AI Zone Admin Forum Add your forum

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

Slight Pattern Mismatching in ChatScript.
 
 

I have hosted a bot with the topic file found in the following link.

https://drive.google.com/file/d/0Bxx_C0y8V8hscllNRXZUbnhTdUU/view?usp=sharing

The idea behind the chat bot is to make it deliver details about a book or an author when user asks.


Chatbot is responding correctly for some input pattern but for some patterns it matches irrelevant outputs and responds with the same.

“http://ec2-52-89-79-125.us-west-2.compute.amazonaws.com/index.php” is where you can find th e chat-bot and compare output with the topic file.

For instance,
look at the snaps..“malyn monroe returns” isn’t matching its output.
but the “Dr.Brian weiss” matches correctly..

I couldn’t sort out why..

Image Attachments
3.PNG5.PNG4.PNG
 

 
  [ # 1 ]

Maybe you have to backslash the : in Marilyn…..Returns?
No sorry. Just checked and raw :  is accepted in the pattern….But why have everything in brackets? Thats probably what leads to your mismatch.

I tried Marilyn Monroe Returns
Marilyn Monroe Returns:

and the match was ok.
but Marilyn Monroe Returns: The and
Marilyn Monroe Returns: The Healing had different matches.
You may have to shift the order of patterns if you are going to use square brackets.

 

 
  [ # 2 ]

Instead of going to use brackets to only type part of the book name try using patterns of partial names with with reuse and DONT use square brackets there.
For partial names that match more than one book you can use rejoinders to specify.

 

 
  [ # 3 ]

proper names with embedded punctuation should be inside quotes. (since you don;t know how CS will tokenize it)

And generally so should all multiple word proper names.

 

 
  [ # 4 ]
Bruce Wilcox - Dec 26, 2015:

proper names with embedded punctuation should be inside quotes. (since you don;t know how CS will tokenize it)

And generally so should all multiple word proper names.


No that is not working since it is a input pattern..

I would like to know if there is any way to group author and books ...

For example,

author1 has book1,book2,book3
author2 has book2,book15,book7

 

 
  [ # 5 ]

Looking at your topic file…
1. do not put commas between keywords in your topic declaration of keywords
2. ([Dr.Brian Weiss books])  -  [] means find one of. This pattern should have been
u: (“Dr. Brian Weiss”) 
since you should not require books as a keyword.
similarly for others like:
u: ([Many Lives, Many Masters])  that should be
u: (“Many Lives, Many Masters”)

 

 
  [ # 6 ]

In addition, your keywords: hi,welcome,hello for your topic will not work. Those are not normal words like a noun or a verb or adjective or adverb. Those are interjections or dialog acts. You only need the keyword ~emohello which stands for the dialog act hello. 

If you do
:prepare hello
you will see what the actual final input your code would see once CS has standardized your input.

 

 
  [ # 7 ]
Bruce Wilcox - Jan 3, 2016:

Looking at your topic file…
1. do not put commas between keywords in your topic declaration of keywords
2. ([Dr.Brian Weiss books])  -  [] means find one of. This pattern should have been
u: (“Dr. Brian Weiss”) 
since you should not require books as a keyword.
similarly for others like:
u: ([Many Lives, Many Masters])  that should be
u: (“Many Lives, Many Masters”)

Yeah that is working.. But for phrase with ‘:’ and having more than 4 words is showing the error => “Too many words in string *** will never match”

I would like to know the reason..

 

 
  [ # 8 ]

Efficiency reasons. It is unlikely that users would type in: The Doctor and the Soul: From Psychotherapy to Logotherapy

By default the tokenizer splits colons into new sentences, this can be suppressed using #NO_COLON_END added to your $cs_token in your bot definition. But it wont help you here.

I could possibly fix this long book names issue in the future. but it won’t be soon.

 

 
  [ # 9 ]
Bruce Wilcox - Jan 4, 2016:

Efficiency reasons. It is unlikely that users would type in: The Doctor and the Soul: From Psychotherapy to Logotherapy

By default the tokenizer splits colons into new sentences, this can be suppressed using #NO_COLON_END added to your $cs_token in your bot definition. But it wont help you here.

I could possibly fix this long book names issue in the future. but it won’t be soon.

Yeah thank you!

 

 
  [ # 10 ]

Is it good to organize authors and books in tables available in chatscript? I mean in the table format
 
  ~author_name     _~genre   “book_name”  Output


I am refering to the paper “Making It Real: Loebner Winning Chatbot Design” page no.8

 

 
  login or register to react