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..

generating utternaces
 
 

Hello guys,


I would like the process of generating utternaces in generation dialogue systems !!! what is the process of generating dialogues so I can be able to programm !!! plus how I can use the grammar rules !!!

can any one help me and upload some intreseted papers or what ever ???

Thanks in advance

 

 
  [ # 1 ]

Hi Scarycaspir,

What type of responses are you looking to generate? Tell us a little more about what your trying to do.


Vince

 

 
  [ # 2 ]

Hi Vincent,

what I am trying to do is talking with a chatterbots lets say and based on the ontology I am working on he replys. I did some diagram shows how the texts will generated accorinding to the grammar rules but still it is completed and something missing in it.


hope you understand me !!

thansk alot

 

 
  [ # 3 ]

Have you tred setting up an account with one of the existing chatbot services just to get started? You could try pandorabots or Rive Script or Personality Forge.

Vince

 

 
  [ # 4 ]

DETERMINER => A || DETERMINER  => AN || DETERMINER  => THE
NOUN  => CHATBOT
NOUN  => HUMAN
VERB  => MIMICS
NOUNPHRASE  =>  NOUN ||  NOUNPHRASE  =>  DETERMINER  NOUN
VERBPHRASE  =>  VERB ||  VERBPHRASE  =>  VERB  NOUNPHRASE
STATEMENT  =>  NOUNPHRASE  VERBPHRASE
UTTERANCE => STATEMENT

This is a rough draft pseudocode.  Please suggest any improvements or corrections.

 

 
  [ # 5 ]

I used some of grammar rules as 8PLA has written !!! but there is something I do not get it, when my supersiviour said how you going to generate the sentences in what way, what does he mean ?? the system I am going to talk to should be able to generate utternaces proper utterances in terms of the structure and concept as well !!! I hope you guys understand me !!

Here are some grammar rules : please tell me if they are right or do I need to add or change ??


S => NP   VP EOS
S => VP VPEOS  
S => VP PPEOS  
S => NP VP VP EOS
VP => V P
VP => V S
VP => V NP
NP => ProperNoun (PN)| CommonNoun (CN) 
VP => V Adjective
NP => Det Noun  
NP => Det Adjective N
NP => Adjective N
EndStence => Q | E | ST

P.S : I need to specify more types of NP

any suggestion ?? I wpuld be happy for that


PP => P Noun

 

 

 

 
  [ # 6 ]

I’m confused about what you are trying to achieve. Are you trying to build a system that transforms a database into natural language output? Are you trying to parse natural language? I think you need to clearly write down the specific goals you are trying to achieve and what tools you have at your disposal (databases, ontologies, parsers, etc.).

Some places to get started if you aren’t sure what I’m asking:

http://en.wikipedia.org/wiki/Parsing
http://en.wikipedia.org/wiki/Natural_language_generation
http://en.wikipedia.org/wiki/Ontology (While on that page, click on the link labeled “Ontology (information science)” - The forum software has issues with parens in links)
http://en.wikipedia.org/wiki/Commonsense_knowledge_bases

 

 
  [ # 7 ]

CR, I edited your post a bit, to clean up a minor glitch with the forum software. Sadly, there is no work-around for this particular glitch. downer

[edit and off topic]

It looks as if it’s not a “glitch”, after all:

Parenthesis aren’t allowed in URLs in ExpressionEngine because in certain circumstances it can cause parts of the URL to be interpreted as code, and are used as part of cross-site scripting hacks.  Unfortunately because some browsers will even interpret URL encoded parenthesis in this manner and still allow code execution, ExpressionEngine takes the high road erring on the side of caution.  URL encoding makes invalid characters safe for transport in a URL, but note safe to use.  The action ExpressionEngine is taking keeps you and your site’s visitors safe.

—Derek Jones - Ellis Labs

I guess we’ll just have to put up with it.
[/edit]

 

 
  [ # 8 ]

Ha, I had a feeling when I saw the parentheses that it wouldn’t post properly. I wish wikipedia would get rid of parentheses in their urls, especially if they are a security concern. I can’t recall any other major site that uses them.

 

 
  [ # 9 ]

[even more off topic, but it’s the last, I promise!]

Whilst considering this particular dilemma, I considered creating a script on my website that would act like tinyurl does (sort of), where you could enter the wikipedia URL into a form, and it would spit out a “simplified” URL that my site would then use to redirect back to the proper wiki URL, but those same security concerns made me think otherwise. smile

[Ok, no more cluttering up the thread with OT material]

 

 
  [ # 10 ]

what I am trying to do is generating the sentences so I have to do several analysis : starting with lexical, syntactic and semantic one as well !!!! and to create sentences I need grammar rules as well !!!

I need dictionary aswell (built in dictionary) for these sentences to be generated !!

do you think guys WordNet is a good choice for using as dictionary or not ??

 

 

 
  [ # 11 ]

WordNet is a good start. There are a number of free ones available. I’ve compiled a list of them on my website at http://wixml.net along with conversions of some of them into a more convenient relational database format.

 

 
  [ # 12 ]
Andrew Smith - Mar 16, 2013:

WordNet is a good start. There are a number of free ones available. I’ve compiled a list of them on my website at http://wixml.net along with conversions of some of them into a more convenient relational database format.

Interesting. Is this sql version different from the one I am using (I didn’t create it , just got it from the net)?  More specifically, does it have the synonyms in the correct order?

 

 
  [ # 13 ]

WordNet is great for determining parts of speech, synonyms, some hierarchal relationships, etc. I use it in my project for parsing purposes to build up a knowledge base, to perform morphological functions, etc. But WordNet doesn’t supply the type of “common sense” or relational knowledge you would need to put together sentences that make sense. That is, WordNet doesn’t associate the word “dog” any more closely with “bark” than with “hover”. smile

However, if you want to generate meaningful sentences, then it would be better to start from an ontology. (See, for example, OpenCyc.)  Your project would then be focused on converting from the internal database structure of the ontology into a natural language string. Fortunately, you would only need to concern yourself with the subset of grammar represented by the ontology relations.

What I would recommend is putting together a list of resources you would like to use. Then figure out which languages or wrappers you would need to put these resources together in a single program. (What programming languages do you know, btw? Do you have a preference?) Then it’ll be more straightforward to fill in the “missing” pieces with your own code.

 

 
  [ # 14 ]

Jan I don’t know which version of the WordNet database you would have. I wrote my own conversion software for all the databases on my site because I found in a lot of cases that existing conversions were inadequate. However someone may have copied them from my site and made them available for download somewhere else, so you would have to directly compare them to find out if they are the same or not.

 

 
  [ # 15 ]

Guys I tried to find a dictionary but I did not find any !!! I built a temrary dictionary with 3.000 words but it is not enough for the project !!! does any one have an idea how I can get a built-in dictionay consists of words, meaning, type (n, v, adj, ..ect).

 

 

 1 2 > 
1 of 2
 
  login or register to react
‹‹ Phrases      PC-PATR: A syntactic parser ››