Slashdot Mirror
Randomly Generated Paper Accepted to Conference
mldqj writes "Some students at MIT wrote a program called SCIgen - An Automatic CS Paper Generator. From their website: SCIgen is a program that generates random Computer Science research papers, including graphs, figures, and citations. What's amazing is that one of their randomly generated paper was accepted to WMSCI 2005. Now they are accepting donation to fund their trip to the conference and give a randomly generated talk."
At the larger conferences they make some attempt at screening out the known crackpots. The amount of effort varies.
non-reviewed papers do not mean that they haven't been read. It means that it hasn't been reviewed. In the case of scientific articles, review means that your peers follow the same process and methods and see if they come up with the same conclusions.
This is my last post.
[6th Estate]
So, this doesn't come close to the sucess of Transgressing the Boundaries: Towards a Transformative Hermeneutics of Quantum Gravity which got into a peer reviewed journal.
_O_
.|< The named which can be named is not the true named
It gets worse... they submitted another paper that was rejected, they asked why, and got this in reply (several paragraphs, complete with random statistics, to say "it's too much work for us to tell you.")
Save Maine's economy: write stuff down. All comments are exclusively my own, not my employer.
One Mark V Shaney, if anyone remembers that Usenet thing.
Free of Flash! Free of Flash!
These junk conferences are organized for no reason other than profit. Accepting everything that is submitted is consistent with their objective.
The deal is, in an effort to get tenure or grants in a publish-or-perish world, mediocre researchers submit to these things. They are published if and only if they pay the registration fee. For this particular conference, the fee is a mere $US 390.
And there are no quantity discounts. If you have n papers you pay n times the fee.
Hmm, it made it to the conference, but it's non-reviewed. So what? The server is /.ed, can't read the correspondence, however, there's little merit for an author to get a paper into a non-refereed publication, I guess.
Alan Sokal did better back then, when the NY-based physicist wrote up an article that got published in a journal (Social Text, IIRC) - journals are supposed to be rather strict in what they accept.
The nice thing here is that they wrote a probably neat NLG (natural language generation) system to write the paper - it seems to be more practical than previous multimodal NLG systems that are much more domain/application-dependent, but generate stuff that makes sense.
Looking forward to that random talk...
I don't know about computer generated random music, but there are a number of composers who have experimented with introducing random elements in their music. John Cage, for one.
I can't speak for how it sounds, I just know that it exists.
Anakin Simpson: If you're not with me, then you're my enemy--ooh, donuts!
For one thing, if you visit the site, the paper that got accepted was accepted as a "non-reviewed" paper.
Even so, before you go off the deep end on this, in my field (which is EE, not CS) it is generally accepted that the conferences are for preliminary results, and the journals are for final results. As a result, conference submissions tend to receive cursory reviews, and journal submissions receive highly rigorous reviews.
At many (but not all) conferences, authors tend to be given the benefit of the doubt, so long as the paper is not obviously ridiculous or plagiarized.
I attended a recent conference at a major university where, rumor had it, 200 papers were accepted and only four were rejected. In spite of this, I found the quality of the conference quite high. You have to go into such things realizing that some crap is going to get through the filter. However, it's nice to hear what everyone is working on, even if the ideas are not completely finished and some of the work might not be going anywhere.
You give the author the benefit of the doubt in a conference submission. The time to be rigorous is at the point of submission to a journal, and in my field, acceptance to a journal is normally crucial to having an idea accepted by the entire community.
Toronto-area transit rider? Rate your ride.
Some psychology papers are based on data collected over decades. Peer review doesn't mean "try this and make sure I was right" as much as it's "make sure my conclusions line up with my methods, the content is relevant in the context of the literature, and so on."
Indeed, even top journals in Psychology will publish papers with mistakes that the reviewers missed. Sometimes it's hard to keep up. I think they give a little slack to the established authors in the field, assuming that easy mistakes won't be made.
My wife, a grad student, discovered a problem in a top researcher's paper. She took it to her advisor and some other professors and they discovered she was right and the author was wrong. Of course, she's scared as hell about writing a criticism of such a well-known and respected researcher.
Smart idea, except that you'd have to pay a copyright fee for each of those "songs".
Why don't you register "millions or billions" of domain names while you're at it?
Mod parent idiotic.
I used to be an organic chemist, and absolutely every paper for refereed journals was reviewed by a third party in the lab to ensure the results could be duplicated. Our lab did a few, and our lab's papers were done by others.
It is expensive and time-consuming. That's why journals like JACS, JOC, Tetrahedron, etc. are respected so widely: the research in them is rock-solid and proven to work.
p
In Korea, long hair is for old people!
Actually, I doubt that the graduate students presenting this talk would care to work with anybody that is actually attending SCI! I think people missed the fact that this is to point out the fallacy of for-profit conferences like SCI.
Those who use "copywritten" to mean "subject to copyright" tend to look like they haven't studied much of copyright law. The adjective is "copyrighted".
What you said is true, that copyright exists from the moment a work is fixed in a tangible medium, but in the United States. But you can't sue until you've registered the copyright in the Copyright Office, and you can't recover statutory damages or attorney's fees for infringements more than three months after first publication unless you registered the copyright before the infringement occurred. In addition, "intellectual property tax" legislation is under consideration that may make the copyright expire sooner if it isn't registered with taxing authorities.
You would need quite a few. Just the combination of the first 8 notes is 26^7=8,031,810,176, assuming the first note's placement is irrelevant, and assuming up to an octave's jump in value either way. That is discounting rythmic variations, which would add quite a few extra combos.
Remember that not all the melodies on an album have to match for there to be grounds for a lawsuit. If just one of the two or three melodies in just one of the 10 or 12 songs in just one of the thousands of albums released annually matches your work, then you've got yourself a case.
Further analysis of this issue is in yerricde's journal. It seems to disregard accidentals (notes not in a given diatonic mode) but takes rhythm into account.
to win a copyright case, you have to prove *copying*. If someone else independently came up with the same tune as you, you'd be unlikely to win unless you could prove they had access to your musical work
If you've heard a musical work even once in a grocery store or on the car radio ten years ago, you are deemed to have had access to the work. And once the plaintiff demonstrates evidence of access and similarity, the judge is likely to rule that copying occurred. See Bright Tunes Music v. Harrisongs Music .
Given all quarter notes or what? Because for every melody, you can create a new one by splitting any given note into two different notes, the sum of whose durations is equal to the original. And you could split it into 3 notes, or 4 notes ... and each of those into briefer notes, etc.
There are three parts to a musical misappropriation case: defendant's access to the plaintiff's work, probative similarity, and substantial similarity. Lack of intent is no defense. Access and probative similarity are circumstantial evidence of whether copying occurred; substantial similarity determines whether the copying is actionable infringement. Access can often be assumed if a work has been in rotation on commercial FM radio. Probative similarity involves testimony of an expert witness, but substantial similarity refers to the impression on somebody with less musical training. For instance, laymen tend to simplify the model of rhythm down to just (say) short, medium, and long notes within a work. Unfortunately, even if such simplifications of the musical model don't make it possible to enumerate all possible melodies, they make it possible to enumerate enough melodies to make music publishing a legal minefield for people outside the cartel, comparable to the software patent situation.
To say nothing of stuff like fermatas, key signitures, etc.
Which are completely ignored in the substantial similarity phase.
The article was accepted by one Nagib Callaos. Odd name ... who is this guy?
Looked him up on Google to find that he mainly organizes academic computer conferences. The only research papers that I find from him are at conferences that he's organized.
Odd.
These "random" elements which John Cage used in much of his music are a far cry from the "randomness" that would be generated from a computer program using algorithms to calculate random instances of pitch, duration, tempo, velocity, etc.
The latter would probably end up looking and sounding, ironically, nearly identical to music composed using serialism, set theory, 12-tone music, etc. in which all 12 notes of the chromatic scale are arranged into a "row", which can then be used in retrograde, inversion, rotation, transposition, among others, all at the compsoer's discretion. The music of Schoenberg, Webern, Berg, and other serialists tend to be more respected among mathematicians these days.
John Cage's "randomness" stems from his intense studies of Eastern Religions, especially Zen Buddhism. For a large portion of his life, much of his music was derived, at least in part, from quasi-random decisions determined in the I Ching (The Chinese Book of Changes). Much has been written by and about John Cage on using random (aleatoric, as we musicians refer to it) elements, and of his philosophies on music in general
To give you an example of his aleatoric compositions:
4'33 - in 3 movements, the performer is instructed to sit silently at the keyboard for 4 minutes and 33 seconds, closing and opening the lid between each movement. the interpretations are too many to list here.
Imaginary Landscape No.4 - the score calls for the prescribed manipulations of knobs on 12 radios. The aural result is dependent on what happens to be on the airwaves at the instant of performance.
Other works have been "composed" by filling in notes, articulations, etc. wherever tiny imperfections appear on a sheet of manuscript paper.
You picked a bad example. Special Relativity was "in the air" in 1905, and if Einstein had decided to take a vacation, any of a half-dozen other Physicists would have published SR within a year or two. Lorentz is a prime example, and the heart of SR is still known as "Lorentz symmetry". FitzGerald probably would have beaten Einstein to the punch except he had the misfortune to die first. Effectively, the combination of Maxwell's Equations of electromagnetic waves plus the result of the Michaelson-Morley experiment showing that the speed of light was invariant made SR inevitable.
Now if you'd said GENERAL Relativity, then I'd have agreed with you.
-- John Dierdorf, Austin TX
Well, I guess that this only demostrates what is already known: academic spam exists. WMSCI is a spam conference organized by a ghost organisation called IIISCI, just as those organized by WSEAS. (And by this I don't mean that some "reals" papers don't sound as random generated to me also). Read more about academic spam at: http://www.humanities.mcmaster.ca/~sgs/index.php?p =48
Actually, if you read WMSCI's mission, it looks randomly generated too:
So CS might have problems, but you cannot argue that based on WMSCI.
unless you can prove that someone derived their work from yours, you have no damages. Copyright has an originality requirement, not a novelty requirement.
True, copying a work into another work requires both access to the work and substantial similarity between the two works, but having heard a song once on the radio or on a grocery store's background music is enough to count as "access" to the song under copyright law. George Harrison got in trouble for this; in Bright Tunes Music v. Harrisongs Music he was ordered to pay over a million U.S. dollars in damages to Bright Tunes Music because he had subconsciously copied "He's So Fine" written by Ronnie Mack into his own "My Sweet Lord".
You and I can each have a copyright on the same thing provided we each came up with it independently.
Given the pervasiveness of commercial radio, how is "independently" possible anymore?
A year and a half ago, someone I knew received an invitation to speak at a conference. He felt honoured and accepted, but became somewhat suspicious at the requested "paper submission fee". He and I started investigating the conference, searched the net for previous years' editions and comments on them, checked the organisation etc. It all turned out to be a scam, pretty unique in its method.
Three or four conferences on distinct technical subjects and apparently unrelated to each-other were being organised at the same time in the same hotel in Miami. All conferences had pompus websites made from the same template and all were served from the same IP address without a reverse
record in Venezuela. All the websites of previous years' conferences were gone. Some conferences gave the same Florida phone number to the secretariat and others gave no phone number at all. The Florida number in question was forwarded to Venezuela.
Common to all these conferences was that they were headed by professor Nagib Callaos, the same one who accepted the SCIgen paper. I searched the net for his credentials; I found none. I phoned his office in Venezuela and asked for them; I was met first with polite evasions and then with hostile evasions. One of his conferences stated boldly that it was organised "under the auspices of the University of Texas in Austin". I checked with the university; the university had never heard of the conference, nor of "professor" Callaos. Shortly after my phone calls to UT, the website of the conference "under the auspices of UT" disappeared, although the conference itself was still ahead in time.
The catch is the submission fees, 250-600 dollars per accepted paper, allegedly to cover the costs of publishing the papers in book form. Presumably nobody ever attends these conferences except the speakers themselves. If the SCIgen gibberish paper is actually read at WMSCI 2005, it will serve the rest of the speakers as a reminder that greed for recognition works just as well as greed for money in the 419-world.