[OliveOyl3471]

What do you all think about natural language interfaces?

This is for an assignment. I'm supposed to find out all I can about them, so I thought, why not ask the smartest IT people I know?

Personally, I think a natural language interface is pointless. The reason to use one would be for simplicity, but I don't see how it could be simple. You (or a user) wouldn't have to learn, say, DOS commands, but you would still have to learn the syntax that would be accepted by the computer as commands. There are too many different dialects, and different ways of saying things, that a human would understand but a computer would not.

Now, if we could make a computer like the one on the Enterprise, that might be a good use for a natural language interface. Like when Picard says, "Computer, locate Dr. Crusher." And the computer replies, "Dr. Crusher is not aboard the Enterprise." Now that would be cool!

The question also asks what applications a natural language interface would be most appropriate for...but I just can't think of any besides the Enterprise computer I already mentioned.

So do you all have any other ideas for me?

I don't want anyone to do my homework for me, but I am supposed to do research, and I think a part of that is asking professionals about the subject. So this would qualify.

The other part of the question was "Are there any successful systems?" I would say no, but my book says there are some with limited success. I still think it's not even worth looking into, unless you can make an Enterprise computer.

Am I right or wrong? What do you think?

This post has been edited by OliveOyl3471: 04 February 2009 - 12:56 AM

[DoubleFission]

If I had a computer that could understand common Unix commands and executed them with a sarcastic manner (All the while interjecting popular culture references) i'd be the happiest man alive:

"Grep keys pockets"
$> You left your keys in the car dumbass
"Sh*t"
$> Yes...

And now for something slightly more serious:

Generally speech recognition/human interface's are currently rubbish IMO.

To actually be able to implement one successfully, the device would have to be able to not only understand different dialects but also different accents. But that'd be almost impossible, however if you got past that there would still have to be at least SOME syntax that people must follow to be able to successfully communicate with the machine.

For example, the Star Trek way:

Computer (Addressing the computer, the equivalent of typing in the terminal or using the mouse), Locate (Command which would execute a certain method accessing different internal ship sensors) Dr. Crusher (Parameters of the method, arguments being passed).

These commands could easily be programmed into a computer through a Hash table for effective searching (of possibly limitless commands).

However that is a downfall of the Hashtable, the programmer should know how much data is going to be contained in it. If commands are going to be added, changed taken away the program behind the H.I. would have to change the hashtable to accommodate (If it's a large change anyway).

Finally, verbal communication is generally a slow medium, many people can communicate faster and more accurately in other ways. Therefore I personally believe the next biggest steps in Human Interface should be made with touch/virtual technologies.

If it was possible to have a virtual reality operating system, those who use it would be much better equipped to be able to interface with a computer than to get a computer to interface with us.

So my theory is, it's probably easier and more practical to integrate Humans into computers than to get the computers to be able to adapt to interface properly with us.

[OliveOyl3471]

Thank you for the response. It appears that you agree with me.

I suppose I should have included some code, in order to get more people to respond.
[rules][/rules]