Slashdot Mirror
Researchers Pull Out of Talks With Publishers On Text-Mining
ananyo writes "Disagreement between scientists and publishers has grown on a thorny issue: how to make it easier for computer programs to extract facts and data from online research papers. On 22 May, researchers, librarians and others pulled out of European Commission talks on how to encourage the techniques, known as text mining and data mining. The withdrawal has effectively ended the contentious discussions, although a formal abandonment can be decided only after a commission review in July. Scientists have chafed for years at limitations on computer-aided research. They would like to use computer programs to crawl over thousands or millions of articles and other online research content, extracting data to build up databases or to pick out patterns such as associations between genes and diseases. But in many parts of the world, including Europe (though perhaps not in the U.S. — the situation is unclear), this sort of use currently requires permission from the content's copyright owner. Even if an institution has paid to access a journal, its academics do not necessarily have permission to mine the text."
And you have failed to notice that the researchers are all for this, which is why they pulled out when publishers weren't cooperating.
The researchers are protesting that they are not being allowed to mine the content that they have already paid to get. They are not arguing that the content should be available for free.
The people who do the science and write the papers produce the content. Yet somehow the publisher controls how it gets used thereafter.
Everyone is so damned beholden to copyright that it more or less constrains how you do anything.
And they wonder why people are pushing for open access -- it's time to cut the buggy whip makers out of the equation.
If you took public money to do this, it should be open. If you want it to be locked down and proprietary, don't publish.
Lost at C:>. Found at C.
Translation: Invent the wheel many many times! Don't you DARE share the data on wheels with others without first getting permission to replicate data from the spoke makers, and rim makers!
Fuck off AC. Look at the internet as a model on how unfettered data proliferation prevents biases from dominating information use. (What's that barbara striesand? That pictue of your beach house is STILL on the internet? Fancy that!) Allowing researchers to share and vet each of these databases you want them to all make independently is EXACTLY how this technology should be used, BECAUSE it prevents usedful data from being hushed up, or forgotten, and gives that data its due. The scientists that created the data want the data shared. The scientists that ewant the data, want it shared.
The only group that does NOT want the data shared, is the publishing industry, because if the data leaves their grimy little fingers, they can't charge rent.
That's the real issue here.
I think you've got the wrong end of the stick.
It's the researchers who want the text-mining capability.
The publishers, who add no value whatsoever, are standing in the way.
Name a journal that has paid a researcher to publish a paper. I'll tell you, there isn't one, researchers have to pay a "submission fee" to have their paper even considered if accepted copyright is often deferred to the journal, then they have to subscribe to the journal to read it. Infact the only thing the actual publisher pays for in this whole mess is the paper and ink to print the thing. I'm going to guess this is just another nail in the coffin of traditional academic journals as the researchers start taking more of their papers elsewhere for publishing.
Knowledge mining does not extract content. Knowledge mining extracts facts. Text mining analyzes and classifies documents, clusters them into groups, and tries to support further knowledge mining. About the only activity that I can think of that could qualify as potentially violating copyright is summarizing. Content may be protected by copyright laws, but the facts can't, and your comment isn't therefore very relevant. I really wonder how the publishers argue that copyright applies to this. Where I live (.cz), copyright not only explicitly applies to making copies; in fact, I believe there's a clause explicitly allowing using "using the work in scientific research", if the use of parts of the text "doesn't exceed the extent necessary for the intended goal". One could argue that mining facts, entities, and their relationships qualifies under this, and that it expressly doesn't qualify as making copies of a work perceived as an authored text.
Ezekiel 23:20
Give the researchers a few years with the current trends, when it becomes clearer that if nobody associated with their work is getting paid for it, they won't be either.
The researchers are paid with grants, they're not paid directly through publishing. If I publish a paper in Nature, it gets included in text mining, and people cite it from the text mine, that benefits me EVEN if no one ever actually reads the paper. If zero people pay for access to my article, that doesn't matter to me. If a billion people pay $30 to see my article, that doesn't matter to me. It matters only to the publisher.
And data mining can't replace most researchers doing benchwork. Barring AI, data mining is not going to come up with brilliant theories or insights, and barring robots, data mining is not going to do benchwork.
Publishers have a lot to fear from this, not researchers.
I guess the GP is proposing that the institutions should pay further sums of money to the publishers to buy this new privilege.
In my view the researchers should man the fuck up and start data mining without permission. I'd love to see the publishers trying to sue, say, Cancer Research UK. The public at large would begin to understand the rottenness at the heart of big copyright.
Pulling out is not an effective method of prevention.
Google doesn't place ads on Scholar results. Publishers do not provide financial incentives to scientific authors. This isn't about any particular company acquiring free rights to scientific texts. This is about computer science and bioinformatics researchers in the text mining area acquiring rights to texts which they themselves (or the scientific community in general) have created in the past. Publishers do not create scientific texts.
An example of this type of text mining work is the detection of protein-protein interactions by looking for papers in which two proteins are mentioned together, perhaps near the words "binding" or "interacting." Of course, there are many more sophisticated strategies as well.
But your argument is that no one should be allowed to benefit from science if they plan on profiting from science, which is somewhat odd.
.: Semper Absurda
It's definitely another point against subscription-fee journals (the "traditional" label is ambiguous; are open-access journals with standard review structures "traditional?").
That aside, I want to clarify: subscription-fee journals do not charge a submission fee, although they often charge for extra pages or color figures. The price then is signing over copyright on your work. In contrast, the open-access journals do charge a (quite large) submission fee. PLoS One for example charges $1,350. The OA journal then lets you keep copyright while licensing under Creative Commons or similar.
.: Semper Absurda
Grants will cease to exist when all value accrues to the companies mining and distributing the data by the most efficient means possible, such as Google... If X is a university or government agency providing funding or a grant, the economic process remains the same. Ultimately there is no reason for any value to accrue to any other entity, if the answer for Big Data Corp. is always "wait for somebody to provide content for free, or mine it, and slap an ad on it".
That doesn't make any sense. The government is not looking for a direct return on the grant, particularly not in the form of publication income. The incentive for the government agency to provide the grants is still there: to advance science. If it ends up open access, published by elsevier or others, or packaged into google's data, the research is still being done.
That's my point, as expressed. Frankly your evaluation of its correspondence with this particular case is of marginal interest to me. :p
So you're just interested in spouting general economic principles and ignoring whether or not they apply to the topic of conversation? Because in this case, it doesn't. What you're suggesting is nonsense, but you don't care to hear that it's nonsense? Well then you'll never learn.
You are ignorant of how scientific publishing works. The publishers are the free loaders. Scientists did the research, wrote the papers, edited and peer reviewed them on a volunteer basis and, indeed, typeset the final print versions.
The large scientific publishers are parasites who abuse their oligopoly powers to extract rents on the labor of the scientist.
.: Semper Absurda
As an NLP Bioinformatics guy, I believe the real crime Aaron Swartz committed was being in the news.
He isn't the first to have that dataset and he wont be the last.
We write papers using massive NLP scans of publications rather routinely.
Most of the time, the papers are downloaded from PubMed (public funded) so they can't even complain about bandwidth costs, etc.
For anyone who didn't know already, most subscription Publishers don't **DO** anything.
They are only slightly better than patent trolls, and in some cases, worse.
In order to have something "distributed" you have to sign a "license."
.: Semper Absurda
I don't think it's exactly feasible to download millions of articles by hand. You can scrape, like Aaron Schwartz did, but that doesn't seem like a great idea these days.
I agree with the sentiment, but we're talking about vast databases held internally by these publishers, not available information encumbered with a little fine print.
.: Semper Absurda
academic culture and the academic generation gap.
Hiring and tenure still involve large percentages of faculty that "came up" under the old system, and don't see the problem (don't have time to see the problem) that has emerged in academic publishing culture over the last couple of decades in particular. They don't see work published outside of the big name journals/publishers as "serious" or "academic" for the moment. So young academics wanting to build a career continue to support them and publish in them, as a pragmatic career-building move.
But young academics by and large (at least in my wing of the social sciences) are incredibly jaded about academic publishing and are absolutely willing to shift the culture away from publishing with big journal mills—they just have to get hired, get tenure, and become "the academics of the world" first. Then, as they begin to be the ones making the hiring and tenure decisions, you can bet that as they consider the next crop of youngsters, they won't place the same premium on Springer, Elsevier, et. al. journals.
The publishing mills are not long for the world, and they know it, which is why they're all trying to expand/reshape their product lines, business models, etc. away from straight print content licensing and toward academic SaaS and other similar offerings.
STOP . AMERICA . NOW
One of the concerns (read: lame excuses) given by the publisher side of this is fear that large scale downloads will cripple their web servers. Private torrent trackers for scientific work is the obvious solution. With university and institutional seeds, this solution would be efficient, equitable and fast.
.: Semper Absurda
The idea that the government need not return value from investment, per the perceptions of the public based on their expectations of private industry, is fiction.
Which is why I never suggested anything of the sort. To quote myself "The incentive for the government agency to provide the grants is still there: to advance science."
The government still gets their return on investment, even if google mines the data: the science is done. Whether the results appear only in Elsevier or whether they're included in a larger data set, the government and scientists get value out of the grant process. Only Elsevier and other publishers are hurt, which is why they're the only ones opposed to it.
Come on. The description of research methods , procedures, tests and results scientific papers, exists for the betterment of humankind, not to make people who own it rich. Get rich by Making Stuff, not exerting a monopolist's control on Knowledge.
How hard is this? All research and results conducted by higher ed should be available for free and the costs rolled into the tax base.
This is as basic as it gets. Roads bridges security and advances in knowledge.
That doesn't make any sense, because you're still refusing to give any examples.
The NIH has an interest in finding a cure for cancer. They give money to researchers to add chemicals to cells and look at them under the microscope. Google repackaging published papers is not going to compete with that.
You have to enter into a contract to be given access, the terms of that contract certainly prohibit downloading the entire database. Just ask Aaron Swartz if he understands that yet.
It's my understanding that for me as a physical person, the law of my country voids any such contract clauses in the copyright law explicitly - meaning that doing so *could* be construed as violating terms of contract but it would never pass as copyright infringement (and yes, there's been a ruling of our Supreme Court on that, so there's hardly any wiggle room here).
Ezekiel 23:20
The NIH has an interest in finding a cure for cancer. They have a budget of $20 million to spend. Company A hires lots and lots of top-notch researchers to do the work for their proposal, spends $5 million on them, still turns a profit on their bid after paying all these researchers. Company B hires one research guy as a figurehead, waits 2 years for the exact same research to be done in a unrelated biology research project in academia, pays $100 for the data-mined results, pulls in $9 million profit direct to the shareholder's and executives bottom line. Out of generosity they kick back $1 million on PAC money to the politicians supportive of the clear "health care research efficiency benefits" they've provided and who also conveniently control future funding to the NIH. Next year, the grant is going down to $5 million, Company A is going out of business, or both.
Two things, one, grants from the NIH are judged by scientists in the field. Currently, the process can distinguish between company A doing actual research and company B rehashing research quite well.
Two, Company B can already get company A's results. They publish the results. Company B could already steal A's results and rehash them. Yet we don't see the problem you suggest.
For further examples, just look most anywhere in the U.S. The value science provides becomes increasingly accumulated by a narrower and narrower subset of the population, leaving a growing expanse of the unemployable whose skills aren't sufficiently differentiable, at least to CEO's, than hiring a new graduate at bottom dollar and plugging them into Google.
You're still making a general statement and insisting it applies to this specific area when it clearly doesn't. Data mining does not make lab work redundant. Neither the NIH nor any CEO worth considering will say "Hey, why pay a million dollars to screen novel drugs that can fight cancer when we could spend five dollars to do a google search for it." Because that's nonsense.