Polymorphism is in no way necessary or sufficient for object-orientation. It's more orthogonal to it, just like static typing or native support for parallelism. There have been many languages around with various degrees of polymorphism that were absolutely not object-oriented. SIMULA didn't have it, and the original C++ didn't have it.
Objective-C has a limited form of polymorphism, and you could make general container classes, but limited to (Obj-C) classes. E.g., if you wanted to write a general sorting function, you could demand that your objects have function for comparing them to other objects. It's not as elegant as C++'s template mechanism (which is basically the only thing I really like about the language, in spite of STL's use of it), but it gives unlimited freedom for tasks that C++ would need to handle in a very, very complex way.
You want an example? Take a look at Cocoa's bindings and Core Data. I've got not idea how you would implement that in C++, except by putting all functionality in one top-level object with tons of virtual functions. Yuck! And then everything (yes, everything) would have to be recompiled, of course, every time you add a feature. Not so in Objective-C.
What Objective-C lacks is overloading, but hey, you give some, you take some. Each language has it's own strength, and you should use the language you are at most comfortable with for a specific task. For me, Objective-C is good for writing UI stuff, C++ for the core administration and computations.
There are other installs you don't see. I recently installed Fedora Core, and it's part of the package. I know that not many people install Linux on their machines, but still. So, basically, all there is is either download or web site visit statistics (and the latter show a very great variability).
You're being way too literal. When some company is granted a patent, it can start asking for money for its application. In a neo-con society, government services will start to charge for all services, even the basic ones. Do I need to mention the state of health services in the USA here? So, if the dark powers of hard corporatism get it their way, the given example might become a reality. No, not with a patent number, but an insurance number of some kind.
"Turn the other cheek" isn't what I had in mind. What I meant with "do not do unto others..." is that you might be harming your fellow web surfers; I am not in the least concerned about the pointy-haired marketing geniuses that try to illuminate us with their 160 by 40 pixels of art and try to make a little money on the side... (spot the sarcasm!).
Is it just me, or are there really recognizable parts of the original film music in the teaser? I cannot recognize the choir bits, but the fanfare is definitely from the first movie (i.e., nr 4). And everything sounds very well done, and somehow I don't think they had an orchestra and choir to their disposal...
The "social contract" approach is a nice way of speaking very philosophically about this, but the real issue is: when everybody blocks ads, most of the web-sites (like Slashdot) will lose income, and most of them will simply stop publishing or start asking money to their readers.
So, if you block ads, you have to ask yourself if you would like to do that.
For the philosophically inclined: look up Kant's ethics, or for the religiously inclined: think of Jesus' "Do not do onto others..." (if that's the correct wording).
The (cognitive) psychologicy and linguistic journals I know have the same system. The evaluation is done by unpaid reviewers, although sometimes one or two members of the board are scientists paid for their troubles. Suppose each of them gets a $50,000 salary. That makes around $150,000 (yes, that's more than two times 50k, but there are more costs involved in hiring somebody). If your journal reaches 150 university libraries, they'll have to contribute $1,000 each, and that's excluding printing, distribution and other editing costs. There might be more universities (if we assume there's one per 1 million inhabitants, there are around 800 of them in Europe and the USA; the rest won't be able to afford these journals...), but unless you've got the ultimate top-journal for every field, not all of them are going to subscribe to your journal. Looking at it in that way, $1000 no longer seems that expensive...
On the other hand, the grand-parent of this post shouldn't forget that Elsevier-Reed (most likely the publisher (s)he's working for) makes quite a profit...
You're pretty stubborn, isn't it? The theorem is true as far as any theorem is known to be true. Many mathematicians have tried to find flaws in the proof, nobody has managed to conjure up a counter-example so far. Just accept it: planar graphs are four-colourable, and any contiguous map is a planar graph.
No, it doesn't. Color Montana red, Idaho blue, Utah red, Colorado blue, Nebraska red, South Dakota blue, North Dakota yellow, and you only need a fourth color for Wyoming. Purple, perhaps...
No, this theorem is pretty valid, and it is intuitively understandable. The only problem is that the only existing proof is rather long...
Also check out the proceedings of conferences like the ANLP (applied natural language processing). I'm in the one of 1992, can't remember the title though, and I would have to search deep in the archives to find it back...
You can also use "Google scholar", e.g. like this: http://scholar.google.com/scholar?hl=en&lr=&safe=o ff&q=%22k+jensen%22+language&btnG=Search That will show you articles by Jensen and articles refering to her work (click on "Cited by"). On the first page, you'll find an early article (1983), but also a reference to the book on PNLP (Penelope, 1993). If you're really interested, I can send you the reference list from my 1994 thesis, but it's obviously outdated (type "Vosse" in Google Scholar, and skip the biochemical stuff, it isn't me; you'll probably also want to skip the psycho-linguistic stuff, as it's not very applicable).
Depending on your interests, you might also want to search for "constraint grammar", which is a completely different way of checking a text, but widely in use in commercial systems (quite a few of MS Office's grammar checkers are built on this idea).
> Mentioning your PhD and references...
It's a pity that it's like that, but I have to agree that a lot of/. readers write authoratively in style without being hindered by knowledge.
> On a side note, what do you think of the neural net approach to help computers understand natural language?
The basic approach "feed the network with raw input and expect it to come up with a good description" doesn't work in natural language. If you check out experiments on recursive networks (RNN, start with Elman), you'll see that the unstructured learning approach is not suited to something with the complexity of natural language. Although some articles are quite positive, they only show work on a very, very small subset of language.
On the other hand, I am currently working on a project in which I'll continue the development of a theory of language processing using a (fixed) neural network, but with a structure that will be specifically devised for the task of combining syntactic and semantic relations. The structure will be fixed, but (in time) the network will be able to learn characteristics of a specific language. But I only started on that recently, so there is nothing to show yet. If you want to know more about it, it's the psycholinguistic stuff that Google Scholar digs up when you search for my name.
The thing that (unstructured) NNs are good at though is statistical association. So in principle, NNs are able to get the same kind of performance as other statistical models, which does capture a lot of detail, but hardly any structure. There are some good books on statistical language processing, which might be a better read than articles on NNs. A book I like is : Foundations of Statistical Natural Language Processing, by Manning and Schutze (1999).
Check the literature, starting with Jensen, around 1990, or Heidorn, PENELOPE.
By the way, if I had written "Hagend, G. 1998. Repair of ungrammatical constructions, Journal of the Assiociation of Computational Linguistics, p. 300-331", would you have checked it, or simply complained? Or should I have written that I have a PhD in the field and in fact built a reasonably successful system myself?
To me it seems that mr Sandeep K is fighting windmills. There have been some good grammar checkers since the 90s and technology certainly has improved since then.
However, there is a pretty good reasons not to build them into MS Office: these programs are slow, rather expensive to maintain, and would generate huge amounts of warnings and errors on an average piece of text (note that the demo texts used by Sandeep are by no means representative). That's not what Microsoft's users want, and MS knows that. They want something that's fast, not intrusive, and helps them out in certain cases.
Although I do applaud any attempt to get better language technology into commercial software, achieving all goals at the same time is simply too costly. Or doesn't an Associate Professor of Marketing and E-Commerce understand what that means?
it was good enough for everything... except reading long pages.
You're not supposed to wear them as glasses. You should stick to your old glasses or contact lenses and use the mouse for navigating the document.
Re:sane error messages when using templates
on
GCC 4.0 Preview
·
· Score: 2, Informative
I only suffer this problem when the line endings in an include file is not "native" to your platform. One include file with LF instead of CR endings is sometimes can be enough.
I don't know precisely about STL, since I avoid relying on it, but you might check all include files. If you're on a unixy platform, try to get the list of all include files (via a "depend" like command) and check 'm all for line endings (the "file" command can be helpful).
That's because Markov chains are memory-less: they don't store "long distance dependencies", only local dependencies. So, depending on the number of states, notes will have a relation to the one, two or three preceding and following notes, but not to notes beyond that window.
BTW, it is possible to store a limited type of long distance dependency in a Markov chain, by partitioning state spaces, but the usual statistical training doesn't give you that kind of representation and even then it won't give you the musical sophistication you're looking for.
What you need are recursive rules, and training them is a lot harder.
The translation depends on the semantic class of the subject (is the subject a potential baker -> use translation nr. 1, is the subject something that's usually being baked -> use translation nr. 2). So, ignoring other issues, this particular problem is easy to solve, it only takes a lot of work.
BTW, the fact that it isn't entirely grammatical has nothing to do with it (if we understand it, we can translate it, so any MT faces the same challenge).
I don't think this is a primer. I remember a magazine (perhaps Keyboard Magazine) that had a disk with software in the 80s. And of course, there was the Dutch radio that broadcasted software over FM...
Philosophy was always considered "queen of the sciences", even in the beginning of the 20th century.
Anyway, your offering is not much better: it consists of a wikipedia reference that's totally free of argument of the validity of post-modern interpretations. As a matter of fact, it literally says "Postmodernism therefore has an obvious distrust toward claims about truth, ethics, or beauty being rooted in anything other than individual perception and group construction." That basically allows for ad hominem arguments, apart from perpetuating the ridiculous idea that there is not one thruth.
Comparing truth with esthetics makes an end to all argumentation (you cannot discuss about taste, as the Germans say) and consequently disables (or deconstructs) itself, not convincing anyone outside the circle of believers.
Anyway, it's worth reading from an academic point of view: to see what crazy ideas other people can come up with, but post-modernism is nothing to take too seriously: it's a toy for philosophers who've lost the competition with other scientists.
I did use patch and rn, and they were simply simple tools waiting to happen. Patch was definitely not the only tool, and rn had many, many, many competitors. Although useful, as a contribution they do not compete with the likes of RUP, Unix, OO or LISP.
"Patch" is not a great contribution to software. Neither are rn, metaconfig, and, IMHO, Perl. And no, Python isn't either. They're just fringe languages and nifty scripts, that happen to be important in your niche of the universe.
Slashdot Mirror
Polymorphism is in no way necessary or sufficient for object-orientation. It's more orthogonal to it, just like static typing or native support for parallelism. There have been many languages around with various degrees of polymorphism that were absolutely not object-oriented. SIMULA didn't have it, and the original C++ didn't have it.
Objective-C has a limited form of polymorphism, and you could make general container classes, but limited to (Obj-C) classes. E.g., if you wanted to write a general sorting function, you could demand that your objects have function for comparing them to other objects. It's not as elegant as C++'s template mechanism (which is basically the only thing I really like about the language, in spite of STL's use of it), but it gives unlimited freedom for tasks that C++ would need to handle in a very, very complex way.
You want an example? Take a look at Cocoa's bindings and Core Data. I've got not idea how you would implement that in C++, except by putting all functionality in one top-level object with tons of virtual functions. Yuck! And then everything (yes, everything) would have to be recompiled, of course, every time you add a feature. Not so in Objective-C.
What Objective-C lacks is overloading, but hey, you give some, you take some. Each language has it's own strength, and you should use the language you are at most comfortable with for a specific task. For me, Objective-C is good for writing UI stuff, C++ for the core administration and computations.
There are other installs you don't see. I recently installed Fedora Core, and it's part of the package. I know that not many people install Linux on their machines, but still. So, basically, all there is is either download or web site visit statistics (and the latter show a very great variability).
You're being way too literal. When some company is granted a patent, it can start asking for money for its application. In a neo-con society, government services will start to charge for all services, even the basic ones. Do I need to mention the state of health services in the USA here? So, if the dark powers of hard corporatism get it their way, the given example might become a reality. No, not with a patent number, but an insurance number of some kind.
Satisfied?
Although it's funny -- and ironic and sarcastic -- this post sadly deserves something more than "funny".
"Turn the other cheek" isn't what I had in mind. What I meant with "do not do unto others ..." is that you might be harming your fellow web surfers; I am not in the least concerned about the pointy-haired marketing geniuses that try to illuminate us with their 160 by 40 pixels of art and try to make a little money on the side... (spot the sarcasm!).
With the display, it's more expensive, and an iMac G5 (ok, not as fast) in this configuration should do around $3,000, perhaps $3,200. Really.
Is it just me, or are there really recognizable parts of the original film music in the teaser? I cannot recognize the choir bits, but the fanfare is definitely from the first movie (i.e., nr 4). And everything sounds very well done, and somehow I don't think they had an orchestra and choir to their disposal...
The "social contract" approach is a nice way of speaking very philosophically about this, but the real issue is: when everybody blocks ads, most of the web-sites (like Slashdot) will lose income, and most of them will simply stop publishing or start asking money to their readers.
So, if you block ads, you have to ask yourself if you would like to do that.
For the philosophically inclined: look up Kant's ethics, or for the religiously inclined: think of Jesus' "Do not do onto others..." (if that's the correct wording).
The (cognitive) psychologicy and linguistic journals I know have the same system. The evaluation is done by unpaid reviewers, although sometimes one or two members of the board are scientists paid for their troubles. Suppose each of them gets a $50,000 salary. That makes around $150,000 (yes, that's more than two times 50k, but there are more costs involved in hiring somebody). If your journal reaches 150 university libraries, they'll have to contribute $1,000 each, and that's excluding printing, distribution and other editing costs. There might be more universities (if we assume there's one per 1 million inhabitants, there are around 800 of them in Europe and the USA; the rest won't be able to afford these journals...), but unless you've got the ultimate top-journal for every field, not all of them are going to subscribe to your journal. Looking at it in that way, $1000 no longer seems that expensive...
On the other hand, the grand-parent of this post shouldn't forget that Elsevier-Reed (most likely the publisher (s)he's working for) makes quite a profit...
You're pretty stubborn, isn't it? The theorem is true as far as any theorem is known to be true. Many mathematicians have tried to find flaws in the proof, nobody has managed to conjure up a counter-example so far. Just accept it: planar graphs are four-colourable, and any contiguous map is a planar graph.
No, it doesn't. Color Montana red, Idaho blue, Utah red, Colorado blue, Nebraska red, South Dakota blue, North Dakota yellow, and you only need a fourth color for Wyoming. Purple, perhaps...
No, this theorem is pretty valid, and it is intuitively understandable. The only problem is that the only existing proof is rather long...
Also check out the proceedings of conferences like the ANLP (applied natural language processing). I'm in the one of 1992, can't remember the title though, and I would have to search deep in the archives to find it back...
o ff&q=%22k+jensen%22+language&btnG=Search
...
/. readers write authoratively in style without being hindered by knowledge.
You can also use "Google scholar", e.g. like this: http://scholar.google.com/scholar?hl=en&lr=&safe=
That will show you articles by Jensen and articles refering to her work (click on "Cited by"). On the first page, you'll find an early article (1983), but also a reference to the book on PNLP (Penelope, 1993). If you're really interested, I can send you the reference list from my 1994 thesis, but it's obviously outdated (type "Vosse" in Google Scholar, and skip the biochemical stuff, it isn't me; you'll probably also want to skip the psycho-linguistic stuff, as it's not very applicable).
Depending on your interests, you might also want to search for "constraint grammar", which is a completely different way of checking a text, but widely in use in commercial systems (quite a few of MS Office's grammar checkers are built on this idea).
> Mentioning your PhD and references
It's a pity that it's like that, but I have to agree that a lot of
> On a side note, what do you think of the neural net approach to help computers understand natural language?
The basic approach "feed the network with raw input and expect it to come up with a good description" doesn't work in natural language. If you check out experiments on recursive networks (RNN, start with Elman), you'll see that the unstructured learning approach is not suited to something with the complexity of natural language. Although some articles are quite positive, they only show work on a very, very small subset of language.
On the other hand, I am currently working on a project in which I'll continue the development of a theory of language processing using a (fixed) neural network, but with a structure that will be specifically devised for the task of combining syntactic and semantic relations. The structure will be fixed, but (in time) the network will be able to learn characteristics of a specific language. But I only started on that recently, so there is nothing to show yet. If you want to know more about it, it's the psycholinguistic stuff that Google Scholar digs up when you search for my name.
The thing that (unstructured) NNs are good at though is statistical association. So in principle, NNs are able to get the same kind of performance as other statistical models, which does capture a lot of detail, but hardly any structure. There are some good books on statistical language processing, which might be a better read than articles on NNs. A book I like is : Foundations of Statistical Natural Language Processing, by Manning and Schutze (1999).
Check the literature, starting with Jensen, around 1990, or Heidorn, PENELOPE.
By the way, if I had written "Hagend, G. 1998. Repair of ungrammatical constructions, Journal of the Assiociation of Computational Linguistics, p. 300-331", would you have checked it, or simply complained? Or should I have written that I have a PhD in the field and in fact built a reasonably successful system myself?
To me it seems that mr Sandeep K is fighting windmills. There have been some good grammar checkers since the 90s and technology certainly has improved since then.
However, there is a pretty good reasons not to build them into MS Office: these programs are slow, rather expensive to maintain, and would generate huge amounts of warnings and errors on an average piece of text (note that the demo texts used by Sandeep are by no means representative). That's not what Microsoft's users want, and MS knows that. They want something that's fast, not intrusive, and helps them out in certain cases.
Although I do applaud any attempt to get better language technology into commercial software, achieving all goals at the same time is simply too costly. Or doesn't an Associate Professor of Marketing and E-Commerce understand what that means?
The fact that it has been done doesn't count?
You're not supposed to wear them as glasses. You should stick to your old glasses or contact lenses and use the mouse for navigating the document.
I only suffer this problem when the line endings in an include file is not "native" to your platform. One include file with LF instead of CR endings is sometimes can be enough.
I don't know precisely about STL, since I avoid relying on it, but you might check all include files. If you're on a unixy platform, try to get the list of all include files (via a "depend" like command) and check 'm all for line endings (the "file" command can be helpful).
Here you can find the contact information, including e-mail addresses, of the Dutch members of the European parliament:
P ag e.jsp?cat=EP&contentCode=MENU_86
http://www.europeesparlement.nl/content/content
That's because Markov chains are memory-less: they don't store "long distance dependencies", only local dependencies. So, depending on the number of states, notes will have a relation to the one, two or three preceding and following notes, but not to notes beyond that window.
BTW, it is possible to store a limited type of long distance dependency in a Markov chain, by partitioning state spaces, but the usual statistical training doesn't give you that kind of representation and even then it won't give you the musical sophistication you're looking for.
What you need are recursive rules, and training them is a lot harder.
The translation depends on the semantic class of the subject (is the subject a potential baker -> use translation nr. 1, is the subject something that's usually being baked -> use translation nr. 2). So, ignoring other issues, this particular problem is easy to solve, it only takes a lot of work.
BTW, the fact that it isn't entirely grammatical has nothing to do with it (if we understand it, we can translate it, so any MT faces the same challenge).
I don't think this is a primer. I remember a magazine (perhaps Keyboard Magazine) that had a disk with software in the 80s. And of course, there was the Dutch radio that broadcasted software over FM...
Philosophy was always considered "queen of the sciences", even in the beginning of the 20th century.
Anyway, your offering is not much better: it consists of a wikipedia reference that's totally free of argument of the validity of post-modern interpretations. As a matter of fact, it literally says "Postmodernism therefore has an obvious distrust toward claims about truth, ethics, or beauty being rooted in anything other than individual perception and group construction." That basically allows for ad hominem arguments, apart from perpetuating the ridiculous idea that there is not one thruth.
Comparing truth with esthetics makes an end to all argumentation (you cannot discuss about taste, as the Germans say) and consequently disables (or deconstructs) itself, not convincing anyone outside the circle of believers.
There you have it.
Shouldn't that be "Habermas"?
Anyway, it's worth reading from an academic point of view: to see what crazy ideas other people can come up with, but post-modernism is nothing to take too seriously: it's a toy for philosophers who've lost the competition with other scientists.
I did use patch and rn, and they were simply simple tools waiting to happen. Patch was definitely not the only tool, and rn had many, many, many competitors. Although useful, as a contribution they do not compete with the likes of RUP, Unix, OO or LISP.
Preposterous.
"Patch" is not a great contribution to software. Neither are rn, metaconfig, and, IMHO, Perl. And no, Python isn't either. They're just fringe languages and nifty scripts, that happen to be important in your niche of the universe.