No, It wouldn't. Chess is a much more complex task than spam filtering.
A baysian filter does not learn sequences. Spam filtering is a binary classification task. You give it message and the task is to decide whether that message is spam or not spam. This is based on a bunch of pre-defined features - usually the words that occur in the text - and the probabilities of those features turning up in spam vs. non spam.
Playing chess is not something that you can naturally represent as a binary classification task. Any approach that attempts to represent playing chess in this way is going to perform poorly.
Drupal is an excellent piece of software. Compared to other CMSs it is fast, modular, has a clean codebase and a gentle learning curve. I recently started using it after messing around with various other CMS systems over the last couple of years.
When I started using drupal I got the same feeling as when I started using Mac OS X. To continue the OS analogy postnuke and phpnuke are more like windows whereas zope and plone are kinda linux of the CMS world.
So the old server was a Pentium Xeon 3Ghz with 1 GB of RAM. This server wasn't even dedicated to drupal - I believe it was shared with approximately 20 other sites.
Now sun have donated a server with dual Opteron and 4G RAM. This alone would probably have been enough to host the drupal site wiht a serious improvement in performance. But they've also purchased 3 Dell 1850s with dual Xeons and 2G RAM.
Given what was serving the site before, do they really need all this horsepower? With the unexpected server donation from Sun, could the money raised have been better spent on something other than more servers?
They might be overkill for a blog, but if you want a python-based CMS you should look at zope and plone. They are both extremely powerful. Unfortunately a lot of people seem to be put off by the steep learning curve and patchy documentation.
The Drupal website has a list of sites running Drupal. The list is dynamically generated by one of the Drupal modules. Any site that enables this module appears almost immediately. Pretty good for improving your search engine ranking.
"The iTunes Music Store is only available in the U.S. To get running all you need is a Mac with Mac OS X (version 10.2.5 or later preferred), and an Internet connection (DSL, Cable or a LAN-based connection recommended for streaming and downloading music). Just download iTunes 4, click the Music Store icon, and you?ve got the world?s most accessible music store, right on your screen. Feel free to browse for as long as you want. There?s no pressure to buy, no annoying pop-up ads, and no confusion about what?s offered."
They allow you to print and save stuff to disk, but not to automatically spider large blocks of content. They have some heuristics running to detect spidering. If this happens, you get locked out of your account until you contact their support and explain yourself.
If people are going to use this archive to automatically induce rules for recognising junk mail (e.g. using naive bayes or ripper), then they will also need at least as many examples of legitimate mail. Of course it could be useful for evaluating classifiers built using smaller corpora.
I work in an university in Ireland and recently a patent law firm visited our department(Computer Science) to 'educate' us about the benefits of patenting any software that we write. From what I remember you can patent software in most European countries, but there is no central european mechanism that can grant you a patent across the entire EU.
Their advice was to apply for a patent in your own country, and more importantly a patent in the U.S. Since most large software companies are U.S. based, and most european companies have customers in the U.S, having a U.S. patent is essential to most european companies.
Amazon Ships to Sorting Machine Beat
By SAUL HANSELL
FERNLEY, Nev., Jan. 18 -- Ever since it built five vast warehouses in 1999, Amazon (news/quote).com has boasted of the wonders of the machinery inside them -- 10 miles of conveyer belts and myriad other gadgets.
What Amazon was not so vocal about was how many people it took to operate those machines, especially during the holiday rush. In 2000, for example, Amazon had to hire 7,200 temporary workers to supplement the 4,400 people working in its warehouses in the United States.
Now, Amazon.com, once the champion the strategy of "get big fast," has learned how to become small. On Dec. 11, its busiest day last year, Amazon's warehouses employed only 4,000 temps and 3,700 full-time employees. With one-third fewer people than the year before, the warehouses processed what analysts estimate were 10 to 15 percent more items.
Amazon, which plans to release its fourth- quarter results on Tuesday morning, needs every dollar it can save. A year ago, the company -- which has lost $2.8 billion since its founding in 1995 -- promised investors it would turn an operating profit in the fourth quarter of 2001 (at least by its own "pro forma" calculation).
That goal was made harder because Amazon's sales grew at only half the rate it predicted at the beginning of the year, dragged down by the recession, the aftermath of Sept. 11 and some of the company's own missteps.
If, as analysts expect, Amazon nonetheless hits its fourth-quarter profit target, a key reason will be the savings from its yearlong campaign to reorganize the people and the machines in its warehouses.
"They are focused on productivity in a very structured way, and it appears they have made good progress," said Anthony Noto, an analyst for Goldman, Sachs. Mr. Noto estimates that order-fulfillment costs absorbed 11 percent of Amazon's sales in the fourth quarter, down from 13.5 percent a year earlier. Still, he said, those costs need to fall below 9 percent for the company to thrive.
Walking amid a forest of bookshelves and climbing metal bridges over the rivers of conveyer belts in the warehouse here 40 miles east of Reno, Jeff Wilke, Amazon's senior vice president for operations, pointed to dozens of improvements big and small.
One big goal had been to reduce errors in keeping track of the several million items continually being placed onto and pulled off of hundreds of thousands of bins on metal shelves. In theory, Amazon's computers know exactly where each item is at any moment. But in 2000, the computers were wrong more than 10 percent of the time, causing delays as workers searched for missing items and restocked spares.
"We had a whole secret plant, not our main focus, putting stuff back," Mr. Wilke said.
To reduce errors, Amazon wrote new software to take better advantage of the gizmo that each warehouse worker was already carrying -- a shoehorn-size device that combines a bar code scanner, a display screen and a two-way data transmitter.
The new software beams far more explicit instructions to workers about where they should go and what they should do. And it checks their work by forcing them to scan each item every time they put it on or take it off a shelf. Errors have fallen to below 5 percent, from 10 percent, Mr. Wilke said.
This new system also helps with another of Mr. Wilke's main goals -- improving the productivity of seasonal temporary workers -- by giving them more direction. It also monitors their performance, so those who cannot get up to speed in a week or so are given help -- then fired, if necessary. Amazon also instituted a formal training program for temps. As a result, the average productivity of each temporary worker has doubled.
Many of Mr. Wilke's efforts reflect the highly quantitative bent expected of an M.I.T.-trained engineer who ran chemical plants for Allied Signal before joining Amazon in 1999. But when he talks about the biggest change here in Fernley, he uses the language of music, not manufacturing.
"We needed to build cadence," Mr. Wilke said, "to operate to the drumbeat of the constraint."
The drumbeating constraint is the $25 million Crisplant sorting machine at the center of Amazon's automated approach. Working with batches of 500 to 2,000 orders, the employees with the hand-held terminals feed items onto a network of conveyor belts into the sorting machine. The machine reads the bar code on each item and routes it into one of 2,100 chutes, each chute representing an order for a single customer. When all the items in an order are in the chute, a light flashes, and a worker rushes to put them in a box. They are then sent on other conveyers to machines that print packing slips, seal the boxes and send them off to shippers' trucks.
Adopting such an expensive and complex machine was controversial for Amazon.
Mr. Wilke acknowledges that he was skeptical of the Crisplant machines when he joined Amazon, inheriting the warehouse designs of his predecessor, Jimmy Wright, a former Wal-Mart (news/quote) executive. In fact, Mr. Wilke arrived in time to delete the machines from the designs for Amazon's warehouses in Europe.
But in the last two years, Mr. Wilke says, he has come to believe that the sorting machines were a good choice. He has also concluded, though, that because they are so expensive and so central to the business, all other parts of the warehouse need to operate with the goal of avoiding backlogs and delays that would prevent the Crisplant machine from running at peak efficiency.
So Mr. Wilke created a new job -- flowmeister -- making one person the orchestra conductor of the warehouse, to keep each section of the operation in rhythm with the sorting machine. In Fernley this day, the flowmeister was Andy Warren, a former logistics consultant who took a career detour as a lawyer. His podium was a metal table topped with seven computer screens that monitor all the key processes of the warehouse.
As Mr. Warren conducted, a graph showed that the people taking items from the chutes and putting them in boxes were not keeping tempo with the ones putting items into the sorting machine. So he had a worker move from the "induction" area to work the chutes, heading off a backlog.
Mr. Wilke's "cadence" talk was hardly music to the ears of the people who worked in the warehouses, because he was essentially insisting they could handle far more volume with no new equipment.
"I felt like Scotty in `Star Trek' saying, `I can't push her any further, captain,' " said Greg Bennett, the manager of the Fernley warehouse. Yet by keeping the Crisplant operating at full speed for two 10-hour shifts a day, the warehouse was able to pack more than 200,000 items on peak days in December, 30 percent more than the year before.
But Mr. Wilke says the pressure is not off. His calculations show that many more incremental improvements could eventually double the productivity of the warehouses.
"The sum of all the little mistakes," he said, "is big."
"In general, the more links there are to a particular page from other pages, the higher it ranks in Google's hierarchy on that search term."
Most people have a vague idea of how google works. This paper goes into some detail.
Dont mean to be pedantic but VM here is virtual memory. The java VM is virtual machine. Quite different. Your average linux user doesn't need to know about the virtual memory system. But if your programming java, you should have an idea how the java VM works.
Slashdot Mirror
A baysian filter does not learn sequences. Spam filtering is a binary classification task. You give it message and the task is to decide whether that message is spam or not spam. This is based on a bunch of pre-defined features - usually the words that occur in the text - and the probabilities of those features turning up in spam vs. non spam.
Playing chess is not something that you can naturally represent as a binary classification task. Any approach that attempts to represent playing chess in this way is going to perform poorly.
To be honest its the first one that has really impressed me. I looked at slashcode, scoop, zope, plone, postnuke, mambo.
When I started using drupal I got the same feeling as when I started using Mac OS X. To continue the OS analogy postnuke and phpnuke are more like windows whereas zope and plone are kinda linux of the CMS world.
Now sun have donated a server with dual Opteron and 4G RAM. This alone would probably have been enough to host the drupal site wiht a serious improvement in performance. But they've also purchased 3 Dell 1850s with dual Xeons and 2G RAM.
Given what was serving the site before, do they really need all this horsepower? With the unexpected server donation from Sun, could the money raised have been better spent on something other than more servers?
They might be overkill for a blog, but if you want a python-based CMS you should look at zope and plone. They are both extremely powerful. Unfortunately a lot of people seem to be put off by the steep learning curve and patchy documentation.
The Drupal website has a list of sites running Drupal. The list is dynamically generated by one of the Drupal modules. Any site that enables this module appears almost immediately. Pretty good for improving your search engine ranking.
The music store is only available in the U.S.
Damn, damn, damn.
"The iTunes Music Store is only available in the U.S. To get running all you need is a Mac with Mac OS X (version 10.2.5 or later preferred), and an Internet connection (DSL, Cable or a LAN-based connection recommended for streaming and downloading music). Just download iTunes 4, click the Music Store icon, and you?ve got the world?s most accessible music store, right on your screen. Feel free to browse for as long as you want. There?s no pressure to buy, no annoying pop-up ads, and no confusion about what?s offered."
They allow you to print and save stuff to disk, but not to automatically spider large blocks of content. They have some heuristics running to detect spidering. If this happens, you get locked out of your account until you contact their support and explain yourself.
If people are going to use this archive to automatically induce rules for recognising junk mail (e.g. using naive bayes or ripper), then they will also need at least as many examples of legitimate mail.
Of course it could be useful for evaluating classifiers built using smaller corpora.
.. between this and nomad v2? Basically it's got a firewire port and lithium ion batteries. Some more details available here.
I work in an university in Ireland and recently a patent law firm visited our department(Computer Science) to 'educate' us about the benefits of patenting any software that we write.
From what I remember you can patent software in most European countries, but there is no central european mechanism that can grant you a patent across the entire EU.
Their advice was to apply for a patent in your own country, and more importantly a patent in the U.S. Since most large software companies are U.S. based, and most european companies have customers in the U.S, having a U.S. patent is essential to most european companies.
Here's why (Deconstructing Katz by Lloyd Wood)
Amazon Ships to Sorting Machine Beat
.com has boasted of the wonders of the machinery inside them -- 10 miles of conveyer belts and myriad other gadgets.
By SAUL HANSELL
FERNLEY, Nev., Jan. 18 -- Ever since it built five vast warehouses in 1999, Amazon (news/quote)
What Amazon was not so vocal about was how many people it took to operate those machines, especially during the holiday rush. In 2000, for example, Amazon had to hire 7,200 temporary workers to supplement the 4,400 people working in its warehouses in the United States.
Now, Amazon.com, once the champion the strategy of "get big fast," has learned how to become small. On Dec. 11, its busiest day last year, Amazon's warehouses employed only 4,000 temps and 3,700 full-time employees. With one-third fewer people than the year before, the warehouses processed what analysts estimate were 10 to 15 percent more items.
Amazon, which plans to release its fourth- quarter results on Tuesday morning, needs every dollar it can save. A year ago, the company -- which has lost $2.8 billion since its founding in 1995 -- promised investors it would turn an operating profit in the fourth quarter of 2001 (at least by its own "pro forma" calculation).
That goal was made harder because Amazon's sales grew at only half the rate it predicted at the beginning of the year, dragged down by the recession, the aftermath of Sept. 11 and some of the company's own missteps.
If, as analysts expect, Amazon nonetheless hits its fourth-quarter profit target, a key reason will be the savings from its yearlong campaign to reorganize the people and the machines in its warehouses.
"They are focused on productivity in a very structured way, and it appears they have made good progress," said Anthony Noto, an analyst for Goldman, Sachs. Mr. Noto estimates that order-fulfillment costs absorbed 11 percent of Amazon's sales in the fourth quarter, down from 13.5 percent a year earlier. Still, he said, those costs need to fall below 9 percent for the company to thrive.
Walking amid a forest of bookshelves and climbing metal bridges over the rivers of conveyer belts in the warehouse here 40 miles east of Reno, Jeff Wilke, Amazon's senior vice president for operations, pointed to dozens of improvements big and small.
One big goal had been to reduce errors in keeping track of the several million items continually being placed onto and pulled off of hundreds of thousands of bins on metal shelves. In theory, Amazon's computers know exactly where each item is at any moment. But in 2000, the computers were wrong more than 10 percent of the time, causing delays as workers searched for missing items and restocked spares.
"We had a whole secret plant, not our main focus, putting stuff back," Mr. Wilke said.
To reduce errors, Amazon wrote new software to take better advantage of the gizmo that each warehouse worker was already carrying -- a shoehorn-size device that combines a bar code scanner, a display screen and a two-way data transmitter.
The new software beams far more explicit instructions to workers about where they should go and what they should do. And it checks their work by forcing them to scan each item every time they put it on or take it off a shelf. Errors have fallen to below 5 percent, from 10 percent, Mr. Wilke said.
This new system also helps with another of Mr. Wilke's main goals -- improving the productivity of seasonal temporary workers -- by giving them more direction. It also monitors their performance, so those who cannot get up to speed in a week or so are given help -- then fired, if necessary. Amazon also instituted a formal training program for temps. As a result, the average productivity of each temporary worker has doubled.
Many of Mr. Wilke's efforts reflect the highly quantitative bent expected of an M.I.T.-trained engineer who ran chemical plants for Allied Signal before joining Amazon in 1999. But when he talks about the biggest change here in Fernley, he uses the language of music, not manufacturing.
"We needed to build cadence," Mr. Wilke said, "to operate to the drumbeat of the constraint."
The drumbeating constraint is the $25 million Crisplant sorting machine at the center of Amazon's automated approach. Working with batches of 500 to 2,000 orders, the employees with the hand-held terminals feed items onto a network of conveyor belts into the sorting machine. The machine reads the bar code on each item and routes it into one of 2,100 chutes, each chute representing an order for a single customer. When all the items in an order are in the chute, a light flashes, and a worker rushes to put them in a box. They are then sent on other conveyers to machines that print packing slips, seal the boxes and send them off to shippers' trucks.
Adopting such an expensive and complex machine was controversial for Amazon.
Mr. Wilke acknowledges that he was skeptical of the Crisplant machines when he joined Amazon, inheriting the warehouse designs of his predecessor, Jimmy Wright, a former Wal-Mart (news/quote) executive. In fact, Mr. Wilke arrived in time to delete the machines from the designs for Amazon's warehouses in Europe.
But in the last two years, Mr. Wilke says, he has come to believe that the sorting machines were a good choice. He has also concluded, though, that because they are so expensive and so central to the business, all other parts of the warehouse need to operate with the goal of avoiding backlogs and delays that would prevent the Crisplant machine from running at peak efficiency.
So Mr. Wilke created a new job -- flowmeister -- making one person the orchestra conductor of the warehouse, to keep each section of the operation in rhythm with the sorting machine. In Fernley this day, the flowmeister was Andy Warren, a former logistics consultant who took a career detour as a lawyer. His podium was a metal table topped with seven computer screens that monitor all the key processes of the warehouse.
As Mr. Warren conducted, a graph showed that the people taking items from the chutes and putting them in boxes were not keeping tempo with the ones putting items into the sorting machine. So he had a worker move from the "induction" area to work the chutes, heading off a backlog.
Mr. Wilke's "cadence" talk was hardly music to the ears of the people who worked in the warehouses, because he was essentially insisting they could handle far more volume with no new equipment.
"I felt like Scotty in `Star Trek' saying, `I can't push her any further, captain,' " said Greg Bennett, the manager of the Fernley warehouse. Yet by keeping the Crisplant operating at full speed for two 10-hour shifts a day, the warehouse was able to pack more than 200,000 items on peak days in December, 30 percent more than the year before.
But Mr. Wilke says the pressure is not off. His calculations show that many more incremental improvements could eventually double the productivity of the warehouses.
"The sum of all the little mistakes," he said, "is big."
"In general, the more links there are to a particular page from other pages, the higher it ranks in Google's hierarchy on that search term."
Most people have a vague idea of how google works. This paper goes into some detail.
Dont mean to be pedantic but VM here is virtual memory. The java VM is virtual machine. Quite different. Your average linux user doesn't need to know about the virtual memory system. But if your programming java, you should have an idea how the java VM works.