April 22, 2009 at 9:10 pm
Comments posted to this topic are about the item Interfaces
April 23, 2009 at 5:43 am
I'm surprised at your comment about voice recognition software. My wife is a medical transcriptionist who uses Dragon Naturally Speaking with nearly 100% accuracy. I think the misconceptions about Dragon are because most people aren't aware what's involved with putting a system together. If you just install the software and start dictating there will be errors. The system needs to be "trained" by feeding it dictation and correcting the errors. It really does learn. Dragon is also hardware dependent. You need a fast system with plenty of RAM and a GOOD microphone and sound card. There's no question that a properly trained voice recognition system with appropriate hardware will outperform manual typing, particularly with something repetitive such as filling forms.
April 23, 2009 at 6:19 am
I've found that data entry errors are not hardware/software dependent so much as they are user influenced. My company teached people to use software such as Dragon Dictate and Viavoice and with suffucient training we've seen people with 95% accuracy. I've also seen people with a keyboard have terrible accuracy. We've suffered with finding errors in our data and most of it tracks to specific users. Some people are not good at data entry not matter what hardware/software configuration they use. You must have a mind set to sit at a desk and key all day.
Just like not everyone can be a good programmer or a good DBA; you have to enjoy what you are doing.
If you could take the human element out of data entry then you could reduce errors.
Ah a technocracy!
April 23, 2009 at 6:58 am
This topic has two main thrusts, voice input and data validation.
Voice recognition as a general input device will never work. Ever. In certain domains it's a perfect fit--such as giving short commands (a few words) when your hands are busy (driving, for example). In that domain there's simply nothing better.
But general computer input? Nope.
I experimented with a PERFECT V/R input system. It could convert text as fast as I spoke, it understood context at a human level, and also knew exactly what was going on.
In case you haven't guessed the "V/R input system" was a person. 🙂 You can't get better VR than that! My girlfriend at the time was curious about V/R so we tried experiments where we alternated as "input system".
Worked flawlessly. There were some serious problems tho.
1) Input speed. Typing for a skilled typist is simply faster. I can cruise all day long at 50 WPM, and hit bursts of 80. No one talks that fast. 🙂 Inputing numbers via voice is clumsy and error prone and signaling end of field can be tricky. Not to mention the arbitrary punctuation in programming languages.
2) Accuracy. Even a "perfect" system (a human) can be confused by context, pity the poor computer!
3) Your voice gets tired. Speaking for hours at a time? Egads.
4) You bother others around you. Phone centers are one thing, but for work requiring quiet concentration? Say developing an application? 🙂 Or a library? Not to mention programming languages are not suited for VR.
Given 3 and 4 alone VR for general purposes is a bust. And can you imagine entering a password via voice? 🙂 'Nuff said.
Now on to data validation in general.
When you have people in the mix there are going to be problems. The best approach I've found is to catch the error as soon as possible. Simple things like not putting letters in a numeric field have pretty much been covered. Then there's boundry checking, and drop down lists--well, you know the drill.
Back in the punch card days (and yes, I'm dating myself :)) critical data input was done *twice*, by two different people. If the records didn't match they were rejected and sent back for input again. Obviously that's horribly expensive but it is effective.
Keeping people out fo the loop as much as possible is really the only answser. Don't double-enter data, don't have redundant data. SPOT, all the way. 🙂
Of course once data's in it really needs to be verified--especially addresses. We found this out the hard way at work...
April 23, 2009 at 7:00 am
I think voice recognition has it's place. Even in Star Trek, they don't use voice for everything. Things like queries (computer, where is Capt Pickard) or running simple things (Computer, start Word 🙂 ).
But for a lot of things, driving the ship, controlling the engine room - they still use keypads.
I think that will always be - at least until we find a way to get our thoughts directly into the computer. Until then, I think we'll always be able to type faster than we can speak.
My two cents.
April 23, 2009 at 7:17 am
I suppose the only problem I really have with keyboards is that the keys for your left hand slant the wrong way. Considering how narrow keyboards are compared to how wide a person's shoulders are, your arms are natrually positioned like /
but almost all keyboards are layed out like \ \. I can live with the key arrangement, it's OK, I don't necessarily think Dvorak keyboard layouts are the answer, although I've barely ever used them.
The real answer of building better interfaces will be with building smarter systems that can give the user fewer choices, like how modern UIs have drop down combo boxes instead of simple prompts where the user had to remember a bunch of codes and symbols to type in manually. But for long messages or strings of text there will always be the need to type it out character by character on a keyboard.
April 23, 2009 at 7:53 am
I do a lot of DC work. I prefer ADC whenever possible. Take, for example, a fuel delivery person. I know which fueling station is being serviced because of the GPS coordinates when the truck pump is switched on. I have level sensors in each tank. I know the level at turn on and at turn off. It's a simple matter of comparison to know how much fuel was dispensed at each delivery. I know arrival and departure times from Geo-fencing. What is hard to determine is what fuel grade got pumped into which station tank. I still have a manual data entry screen on the truck OBC to help out the driver when her sensor net fouls up.
We have been working with RFID for a while in warehouse environments. I can tell where every forklift is in a warehouse. If I tag all the boxes and pallets I can then sense what is being picked up and put down. The backup to RFID is barcoding and the backup to that is keying. I want to play with VR in warehousing.
I used to do software for dentists. Periodontal charting is a great candidate for VR. There are only 33 teeth with 6 pockets each. The depth is on a limited scale. The readings are taken in a specific order.
ATBCharles Kincaid
April 23, 2009 at 8:08 am
Most interfaces are doomed to irrelevance (if not failure) because they are more 'gee wiz' than practical. Touch screens (except for special apps like ATMs and kiosks) work best for simple menu selections, but often require you to hold your arm in an unnatural position. Gesture based systems may great demos, but are often too vague for realworld stuff. VR has potential in a number of areas where context and distraction is limited (would you rather work in a roomfull of people typing, or a roomful of people dictating to their computers?)
A good interface needs to be fast, unambiguous, not physically tiring, and simple.
[BTW am I the only person who HATES laptop touch pads? They slow me down enormously, and are very fatiguing. I can't imagine doing any serious work with one of those)
...
-- FORTRAN manual for Xerox Computers --
April 23, 2009 at 8:17 am
I am undecided when it comes to typing out full text if I can speak faster than type however I am pretty certain that an individual skilled in 9 key can easily input data faster than dictating it. I think the input environment will dictate speed which will be the ultimate deciding factor. Field workers such as sales personnel and insurance claims reps will greatly appreciate dictation where accountants probably would opt for an old fashioned keyboard.
April 23, 2009 at 8:35 am
jay holovacs (4/23/2009)
[BTW am I the only person who HATES laptop touch pads? They slow me down enormously, and are very fatiguing. I can't imagine doing any serious work with one of those)
You are NOT alone. Every work with a laptop that had the Joy-mouse button in the center of the keyboard? A little red stud that stuck up between the keys. Push it toward the Q row and it moves the cursor up. Push it toward the Z row and it moves the cursor down. Left is toward A and right is toward L. It's slow but much more precise than the pad.
ATBCharles Kincaid
April 23, 2009 at 8:47 am
Interesting responses.
My wife says Dragon is great, but with ambient noise, and enough mistakes, it isn't good enough. There's also the fatigue factor, and it doesn't go nearly as fast as she can type. I think some of it is that you don't necessarily know what you want to write. You put something on the screen and then might want to change it, reword, etc. For that, a keyboard works much better.
Perhaps it's not a bad alternative to switch to at times. I know Jerry Pournelle (sci-fi author) uses it quite a bit and likes it.
Star Trek does use keypads, and I get that, but they're inconsistent in how the computer responds. I see so many places where the computer just "knows" when they're talking to the computer and when they're talking to people. Granted, it's staged, but it does show some of the issues you might have.
April 23, 2009 at 9:29 am
If you really want speed and accuracy for input you should try a wireless brain-wave reader. The only problem is that the words I misspell in my head still come out wrong...nothing you can't fix with a good ol' spell checker.
Going the other way, the image implanter (again, you have to go with the wireless version) works great! You can "see" much more than even what a dual monitor setup can show you. 😉
April 23, 2009 at 9:42 am
All you need to do after the wave reader is hooked up is pipe the data over to Twitter.
April 23, 2009 at 11:20 am
A good interface is what ultimately works for the end user without regard to how hard we have to work to make the data right by scrubbing, storing, backing up, etc. We exist only for the end user. Whomever it is, they are our customers. A good interface that works for the intended users can make our jobs more difficult. It is a challenge I welcome.
April 23, 2009 at 12:08 pm
Charles Kincaid (4/23/2009)
jay holovacs (4/23/2009)
[BTW am I the only person who HATES laptop touch pads? They slow me down enormously, and are very fatiguing. I can't imagine doing any serious work with one of those)You are NOT alone. Every work with a laptop that had the Joy-mouse button in the center of the keyboard? A little red stud that stuck up between the keys. Push it toward the Q row and it moves the cursor up. Push it toward the Z row and it moves the cursor down. Left is toward A and right is toward L. It's slow but much more precise than the pad.
You are definitely not alone. I don't see why they were even invented, but maybe that's just me. 😀
Give me a good 'ol keyboard and mouse. That's all I need. I do recommend the Microsoft ergonomic keyboard which helps with the arm positioning problem and potential carpal tunnel.
Viewing 15 posts - 1 through 15 (of 26 total)
You must be logged in to reply to this topic. Login to reply