<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN">
<html style="direction: ltr;">
<head>
<meta content="text/html;charset=ISO-8859-1" http-equiv="Content-Type">
</head>
<body style="direction: ltr;" bgcolor="#ffffff" text="#000000">
<p style="margin-bottom: 0cm; margin-top: 0pt;"><br>
<br>
Michael Banck wrote:<br>
</p>
<blockquote
cite="mid:20080910162414.GY973@nighthawk.chemicalconnection.dyndns.org"
type="cite">
<pre wrap="">On Wed, Sep 10, 2008 at 11:13:05AM -0500, Yoav Avitzour wrote:
</pre>
<blockquote type="cite">
<pre wrap="">- Added support for titlecase
(<a class="moz-txt-link-freetext" href="http://muffinresearch.co.uk/archives/2008/05/27/titlecasepy-titlecase-in-python/">http://muffinresearch.co.uk/archives/2008/05/27/titlecasepy-titlecase-in-python/</a>)
if present. titlecase does not really resolve the main problem with
capitalization of titles, which is recognizing acronyms that should
remain capitalized.
</pre>
</blockquote>
<pre wrap=""><!---->
Right, that's a hard problem.
Maybe it is possible to see whether there's a title set already, and do
not overwrite it in that case? Cause even with clever titlecasing, the
ISI-provided title might be a regression from a crossref.org one, as
those are usually properly capitalized and have the acronyms correct.
Another possibility could be a general "merge window" popup, where the
user could review the old and new data for the changed field(s), and
reject/approve them manually (maybe having "approve all without further
asking from now on" as an option as well).
</pre>
</blockquote>
The main problem with the first idea is that it completely invalidates
one of what I see as the two main uses of the plugin. Currently, I use
the plugin in two ways:<br>
a) Complement author list on documents for which data was retrieved
from crossref. <br>
b) Retrieve data for documents for which data was not retrieved
automatically, by manually filling up partial title and author
information, just enough such that the ISI query will retrieve the
correct document. <br>
<br>
For use case (b), it is quite a problem to leave the old title in
place, since it may very well be that it is not complete. Also, the ISI
query often fails on hyphenated words and words representing symbols
(such as \alpha). In that case, the user may want to intentionally
cripple the title for the ISI query to find a matching document. So,
the bottom line is that I don't think leaving the original title in
place if it's there is a viable option.<br>
<br>
As for a more elaborate merge window - it's possible, but I'm not sure
it's worth the effort at this point. I'll think about it and consider
it for future versions. Another option I was thinking of is an
expandable private dictionary for each user to fill up through
preferences (with some global acronyms like DNA already there to begin
with). However, all these solutions are awkward attempts to cope with a
silly flaw in the way ISI reports its records. What I really want to do
is to find a way to simply get the correct data from ISI.<br>
<br>
Yoav<br>
<br>
</body>
</html>