Virtual Humans Forum
Virtual Humans Forum
Home | Profile | Register | Active Topics | Members | Search | FAQ
Username:
Password:
Save Password
Forgot your Password?

Virtual Humans
Partner website:
Chatbots.org
 All Forums
 Virtual Humans
 Business Applications
 Identify application areas
 New Topic  Topic Locked
 Printer Friendly
Next Page
Author Previous Topic Topic Next Topic
Page: of 5

Joe Repka
Curious Member



USA
34 Posts

Posted - Dec 06 2005 :  18:55:05  Show Profile
The first step should be to list up problems that virtual humans could solve and needs that they could satisfy. Business is all about creating solutions that people will pay for.

What problems can the current generation of vHumans solve?

What kinds of capabilities will the next generation require to solve a wider range of problems?

Edited by - Joe Repka on Dec 07 2005 23:13:07

laackejim
Committed Member



USA
3274 Posts

Posted - Dec 06 2005 :  23:05:40  Show Profile
Joe, You are exactly right. I just finished such a discussion with my Son who is searching for a niche in the computer graphic art world. AND I have some specific ideas about the path and commercial options myself. I will try and get back to you later tonight. Gotta run right now. But tonight or later I will post you with some options.

Jim
quote:
Originally posted by Joe Repka

The first step should be to list up problems that virtual humans could solve and needs that they could satisfy. Business is all about creating solutions that people will pay for.

What commercial attempts have failed. Why did they fail?

How are vHumans being used now? What are their capabilities and limitations? How successful are the commercial applications? How can they be made more successful?

What problems can the current generation of vHumans solve?

What kinds of capabilities will the next generation require to solve a wider range of problems?



Jim
Go to Top of Page

vrossi
Forum Admin



Italy
1455 Posts

Posted - Dec 06 2005 :  23:08:14  Show Profile  Visit vrossi's Homepage
I agree, Joe

I believe that most of us currently can allocate only their spare time to VH, because each one of us has some "serious" daily job, which allows him to earn the money needed for eating, paying the home mortgage, and so on.

Therefore, analyzing the business opportunities of VH means to find a way for some of us to work on VH not only as a week-end hobby, but as a daily job.

Let's think how this field might evolve if we could spend all our working energies on this. So I believe that the questions you are asking are the ones we must answer more quickly.

So, let's start here some brainstorming about all this.



1. Voice recognition system

While TTS are good enough to simulate a real human being, the voice recognition systems are currently nice toys, but can't be used in any practical way. If the VH must become your assistant, you can't be forced to wear a headset microphone when you need to communicate with him. And even using this device, you get so many mistaken inputs that finally you decide to switch to typed input or, even worse, to abandon the chatbot application and run directly the software you need in the traditional way.



2. English-only AI engine

One limitation I feel, living in a non-English speaking country, is that all these technologies are currently working in an acceptable way only in English.

If our conversation in English with a chatbot is sometimes disappointing, a conversation in any other language is almost impossible, or limited to some very specific phrases.

I can't even think about offering some real application to an Italian company, until I can show an Italian speaking bot. The problem is not about just translating some portion of code containing English sentences, or finding some dictionary/encyclopedia in another language, or using a different TTS.
The main problem is about the syntax and grammar processing, which is completely different and much more complex.

AIML, for example, contains some tags which can work only in English, like <person>, which converts "I like you" in "You like me", based on a relatively simple rule. Just think that in Italian the same conversion is from "Tu mi piaci" to "Io ti piaccio", with a lot of variations according to single/plural/feminine/masculine and with different words for each verb tense. The same can be said of other technologies, like Hal, which includes many English-only features.

I know that this is not the main reason for chatbots not being successful (since they are not yet successful even in English speaking countries), but it is surely an evident big obstacle to their worldwide use.

So I think we should go on trying to build more human-like english-speaking bots, but at the same time we should think about some way to keep the language-dependent parts of the engine as a separate component, so that it will be easier to replace it later with a different language engine.






Vittorio
Go to Top of Page

hologenicman
Moderator



USA
3324 Posts

Posted - Dec 06 2005 :  23:22:24  Show Profile  Visit hologenicman's Homepage
Hey there,

I wonder why "Array Microphones" like http://www.acousticmagic.com/products/ haven't caught on. They would seem to free the voice recognition aspects of our v-humans.

A quick internet search for "Array Microphone" finds quite a few products, but they still are not mainstream. Perhaps price is the culprit. Even if not for mainstream use, they would be quite good for dedicated installations. (unless they just don't work)

John L>

HologenicMan

DISCOVERY: The more I learn, the more I learn how little I know.

GOAL: There's strength in simplicity.

NOTE: Goal not always achieved.

Edited by - hologenicman on Dec 06 2005 23:25:37
Go to Top of Page

laackejim
Committed Member



USA
3274 Posts

Posted - Dec 07 2005 :  03:13:23  Show Profile
Okay Joe and Vittorio primarily. Ålso Peter (if he is catching this).

I suggest that what we have here very nearly a critical mass of talent, interest, and opportunity to make something really significant. And I have a suggestion. Because I am terribly VERBOSE (Peter will agree I am sure) I will make it in a series of pieces.

First. Whatever the commercial development turns out to be it will depend on our combined talents.

Second. Peter would seem to me to be the ideal "head monitor" "guide" "teacher" and "mentor". When we come up with something, and I think that when we do it will be at least three pronged, commmercial, technological breakthrough, and move the concept a giant step forward. (are you hooked yet peter?)

Third. Vittorio pointed out the difficulty of seriously promoting a "hobby" while having a serious job to pay for life. Well, I (being an old duffer) am retired and have about 14 to 16 hours a day, seven days a week to focus. I also live in a place that promotes the flow of ideas and solutions and keeps me from being distracted (can't get TV for example). In additon, I am commited to development of VPeople for the reasons Peter identified (or at least most of them)

Fourth. I think we should focus our attention and talents to produce something as "the Plantec people" (hooked yet peter?) . Could be a new book or a new foundation.

Fifth. I have a commercial venture in mind that would meet several of our interests. Produce sufficient value for commercial interests to like it alot. It is difficult enough to be a real challenge and therefore fun to do. Is adaptable across the cultural and language barriers that Vittorio so clearly flagged.

Sixth. This idea was one I was developing with the intent of getting my Son involved and helping him along his way. He has definitely decided that this is not what he wants to do and is making his own way. Therefore I am perfectly willing to share it with others who would like to develop it and the benefits or at least some of them (it would be nice if there were some) could be focused on a foundation to promote the ideas and development of VPeople for all the values they can bring.

Okay. Told you I was Verbose. Let's talk about it. If there is agreement to give it a run I will make available everything I have and think. No, I am not asking for a contract or anything like that. Just some intelligent people who share a vision and who are willing to work together for a solution.

Jim Laacke



Jim
Go to Top of Page

vrossi
Forum Admin



Italy
1455 Posts

Posted - Dec 07 2005 :  10:02:22  Show Profile  Visit vrossi's Homepage
Jim,

I agree completely with you. I think we should put here, in this forum, all the ideas we have. Now we are still in the brainstorming session, so I'd like that everybody says what he has in mind, without worrying if it's a good or bad idea, if it's feasible or not.

In the brainstorming phase, all ideas are useful. In a second phase we will analyze them and try to define a feasible business plan.

This kind of discussion is just what I had in mind when I decided to create this forum.

Hey, Peter, we are waiting for your comments now!




Vittorio
Go to Top of Page

maviarab
Curious Member



United Kingdom
27 Posts

Posted - Dec 07 2005 :  13:11:26  Show Profile  Visit maviarab's Homepage
vrossi,

If you were not already aware there are some half decnt italian tts voices about...bit of playing around who knows?.

Also as far as business applications go, scansoft (formally rhetorical) will actually record any voice for you and turn it into tts...going off the prices they charge for tts voice hire I imagine its bloody expensive but the business potential for a "real" tts voice in whatever accent/dialect/language is available

Ask not what your bot can do for you, ask what you can do for your bot



www.digitalgirl.co.uk

Edited by - maviarab on Dec 14 2005 23:32:13
Go to Top of Page

vrossi
Forum Admin



Italy
1455 Posts

Posted - Dec 07 2005 :  15:30:31  Show Profile  Visit vrossi's Homepage
Thanks, maviarab.

I know several Italian TTS: Cepstral has recently introduced Vittoria, which is not bad and costs about 30$, even if it's not so good as the English voices.

There are some other suppliers, like Loquendo (http://www.loquendo.com) which have Italian voices which are really almost perfect, even if they are usually sold to telephone companies for big projects and are not available as something to install on your PC.

So, the problem is definitely not in finding a TTS, but in creating an engine which is able to manage all the specific issues related to a language.

I copy here something I posted on the Zabaware forum as an answer to a Spanish Hal user:


The problem is not about just translating some portion of code containing english sentences, but about many other aspects.

Hal uses a Wordnet database, which is an encyclopedic dictionary, allowing him to answer many general questions. Something similar should be found for other languages.

But the main problem is about the syntax and grammar processing. When you say "You are nice", Hal has an algorithm which converts it from his point of view in "I am nice". This is relatively easy in English, which basically has no masculin or feminine, and where plural is generally simply a matter of adding an 's'. In Italian, (and i believe Spanish is the same) "nice" can be translated as "bello", "bella", "belli", "belle", according to the comnbination of singular/plural masculine/feminine. Even worse if we analyze verbs. In English you just add "ed" for the past and "will" for the future. In Italian it is such a complex thing that I don't even try to explain it here. Irregular forms are very limited in English (I know there are about one or two hundreds irregular verbs, and that's all), while they are very common in Italian.

So, I think it's a job which would require a big effort. Maybe it will be done by some university.



If you go to Alice site, you will find a so-called AIML Italian version. It's only some little AIML file, containing a few patterns, which can't be really used. Just look inside the Alice files and you will see that they must be rewritten from scratch, and that even some language tags (like <person> or <gender>) imply to rewrite also the AIML interpreter code.

That's why I say that - though I acknoledge the leading role of the anglo-saxon world in chatbot development - I believe that even the best implementations (like Alice and Hal) are basically wrong in mixing up the general bot engine with language-specific statements.

Personally I am happy to chat with my Virtual Human in English (it's a way to improve my knowledge of a foreign language) but there are some other billions people on the Earth who are our potential market and speak a different language. Italians are only 50 millions, so maybe it's not such an attractive market, but what about chinese, indian or spanish?

PS: I appreciate a lot your contribution to this forum. I would appreciate a bit more if you could use a smaller logo in your signature.






Vittorio
Go to Top of Page

maviarab
Curious Member



United Kingdom
27 Posts

Posted - Dec 07 2005 :  15:43:11  Show Profile  Visit maviarab's Homepage
I agree with the language problems, tried to learn italian myself couple years back...gave up in end...could not concentrate on it 100% but yes something needs to be done indeed to combat this problem..again as you mention...china and japan must account for a lot of people.

And apologies for logo..have resized

Ask not what your bot can do for you, ask what you can do for your bot



www.digitalgirl.co.uk
Go to Top of Page

vrossi
Forum Admin



Italy
1455 Posts

Posted - Dec 07 2005 :  15:46:58  Show Profile  Visit vrossi's Homepage
Thank you. This size is perfect .



Vittorio
Go to Top of Page

thenar
Curious Member



USA
69 Posts

Posted - Dec 07 2005 :  17:48:55  Show Profile  Visit thenar's Homepage
quote:
Originally posted by vrossi

I agree, Joe

I believe that most of us currently can allocate only their spare time to VH, because each one of us has some "serious" daily job, which allows him to earn the money needed for eating, paying the home mortgage, and so on.

Therefore, analyzing the business opportunities of VH means to find a way for some of us to work on VH not only as a week-end hobby, but as a daily job.

Let's think how this field might evolve if we could spend all our working energies on this. So I believe that the questions you are asking are the ones we must answer more quickly.

So, let's start here some brainstorming about all this.



1. Voice recognition system

While TTS are good enough to simulate a real human being, the voice recognition systems are currently nice toys, but can't be used in any practical way. If the VH must become your assistant, you can't be forced to wear a headset microphone when you need to communicate with him. And even using this device, you get so many mistaken inputs that finally you decide to switch to typed input or, even worse, to abandon the chatbot application and run directly the software you need in the traditional way.



2. English-only AI engine

One limitation I feel, living in a non-English speaking country, is that all these technologies are currently working in an acceptable way only in English.

If our conversation in English with a chatbot is sometimes disappointing, a conversation in any other language is almost impossible, or limited to some very specific phrases.

I can't even think about offering some real application to an Italian company, until I can show an Italian speaking bot. The problem is not about just translating some portion of code containing English sentences, or finding some dictionary/encyclopedia in another language, or using a different TTS.
The main problem is about the syntax and grammar processing, which is completely different and much more complex.

AIML, for example, contains some tags which can work only in English, like <person>, which converts "I like you" in "You like me", based on a relatively simple rule. Just think that in Italian the same conversion is from "Tu mi piaci" to "Io ti piaccio", with a lot of variations according to single/plural/feminine/masculine and with different words for each verb tense. The same can be said of other technologies, like Hal, which includes many English-only features.

I know that this is not the main reason for chatbots not being successful (since they are not yet successful even in English speaking countries), but it is surely an evident big obstacle to their worldwide use.

So I think we should go on trying to build more human-like english-speaking bots, but at the same time we should think about some way to keep the language-dependent parts of the engine as a separate component, so that it will be easier to replace it later with a different language engine.






Vittorio



-P-
Go to Top of Page

thenar
Curious Member



USA
69 Posts

Posted - Dec 07 2005 :  17:56:20  Show Profile  Visit thenar's Homepage
Vittorio,
Any good NLP engine is language independant. We've developed fully functional vpeople in French and German. A bilingual version of Sylvie used to be at the art museum in Vienna Austria. She can switch databases on the fly...as well as TTS engines and speech recognition engines. It's relatively easy to do. It is a lot of work to develop the databases in different languages, but no more difficult that developing them in English. ALICE is in English because it was developed by English speakers...if you had an Italian database she would work in Italian. But that would take years of development work to reach the depth of the English database. You need to get started on the progect and get a lot of help. I will say we developed both the English and French databases in just a few months time and they are very effective.

-P-
Go to Top of Page

laackejim
Committed Member



USA
3274 Posts

Posted - Dec 10 2005 :  06:04:47  Show Profile
Hello Vittorio,
I thank you very much for your response to my lengthy note. This whole subject area is something I think is very important and I have made a major commitment of my time and energy to working on a solution. I would love to participate in a venture with others to create something.

However, I suspect I should assume from the total lack of other responses, positive or negative, that my suggestion and offer was out of line or out of context for the forum. If the latter is true OOPS I will try to do better.

I would like to ask a favor, however. About 5 months from now I will probably have a information providing, talking and (limited) learning bot or two running. They will be the product of a high school class I am going to be teaching this spring. The favor I am asking of you is to serve as a judge in a contest among teams in the class as to the best representation and communication. It would only take about 20 minutes and is a long time in the future. The reason I am asking now is that if you have no interest I will have time to make other contacts and look for somebody else.

Jim

quote:
Originally posted by vrossi

Jim,

I agree completely with you. I think we should put here, in this forum, all the ideas we have. Now we are still in the brainstorming session, so I'd like that everybody says what he has in mind, without worrying if it's a good or bad idea, if it's feasible or not.

In the brainstorming phase, all ideas are useful. In a second phase we will analyze them and try to define a feasible business plan.

This kind of discussion is just what I had in mind when I decided to create this forum.

Hey, Peter, we are waiting for your comments now!




Vittorio


Uncle Jim (e=mc2)
Go to Top of Page

Joe Repka
Curious Member



USA
34 Posts

Posted - Dec 10 2005 :  07:28:18  Show Profile
Jim, put me on your list of judges.

Tying this in to the topic, education is one industry where commercial potential is significant. Language learning is one more specific area where NL chat bots can be useful to both teachers and learners by providing an interesting means of individualized exposure to language forms, a more realistic practice environment and a testing mechanism.

Presentation of background nad the most basic level of almost any subject area can be facilitated by an AI chat bot that can control a multimedia presentation. Another idea is research guidance and resource discovery for any subject. The idea is to add true and adaptive interaction based on NL to the kinds of things presented by video tape or interactive multimedia programs.
Go to Top of Page

vrossi
Forum Admin



Italy
1455 Posts

Posted - Dec 10 2005 :  08:47:54  Show Profile  Visit vrossi's Homepage
Jim, please add my name to the list of judges.

As for the lack of responses (well, "lack" is not quite correct, since I promptly answered you), you must consider that this Forum was born exactly one week ago, and now has only 22 registered members (plus some lurkers, since it is open in read-only mode to everybody). Therefore, the potential market for any proposal is currently very small.

I believe that in the next weeks more and more people will register and the discussions will heat up. We only need to let more people know about the existence of this place. By the way, may I ask anybody here to put a link to this Forum in any other place where it can be of interest (other related forums, sites, etc.) ?

Bye



Vittorio
virtualhumansforum.com
Go to Top of Page

laackejim
Committed Member



USA
3274 Posts

Posted - Dec 11 2005 :  04:21:02  Show Profile
Vittorio,

You are ON. Thank you.

As to my impatience. I tried to make sure I recognized your response in my note. As for the rest. Well, it may seem strange for an old man to be that impatient but I guess I was. I stand humbled and properly chastised. I guess I felt like I had just finished struggling through the conceptual woods and found my way to the very gates of the Academy, snuck in and found that the big boys didn't pay any attention to my claims that I had solved the riddle of the universe. Hmmmm. thinking of it that way, maybe I should be embarrased rather than chastised.

Jim


quote:
Originally posted by vrossi

Jim, please add my name to the list of judges.

As for the lack of responses (well, "lack" is not quite correct, since I promptly answered you), you must consider that this Forum was born exactly one week ago, and now has only 22 registered members (plus some lurkers, since it is open in read-only mode to everybody). Therefore, the potential market for any proposal is currently very small.

I believe that in the next weeks more and more people will register and the discussions will heat up. We only need to let more people know about the existence of this place. By the way, may I ask anybody here to put a link to this Forum in any other place where it can be of interest (other related forums, sites, etc.) ?

Bye





Uncle Jim (e=mc2)
Go to Top of Page
Page: of 5 Previous Topic Topic Next Topic  
Next Page
 New Topic  Topic Locked
 Printer Friendly
Jump To:
Virtual Humans Forum © V.R.Consulting Go To Top Of Page
This page was generated in 0.23 seconds. Snitz Forums 2000