AI Zone Admin Forum Add your forum

Chatbot Conference in the Netherlands on Tue 24, 2018, with international line-up! Check it out

How to solve accuracy problem?
 
 

In the Alice aiml converted files to CS there are many sentences like these:

#! what
#u: (what) bandaide.
#! what
#u: (what) cartune.
#! what
#u: (what) flower power.
#! what
#u: (what) pork chops.
#! what
#u: (what) fast food.
#! what
#u: (what) cereal killer.
#! what
#u: (what) canned carrot.
#! what
#u: (what) serial killer.
#! what
#u: (what) I dont know, but its serial.
#! what
#u: (what) I dont know, but its ruff.
#! what
#u: (what) sour milk.
#! what
#u: (what) carnation. 

And when I add some other sentences like this:

u: (what is your namemy name is Alice 

Most of the time the CS doesn’t choose the last one, even when the input is exactly `what is your name`.
Are there any solution for this?

 

 
  [ # 1 ]

I think you have to put it before others for accuracy

 

 
  [ # 2 ]

Thanks but where?
Are there any priority between the .top files?

 

 
  [ # 3 ]

I quote an answer from andy heydon when i had the same problem :
“”“
You want the most specific pattern first.
CS (generally*) starts at the top and works down the topic until it finds a rule whose pattern matches. It will stop when one generates output. So put the harder to match patterns first and the least specific, catch-all patterns at the end.
“”“

Hope it will help you…

 

 
  login or register to react