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..

  [ # 16 ]

Not at all: )
Now I can dive into ChatScript again.


  [ # 17 ]


replacing “Welcome to ChatScript” in
by “Schön, dass du da bist!”
I am recieving a “Bad UTF-8”-message.

Perhaps something for your new release, Bruce.




  [ # 18 ]

You’ll have to save the file as UTF-8.  I use wordpad and save_as and pick the format.


  [ # 19 ]

Thank you and 3 more questions:

I want to widen the bottleneck of a beginning conversation
and created a random-list of intros instead of “Enter your name”,
but I couldn´t find, where I can change it. Could you tell me where it is?

How can I change the name of Harry while saving all of Harrys capacities?
I want to modify them.
Is it done by simply changing the botprompt from HARRY to MALDIX in macros1.txt?

I want to translate and expand the given concepts and share it with you.
What is the most useful way for ChatScript I could do it in your opionion?
Shall I alternate like:

concept: ~sadwords(sad traurig blue bedrückt unhappy unglücklich sorrowful betrübt sorrow Bedauern)





  [ # 20 ]

1. instead of “Enter your name”—— Harry doesnt have this now. So I’m not sure what you are referring to.  All conversation of harry is in

2. Harry’s name is in
#! what is your name $bot = harry
u: ($bot=harry what is your name) My name is Harry.

and in the bot definition of $botprompt = ^“HARRY: ” and the name of the bot internally:
table: defaultbot (^name)
^createfact(^name defaultbot defaultbot)

If you rename the outputmacro defining harry, rename the defaultbot naming harry, rewrite any rule saying $bot=harry, and rewrite the botprompt, you will have renamed your bot.

3. For supplying augmented concepts… since your concepts are german words, I would run extended concepts:

concept: ~sadwordsUniversal (~sadwords ~sadwordsGerman)  and build a sadwords german concept.
That way I could have concept words by language, and maybe make that a compilable option somehow.


  [ # 21 ]

Point 1)
Mycitation was not correct.
It was “Enter user name” what I was looking for.
Now I found it in “main.System.cpp” and “testing.cpp”.
Is it possible to have a random-list there, like
[Hi, please tell me your name.] [Nice to meet you. How shall I call you?] ...

Point 3)
I think I should use English for concepts and topics,
because there are a lot of english loanwords in German
and because I want Maldix to speak both English and German some day. It would look like this:
concept: ~CHILDHOOD (child childhood kid little memory young younger history ~childhoodGerman)
concept: ~CHILDHOODGERMAN(Kind Kindheit Kid klein Kindheitserinnerung jung jünger jüngsten)

I think (please tell me if I´m wrong), after that, I can take the concept to create a topic, too? Like this:
topic: ~CHILDHOOD ()
u: ([young Kind]) ^noerase()
[Magst du Kinder?] [Kinder sind doch was Wunderbares!]
Is it right, that I would recieve one of these german answers this way?

Could you tell us something about the relationship between concepts and topics in ChatScript?




  [ # 22 ]

1.  Currently controlling the input user name prompt is not possible as a stand alone. You can do it with server because you control the webpage. I would not, for a client, try to do this at present.
3.  I am reluctant to make the released concept sets share languages data for the basic sets.  I have no problem with defining supersets, which is why I suggested that ~childhood and ~childhood_german could be defined from ~childhood_universal or whatever. That is, users of the English only form would not want german words tacked onto ~childhood.

As for the relationship of topics and concepts.

1. One cannot use the same name for both a concept and a topic. A name is one or the other.
2. A topic declares itself whose name acts precisely like a concept as well as having a collection of rules.
3. TYPICALLY a concept (particularly ones predefined by me) are a collection of words that essentially mean the same thing or are all instances of something.
4. TYPICALLY the keywords of a topic are words that are associated with it, but do not all mean the same thing. USUALLY to see the keyword of a topic in a sentence is to think maybe that the topic should be checked for reaction.
E.g.  topic: ~coffee (coffee latte cappuchino)
But I wouldn’t put cup or saucer as keywords of ~coffee because merely seeing cup in a sentence does not bring coffee to mind “I borrowed a cup of flour”. 

So no, you cannot declare topic: ~childhood.  You could declare topic: ~childhooda (~childhood ~childhood_german)



  [ # 23 ]

Separating topics and concepts on the one hand
and an english version from a german version on the other hand are very good ideas.

Would you mind if I would take topics as the X-axis of my bots mental activity,
containing associations of the topic,
while concepts would be the y-axis of it in a stricter object-orientated sense like yours?

For example:

topic: ~Stalin ( Moscow Kreml USSR Russia )
concept: ~dicators ( Stalin Hitler ... )
concept: ~sole_ruler ( ~dictators ~kings…)

Doing this, MALDIX could learn (by machine-learning tools)
that he NEVER likes dictators (and could handle every new dictator in an equal way), but he likes kings under some conditions (for example Henry_IV_France because of his humanity)


  [ # 24 ]

that’s fine.


  [ # 25 ]

Working on an ontology in concepts for MALDIX,
I am asking myself some questions:

If I get it right, User Manual (page 14) says,
that a concept is becoming hierarchical by using these [ ] brackets instead of these ( ):

concept: ~pokerhand [ royal flush straight flush 4 of a kind full house ] 

Shouldn´t it look like this:

concept: ~pokerhand [ royal_flush straight_flush 4_of_a_kind full_house ] 

to avoid ChatScript from supposing that “royal” and “flush” are 2 poker-hands?


Is the better, bigger… example always on the left side,
decreasing slowly until the worst or least is reached, like:

concept: ~animalsfrombigtosmall[ elephant horse ... mouse fly] 


concept: ~femaleselectioncriteriatowardsmen [ tall rich ... beautyful ethical ] 


Is it possible to make it standard practice that ChatScript is able
to climb its own ontological ladder of concepts? 2 examples:

“hammerhaed_shark” has no own rule, but can be found in concept “shark”,
but there is no rule for it, so ChatScript executes the rule of the next-higher concept “fish”.


“Paul_Desmond” has no own rule, but can be found in concept “Cool Jazz”,
but there is no rule for it, so ChatScript executes the next-higher concept with a rule: “music”

How could this look like?



  [ # 26 ]

There is no distinction between using [] and using ().  Original chatscript used () for concept and [] in patterns. Someone said it would be more consistent if it used [] for concepts/topics, so now it accepts either for concepts and topics.

There is no requirement on ordering on a concept or topic keyword list. I personally used an order for the poker hands because of the way I intended to use it.  And it could have been greater first or lesser first. that was arbitrary. Merely that it was ordered.

Chatscript automatically matches patterns using the entire ontology hierarchy. This happens automatically for concepts, so if you had a concept ~fish that had the hammerhead shop in it, such rules would match.  The system does NOT automatically do it for the mere word fish, but if you had used a pattern with a specific definition of wordnet that had the hammerhead shark as a subnode, then it could be matched automatically (I think).


  [ # 27 ]

I had a telephone-call right now with Prof. Hinrichs,
the leader of the german WortNet GermaNet,
(which is with costs) and asked him,
if it would be possible to use it for free
under the conditions of our purposes.

Unfornately, he denied.

This confirms me to create my own little ontology
especially for Chatbots: ONON, an Ontology Of Needs,
because I´m sure that this is the point most people
are talking about. If I had the chance to nest concepts
in a hierarchical way, I could do things like:

concept: ~daytodayroutineweekday[[ awake watch_clock stand_up ~fixingbreakfast ... ~gotobed] ]

In this way, I could teach a bot many things
like what comes next and ask for it I don´t know if this is possible without a feature-request
for a visible ordered concept like “[[]]” (just an example).
But how could my bot know that humans cannot fix their breakfast
without a previous stand_up from their bed?


  [ # 28 ]

why does the concept need ordering visually with [[ ]].  Your script can treat any concept as ordered on its own.
If you are saying it wont know which concepts are ordered and which are not, you could always add an extra annotation on a concept. This could be done as a fact, or merely as part of the name.. eg

concept: ~ordered_day [ wake breakfast lunch dinner sleep ]

And you could write code that could know the concept was ordered because the word ordered was in the name.
Or one could tell this way:

concept: ~ordered_concepts (~ordered_day)


  [ # 29 ]

You are absolutely right.
I was nearly sure that you would show me the shorest way to an aim,
that I started to reach via Timbuktu again: )


 < 1 2
2 of 2
  login or register to react