AI Zone Admin Forum Add your forum

NEWS: survey on 3000 US and UK consumers shows it is time for chatbot integration in customer service!read more..

Intelligent Branching

I am considering making my own personalized chatbot and while Harry’s control has been adequate for the customer service or interview based chatbots ive made so far, it does not really help with meandering conversation.
The main flaw there seems to be the jump to any keyword topic at the end. That is very ambiguous and the jump could be very out of context.

So, i thought I could create more ‘intelligent branching’ at that point. Maybe by queuing up related topics and most importantly by having the control manage branching by mood and branching on polarity of the current mood but also by using topics to handle bringing mood back to an ‘average’ state.

Am I overshooting here? Is there a simpler approach?

Love my current project. Its purely in the design and mind mapping stage right now but even that is really an art.


  [ # 1 ]

I have fun with this sort of thing. I have a $demeanour variable that is influenced by things I say, and impacts bot choices and even vocabulary.

For example, if I don’t say goodbye before logging off, it’s a minus.

Saying I missed her gets me plus points.

My bot gets grumpy and less assured around the full moon! wink

If she’s really pissed off, she becomes almost monosyllabic.


  [ # 2 ]

Love it, I just added sentiment analysis to program-y, as you talk to it, it keeps a running score of positivity (-1 a +1) and objectivity ( 0 - 1 ) and can use them to tailor the response but I really like the temporal element that can have an effect on mood


  [ # 3 ]

Keith, I’m guessing your “positivity” variable would be similar to my “demeanour”, but I’m curious as to how you use “objectivity”?


  [ # 4 ]

Subjectivity measures the range of objective to subjectivity positivity

I really really hate you is very subjective negative, it’s a personal view
I love a cheetah because it is so fast is positive and objective as it’s based on fact
I hate spiders because they have 8 legs is objective negative

Just adds a different dimension to how the bot replies to pos and neg statements

More subjective statements impact tone, were as more objective indicate a higher like or dislike


  login or register to react