I'd love to. Sadly, I'm afraid that I'd be sued into the ground by the corporation holding a patent on the method for building a house from mud bricks. So instead, I'm gonna see if that corporation isn't willing to give me a couple of bucks for my brilliant idea. Maybe even hire me. An update to that wonderful book will have to wait.
Unfortunately, the patent is very short on the actual algorithm used to determine clustering and make the inferences. As a result, I can't determine whether the patent on the algorithm would be valid or not. What I can tell though is that as it is worded, the patent describes exactly the work done by the credit card companies. As a result, it is unfortunately irrelevant whether Amazon's work was novel or not - the patent covers topics that had been tackled before.
As for your example - if I take a mud brick, and instead of building a house with it, I build a space elevator, do I get to patent the brick? Think about it.
I'm sorry, but your example is exactly what fraud detection was doing long before 1999. The only difference is that instead of looking for items with high clustering probability, they were looking for items with low clustering probability. In other words, the algorithm could be the exact some one, except that the interesting result was now the high probability, and not the low probability.
Dude - if you're one of the original references, then you should be aware that this is nothing but an inference system, which in turn is what is used by an expert system. And those had been in use for years by medical systems and a locomotive repair software system by GE. Care to explain how your system differed from known AI techniques such as knowledge tree traversal, centroid clustering or any other standard data mining techniques?
From what I'm guessing, there isn't much. Which means that this is a trivial patent on known techniques - aka leeching. This is nothing but more proof that software development will be the province of the rich and lawyer-heavy.
Slashdot Mirror
I'd like you to take a look at the refineries in CA then. Those suckers seem to go down once a year, preferably right when demand is highest.
I'd love to. Sadly, I'm afraid that I'd be sued into the ground by the corporation holding a patent on the method for building a house from mud bricks. So instead, I'm gonna see if that corporation isn't willing to give me a couple of bucks for my brilliant idea. Maybe even hire me. An update to that wonderful book will have to wait.
Unfortunately, the patent is very short on the actual algorithm used to determine clustering and make the inferences. As a result, I can't determine whether the patent on the algorithm would be valid or not. What I can tell though is that as it is worded, the patent describes exactly the work done by the credit card companies. As a result, it is unfortunately irrelevant whether Amazon's work was novel or not - the patent covers topics that had been tackled before. As for your example - if I take a mud brick, and instead of building a house with it, I build a space elevator, do I get to patent the brick? Think about it.
I'm sorry, but your example is exactly what fraud detection was doing long before 1999. The only difference is that instead of looking for items with high clustering probability, they were looking for items with low clustering probability. In other words, the algorithm could be the exact some one, except that the interesting result was now the high probability, and not the low probability.
Dude - if you're one of the original references, then you should be aware that this is nothing but an inference system, which in turn is what is used by an expert system. And those had been in use for years by medical systems and a locomotive repair software system by GE. Care to explain how your system differed from known AI techniques such as knowledge tree traversal, centroid clustering or any other standard data mining techniques?
From what I'm guessing, there isn't much. Which means that this is a trivial patent on known techniques - aka leeching. This is nothing but more proof that software development will be the province of the rich and lawyer-heavy.