Re: mining and rights

[Phil Davis <pmd8@cornell.edu>]

I've never seen a licensing agreement that states *how* an 
information resource can be used.  Textual analysis is use, 
whether it is performed by someone doing keyword searches, or by 
a machine doing sequence similarly matching.  That said, there 
are some unwritten rules about what constitutes *use* and 
distinguishes it from *abuse*.  Without understanding the intent 
of the user, it is impossible to distinguish systematic 
downloading for the purposes of textual analysis, from systematic 
downloading for the purposes of stealing a publisher's content. 
Security software cannot distinguish the intent of data mining 
from stealing -- they both look like systematic downloading, and 
most publishers are pretty quick to stop this form of use.  The 
Spider Activity Reports from Blackwell are a good example of 
this.

While I think the future is wide open for new tools that enable a 
researcher to perform analysis on large literature collections, 
we may need to distinguish the counting of downloads that emanate 
from data mining software from ordinary human searching and 
browsing.  A single individual using data mining software may 
make COUNTER usage reports essentially incomprehensible to a 
librarian.

--Phil Davis


At 05:20 PM 5/30/2006, you wrote:
> Joe Esposito's inquiry -- I would be very interested to hear comment
> from publishers -- about the licensing issues raised by wanting to
> use large databases of journal articles for data mining connects
> with something in an interview with Cliff Lynch in the May/June
> Educause Review.  Excerpts:
>
>  "We now have about fifty years of investment in text analysis
>  and text mining.  THe intelligence community is still spending
>  heavily on these technologies, and industry is getting very
>  interested for lots of reasons.  For example, I'm told that the
>  pharmaceutical industry is very interested in computational
>  mining of the biomedical literature base.  This is an important
>  part of what is at stake in these massive digitization programs.
>  Are we going to be able simply to read the digitized works, or
>  are we going to be able to compute on them at scale as well?
>  (Presumably, Google will be able to compute on everything it
>  digitizes, even the in-copyright works.  Almost nobody seems to
>  have figured this out yet!  What an amazing and unique resource.
>  It's not clear what the academy broadly will be able to compute
>  on.)  The answer will make a big difference for the future of
>  scholarship.  This move to computation on text corpora is going
>  to have vast implications that we haven't even thought about yet
>  -- implications for copyright, implications for publishers,
>  implications for research groups.  In fact, it may represent the
>  point of ultimate meltdown for copyright as we know it today."
>
> Leaving aside the undoubted substantial potential -- are there any
> indications that mining issues are affecting the way publishers are
> granting or withholding access to material?
>
> Jim O'Donnell
> Georgetown U.