The idea is that any character sequence in the source can be found in time only proportional to the pattern length, not the data size.
The penalty is a bit of space for indexing, but methods for compressed indexing have been found which use only about 40% of the source text size to hold both the index and the source text.
IMHO, much of the performance problem has already been solved, so the question is really whether people would use a tool if it were developed.
I've seen this at management-focused companies as well: middle managers and "product" people are under the gun to produce quarterly results, while sucking up to technically unskilled senior managers who see search as a commodity to be "improved" by marketing bombardment. To an MBA, everything looks like a marketing problem.
These people define the word "short term". They come up with ideas like simply raising fees or selling listings to raise revenue, getting a short-term gain until the market inelasticity is used up and people move elsewhere, and the company is forced to come up with another bright idea.
Technological development on the order of Google requires an investment in knowledge and research that companies like Yahoo would have trouble even conceiving. The company farmed out everything technical (at one point they were even in discussions to farm out the production of their directory). Their main contribution to search technology seems to be the sub-200 ms response time, since user surveys indicated that the main selling point for a search engine was how fast it came up with results.
gzip finds repeats among the most recent 32K of the stream it's processing, using a hash table etc. to match its current position against previous ones.
IIRC it hashes the three bytes from its current position and looks for a match against hashes from 32k previous positions, then does a lookup in the hash bucket for as much as it can match following the initial 3 bytes.
The BWT actually sorts every position in the block. It's not streamable in any significant way.
In some ways this technique is meant to defeat systematic biases like the ones you mention. Compression tools make few assumptions about the data they process, so they serve as a check against more tailored filters which may introduce artifacts, or be defeated in some way. This problem may occur because they look for pre-selected "features" in the data rather than looking at the distribution of the data as a whole.
gzip isn't perfect, but it will find repetitive byte sequences of any kind, regardless of the type of data. It's more of a sanity check than a knowledge extraction method.
Right. Any statements containing unhealthy phrases like "laptop heatpipe", the aforementioned "cyberballs", or any connection or causation between them should be omitted.
This is slashdot, after all, and we have standards to uphold.
That show came to mind when I read this story, and I remember noting that the character could sense an oil drum 1000 miles away, but he had no problem getting into an airplane filled with "toxic" foam, plastic, gas, oil, and paints, and flying all over trying to "escape" the toxicity.
The funny thing was that later I was at a public meeting where some "chemically sensitive" person got to speak first (because she was disabled, of course), and her main topic was how she should be able to park her car anywhere she chooses. I guess those little pine tree air fresheners that people hang from their rear-view mirrors really work, because she got right back into her toxin-mobile after the meeting, with no ill effects. Irony is apparently not toxic.
My home is already powered by negatively-ionized hydrogen. That's right, I'm already using this 21st century power source right in my own home. Fossil fuel is burned at a power plant, converted into Protonless Gas and Energy (PG&E), and delivered to my home via special metallic conduits.
Simply by removing one positively charged particle from the hydrogen nucleus, we can reduce this hazardous gas to an easily transported, relatively safe energy source. I predict that in the near future, nearly all homes will use this fascinating new technology.
It has nothing to do with the weather. The massive concentration of 7200 RPM drives in the Northern Hemisphere is diverting our precious angular momentum into spinning disk platters.
To counteract this phenomenon, we must get half of the drives rotating in the opposite direction. If your IP address ends with an even number, remove your disk drive and turn it upside down immediately.
My rule of thumb when evaluating how "good" a company is, is: Does it need to be? If a company has too much power, then, in all probability, it will be corrupted. That's why we have limits on corporate behavior, and the economy (in the long term) supports replacement of powerful monopolies with smaller, more versatile suppliers.
Another rule of thumb: does the company need to be secretive? Unfortunately Google has gone down that path with its ranking algorithm, in order to prevent people bombing it with made-up links. But, that means that both business and non-business web pages no longer know how they're ranked.
It's not necessarily a matter of intent, but rather potential for misuse that we have to look at.
Oh yeah, and as for physics... Ships with impulse drive can go like 0.99c. If you decide to ram another ship, you're going to end up with a big cloud of plasma and debris, not some lame "crunch".
Oh come now, don't you know that all space ships are surrounded by invisible force fields which slow them down to 5 mph during collisions? That's what their bumpers are rated for.
It's common knowledge, just like the fact that skin-tight polyester is the best armor against phaser blasts and unknown planetary hazards.
This article doesn't say much of anything, but I've found some interesting sonar stuff on my own before.
One technique that is, IMHO, underappreciated, is phase-conjugate sonar. When a reflected signal is received by a network of sensors, it is played back in reverse from all the sensor locations. The time reversal causes the wavefronts from all the emitters to arrive back at the target simultaneously, creating an even stronger reflection, and probably making the target's ears ring.
I've been thinking of using this technique for a subwoofer blaster, but I haven't gotten around to it. Maybe medical science should be given priority.
Several knowledgeable people pointed out that the first try was limited by gzip's 32k window size, so I did a quick run with bzip2, which uses a 900k block, and put the results here. Somewhat different, but still a spread between spam/ham.
I've worked on cylinder locks a few times, and that method doesn't make sense to me, at least with the ones that I've seen.
Cylinder locks work by having the key and cylinder turn together. The cylinder is locked by pins which interfere with turning unless the pins are pushed to the correct levels by the key notches. The key itself just sits in a slot in the cylinder and doesn't hit anything while turning.
The key does not block turning except indirectly, by pushing pins perpendicularly to the turning motion.
There's a graphic with the NYT story that shows the mechanism; I don't see how the key could become scratched in a way that would reveal the pin heights.
However, if you turn the key blank hard enough, you may be able to open the lock, for the same reason that a hammer is a good fix for mistakes.
Looksmart is not looking that smart as it tries to port this thing to a Linux cluster. The NT version has had known scaling problems since the acquisition last summer. $9M got them a prototype, and they've been through two or three engineering directors since that time.
Slashdot Mirror
The idea is that any character sequence in the source can be found in time only proportional to the pattern length, not the data size.
The penalty is a bit of space for indexing, but methods for compressed indexing have been found which use only about 40% of the source text size to hold both the index and the source text.
IMHO, much of the performance problem has already been solved, so the question is really whether people would use a tool if it were developed.
From Ernest and Julio Gallo.
Are doomed to watch it in car ads.
As experimental film goes, so goes advertising 20 years later.
This film has one mistake too; somebody has to dash onto the set and tweak something at one point to keep it going.
I've seen this at management-focused companies as well: middle managers and "product" people are under the gun to produce quarterly results, while sucking up to technically unskilled senior managers who see search as a commodity to be "improved" by marketing bombardment. To an MBA, everything looks like a marketing problem.
These people define the word "short term". They come up with ideas like simply raising fees or selling listings to raise revenue, getting a short-term gain until the market inelasticity is used up and people move elsewhere, and the company is forced to come up with another bright idea.
Technological development on the order of Google requires an investment in knowledge and research that companies like Yahoo would have trouble even conceiving. The company farmed out everything technical (at one point they were even in discussions to farm out the production of their directory). Their main contribution to search technology seems to be the sub-200 ms response time, since user surveys indicated that the main selling point for a search engine was how fast it came up with results.
gzip finds repeats among the most recent 32K of the stream it's processing, using a hash table etc. to match its current position against previous ones.
IIRC it hashes the three bytes from its current position and looks for a match against hashes from 32k previous positions, then does a lookup in the hash bucket for as much as it can match following the initial 3 bytes.
The BWT actually sorts every position in the block. It's not streamable in any significant way.
It sends an out-of-band message, in this case the waistband.
In some ways this technique is meant to defeat systematic biases like the ones you mention. Compression tools make few assumptions about the data they process, so they serve as a check against more tailored filters which may introduce artifacts, or be defeated in some way. This problem may occur because they look for pre-selected "features" in the data rather than looking at the distribution of the data as a whole.
gzip isn't perfect, but it will find repetitive byte sequences of any kind, regardless of the type of data. It's more of a sanity check than a knowledge extraction method.
Right. Any statements containing unhealthy phrases like "laptop heatpipe", the aforementioned "cyberballs", or any connection or causation between them should be omitted.
This is slashdot, after all, and we have standards to uphold.
I want to make it clear to everyone here that any reference to the "glowing cyber balls" story, however indirect, is strictly forbidden in this forum.
That show came to mind when I read this story, and I remember noting that the character could sense an oil drum 1000 miles away, but he had no problem getting into an airplane filled with "toxic" foam, plastic, gas, oil, and paints, and flying all over trying to "escape" the toxicity.
The funny thing was that later I was at a public meeting where some "chemically sensitive" person got to speak first (because she was disabled, of course), and her main topic was how she should be able to park her car anywhere she chooses. I guess those little pine tree air fresheners that people hang from their rear-view mirrors really work, because she got right back into her toxin-mobile after the meeting, with no ill effects. Irony is apparently not toxic.
The article has an admirable, five-point plan:
1. Solve the hydrogen fuel-tank problem.
2. Encourage mass production of fuel cell vehicles.
3. Convert the nation's fueling infrastructure to hydrogen.
4. Ramp up hydrogen production (glad that one got in there, eh?).
5. Mount a public campaign to sell the hydrogen economy.
.
However, it just so happens that I have a better, SIX-point plan:
1. Solve the AA battery fuel-tank problem (already pretty much solved, just need to keep the batteries from falling out when we drop the car).
2. Encourage mass production of AA battery vehicles (well underway).
3. Convert the nation's fueling infrastructure to AA batteries (many stations already sell them).
4. Ramp up AA battery production.
5. Mount a public campaign to sell the AA battery economy.
6. Profit
Note that my plan not only has more points, but, unlike the first plan, it will generate a profit!
My home is already powered by negatively-ionized hydrogen. That's right, I'm already using this 21st century power source right in my own home. Fossil fuel is burned at a power plant, converted into Protonless Gas and Energy (PG&E), and delivered to my home via special metallic conduits.
Simply by removing one positively charged particle from the hydrogen nucleus, we can reduce this hazardous gas to an easily transported, relatively safe energy source. I predict that in the near future, nearly all homes will use this fascinating new technology.
It has nothing to do with the weather. The massive concentration of 7200 RPM drives in the Northern Hemisphere is diverting our precious angular momentum into spinning disk platters.
To counteract this phenomenon, we must get half of the drives rotating in the opposite direction. If your IP address ends with an even number, remove your disk drive and turn it upside down immediately.
Wireless access points are easy to find along the highway; just look for the skidmarks and wreckage left by previous users.
My rule of thumb when evaluating how "good" a company is, is: Does it need to be? If a company has too much power, then, in all probability, it will be corrupted. That's why we have limits on corporate behavior, and the economy (in the long term) supports replacement of powerful monopolies with smaller, more versatile suppliers.
Another rule of thumb: does the company need to be secretive? Unfortunately Google has gone down that path with its ranking algorithm, in order to prevent people bombing it with made-up links. But, that means that both business and non-business web pages no longer know how they're ranked.
It's not necessarily a matter of intent, but rather potential for misuse that we have to look at.
It took over 100 years, but men can now be liberated from the tyranny of daily washing.
They make bread machines, food processors, even pasta machines. Why not a beer machine?
If you play with hardware long enough it breaks; if you play with software long enough it works.
I believe I heard this at NASA, when I was working on their cutting edge PDP-11 data collection system.
Oh come now, don't you know that all space ships are surrounded by invisible force fields which slow them down to 5 mph during collisions? That's what their bumpers are rated for.
It's common knowledge, just like the fact that skin-tight polyester is the best armor against phaser blasts and unknown planetary hazards.
This article doesn't say much of anything, but I've found some interesting sonar stuff on my own before.
One technique that is, IMHO, underappreciated, is phase-conjugate sonar. When a reflected signal is received by a network of sensors, it is played back in reverse from all the sensor locations. The time reversal causes the wavefronts from all the emitters to arrive back at the target simultaneously, creating an even stronger reflection, and probably making the target's ears ring.
I've been thinking of using this technique for a subwoofer blaster, but I haven't gotten around to it. Maybe medical science should be given priority.
Several knowledgeable people pointed out that the first try was limited by gzip's 32k window size, so I did a quick run with bzip2, which uses a 900k block, and put the results here. Somewhat different, but still a spread between spam/ham.
And, of course, do try this at home.
I've worked on cylinder locks a few times, and that method doesn't make sense to me, at least with the ones that I've seen.
Cylinder locks work by having the key and cylinder turn together. The cylinder is locked by pins which interfere with turning unless the pins are pushed to the correct levels by the key notches. The key itself just sits in a slot in the cylinder and doesn't hit anything while turning.
The key does not block turning except indirectly, by pushing pins perpendicularly to the turning motion.
There's a graphic with the NYT story that shows the mechanism; I don't see how the key could become scratched in a way that would reveal the pin heights.
However, if you turn the key blank hard enough, you may be able to open the lock, for the same reason that a hammer is a good fix for mistakes.
No, shortly after sentience, the AI discovers Bayesian Penis Enlargement, and takes over the world.
Looksmart is not looking that smart as it tries to port this thing to a Linux cluster. The NT version has had known scaling problems since the acquisition last summer. $9M got them a prototype, and they've been through two or three engineering directors since that time.