selection algorithm which only returns a known response to the phrase which is the closest match to the user's input.

I currently have a ticket open (#2) to address this by allowing the chat bot to track the current conversation by building a profile about the conversation which it can reference in order to maintain accurate responses. I have been planning to implementing NLTK inorder to extract names and sentence topics for the this purpose.

from chatterbot.

ok . I have been reading NLTK! a lot recently and i was thinking to have a bot python program somewhat like you have and on searching i found this project . 👍

Just to add some thoughts here , we should first select what type of bot it is . For example if we select a technical bot , the bot should import all the technical conversations , some historic and some standard books

If bot is for English grammar learning it should import only conversions which have standard English learning trainings .

Just some thoughts .

from chatterbot.

I would definitely be interested in implementing something to allowing the program to extract responses from books. Project Gutenberg is a perfect source to use for this. It might also be interesting to extract information from Wikipedia database dumps https://dumps.wikimedia.org.

from chatterbot.

That will be great . Can i help in some way ?

from chatterbot.

Any help or contributions would be greatly appreciated. I will write up some ideas on implementing this and post them.

from chatterbot.

ok great

from chatterbot.

Alright, I have set up a document which briefly outlines my thoughts on implementing this feature. Feel free to modify or comment on it as needed. https://docs.google.com/document/d/1AWPgrpEyzs2T1gm3Vlp-oBaoMpgO860GfDFWcqCq5To/edit#

from chatterbot.

I have done some edit suggestions in the document . Please have a look . This is first time i am working on a open source project , Please let me know if i do something wrongly .

from chatterbot.

Awesome, sorry for the late reply. I've read over the additions and left a few comments in the document. Let me know what you think. I'm hoping to finish up the changes for the database pull request sometime in the next few days. Once its finished I think we should be good to finalize the knowledge-base training plan. I've also added you as a project contributor, you should now have push access to this repository which will be useful if you're interested in making additions to the code base.

from chatterbot.

Sorry, didn't mean to close this.

from chatterbot.

This has been fixed by the changes made in #27

from chatterbot.

This thread has been automatically locked since there has not been any recent activity after it was closed. Please open a new issue for related bugs.

from chatterbot.