UK chemist on Elsevier's ban on textmining
blogs.ch.cam.ac.uk
blogs.ch.cam.ac.uk
I would implement this as a browser plugin that uploads the content to a server (like Google Translate), and let the journals deal with each rogue user individually.
Telling people what software they can use to read text doesn't scale.
Elsevier is willing to play this game of chicken; they are convinced that their customers (i.e., research universities) cannot do without their product. Hence the OP reporting that twice, they instantly and without warning cut off all access to their product to the entire university, because of detected scraping.
Telling people what software they can use to read text doesn't scale.
Unfortunately, I think it does. If individual users, even a large number of them, use a plug-in, they might get away with it-- but this would result in an incomplete data set. To systematically textmine the corpus (which is the task at hand) requires some kind of systematic access to the data, and this is where Elsevier steps in and shuts it down.
But, 35 years is scary, and if I were in his shoes and the government said "pay a $100,000 fine", I'd probably agree without much argument. And that's exactly the point of saying he faces "up to" 35 years.
Second: as it stands, this looks more to be a contractual than copyright matter (Elsevier is contractually banning use of automated text processing by subscribers), though it might possibly attempt to put teeth into this by asserting copyright. The remedy sought by Elsevier would be not to allege infringement, but to cancel future subscriptions. It might be nice for, say, a large collection of institutions to call Elsevier's bluff.
Third: Copyright law as it exists (and particularly in the US where I understand it reasonably well) is very mechanical: it governs the making of copies of an expressive work. Copyright does NOT govern facts, it does not apply to works which are not expressive, it does not apply to works which are functional in nature.
The real test here would be to put this (and possibly other) contract claims to test in a court. Unfortunately, contracts are governed (in the US) under state, not federal law, and while there's some uniformity of language, it would probably take several such cases (and appeals to at least the Federal Circuit) to establish reasonable case law.
Otherwise, what's significant about this to me is that, once again, it's a case far less about the availability and copying of information (journal articles are routinely copied), than it is about power and control within an information market. This is an area in which conventional economics is far too often lacking (though it's also an area in which much interesting work is starting to happen).
the chemistry papers talked about in this article are an easy domain, but similar shredding of more general documents will be possible by 2018.
developing all this takes people's time and other resources, for which money is a proxy. the best way to beat Elsevier is to build something better -- but that something better is going to take substantial funding.
Scientific papers can be published online at a cost well under $35. Add bureaucracy to that, however, and the cost can increase to 100 times. Conventional funding agencies have a limited interest in long-term programs (as opposed to projects) so that organizations find it difficult to afford to maintain digital libraries after they are built. To bring the evil empire down, somebody needs to figure out how to get that $35.
To make more a available and searchable I actually uploaded everything to academia.edu. I hope they don't get sued by Wiley and/or Elsevier for what the service they offer. If anyone wants to check out some chemistry you can check out what I have here http://unlv.academia.edu/AlexiNedeltchev
Few side thoughts: I think there are many things that can be improved in the science publishing: 1. Articles could be more interactive by providing discussion /commenting section. 2. Currently, if you want to see if the article you are checking out is worth reading you have to refer to the overall rating of the journal (this is known was impact factor). I think every article should have separate ratings. That way you can tell which are high impact articles and which are not. Btw the impact factors is calculated based on how many articles referred the article in hand. Does that sound familiar? It's the same concept as a webpage SEO. The more links pointing in the higher the rank. 3. Publishing process is every inefficient. It takes months to get something published since it was to be peer-reviewed. This obsolete approach that begs to be improved. Any ideas?
You should take the time to discuss a bit with your librarian. As I did my PhD in Denmark (DTU), I naively wrote a robot to download the issues of a well known chemical data journal. In about a week of balanced usage, I went to discuss with our librarian. He had seen my usage, was nice not to talk about it, but told me this: I downloaded more than the entire university in a year… and it was not a lot. It means that at that time, they paid a bit less than the $35 per article price.
What is really important to notice is that Elsevier are not selling knowledge for most of the scientific communities but influence. That is, you are published, cited, you get ranking and your university reward you. This is what we need to address if we want to have really open access. We need a better way to “sell” influence to the university researchers and deans.
As I am building Cheméo http://chemeo.com a chemical data search engine, I suffer too. It is maybe time to unit and propose a legal, efficient and rewarding way for the researchers to publish their papers. We can do that on the side and let our influence grow.
Additional notes for HN readers as yeah, we are a bit more on the programming side. What we basically need is a parallel DOI system easy to use, able to load all the open repositories and able to accept "direct" submissions.
We are not going to solve the problem in a year, this is an influence issue, it will take time, years, to really address it, be it by our own work or by "law".
Nothing very complicated, but it needs to be well done to be of any use.
Thanks for the feedback!
If you want a SMILES detector, you can use my opensmiles-ragel grammar to detect if a word is syntactically correct. (Not grammatically correct; it will allow "c1C" unless you write code to require balanced parentheses and matching ring counts.)
This would be much faster than passing it to any of the cheminformatics toolkits to do the first level detection.
If you're doing similarity searches, you might be interested in my chemfp project.
After it's all been mined stop submitting to Elsvier.
1. http://code.google.com/p/google-ajax-apis/issues/detail?id=1...