0
Answered

how to create chatbot with free flow conversation and not just following one set of rigid flow

adez3d@gmail com 7 years ago updated by Praveen Srinivasaiah (Admin) 7 years ago 5

Hi, I'm newbie here. 

how to create a personal chatbot that can answer a free flow NLP question?

Something similar to Mitsuku or Zo from Microsoft. 


on "on sequel"  it seems we have  to follow one set of flow from start to end in almost linear manner (with branches)


Is is possible to create a bot that can answer random user question freely (like a Q&A) most of the time? 


something like this:

User: How are you?

bot: fine, thanks, how about you?

etc


but some other time can also create a flow of conversation (for instance when talk about something specific like a movie).


Thanks




Answered

Yes, this can be created using the NLP nodes. Here is the link to our guide which will help you further.


http://forum.onsequel.com/forums/2-faqs/categories/16-nlp-natural-language-processing/topics/


Let me know if you have any questions.


Hi Praveen, thanks for the answer. 

I believe I'm quite familiar with all the NLP stuffs.


However to create something like Zo from microsoft > https://www.messenger.com/t/zo

We will need to have "thousands and thousands" possible NLP questions and answers. 


If we use NLP PLAYER DECISION NODE then it will create a very long flat horizontal node (thousand of them)


And if we use NLP RANDOM RESPONSE NODE it will create a very long vertical node. also with thousand questions and answers.

Is it possible to create an NLP node, independent of any flow/sequel?


So if I need to follow certain flow of conversation it will follow a particular flow,

but when it needs to just answer any random NLP question from user, it will only answer just that question without event bother to follow  any particular flow/sequel.





For this specific use case, NLP random message node fits in best where you can have multiple possible questions/words user might input and corresponding responses to each of them. 

Thanks Praveen for the answer