Yoav Avitzour
2008-09-14 17:17:01 UTC
Hi,
Attached is the new version of the plugin. This version includes a
significant change as described below.
It turns out that all the previous versions of the plugin did not query
the ISI database for anything other than the title. This, in turn, was
quite fortunate since it also turns out that ISI requires a very strict
format for author query. The new version queries both author and year in
addition to title, and it uses some logic when to convert the current
author list to an acceptable ISI format. It also converts back to the
author name to a more human readable form in the referencer window (with
dots separating the author initials)
This function that converts authors to the restrictive ISI format,
however, is not perfect. It cannot cover all possible ways to write the
author name, so on some cases the query will return no results since the
author list that was generated by the plugin was incorrect. In these
cases the user may either clear the Authors field altogether or make
sure it is formatted in a sensible way. The authors list must be
separated by "and" as is the bibtex rule.
Sensible ways to write author name include:
(1) Firstname Surname and ...
(2) Surname, Firstname and ...
(3) A. B. Surname and ...
(4) Surname, A. B. and ...
(5) Surname, AB and ...
(6) Surname AB and ...
The plugin returns the author lists formated in format (3).
Known exceptions:
- Authors with more than three names (either due to a suffix such as Jr,
Sr, etc. or simply authors with 4 or more names). The ISI query system
does not support authors with more than 3 names (although the database
itself contains entries with such authors, you just cannot query them).
The plugin attempts to work around it, however, if the three initials
are joined together (e.g. "Surname, ABC" as in format (5) or (6)), the
plugin is not aware that these are initials and will treat it as a 3
letter name (format(1) or (2)).
- Author with a surname that has only two letters and uses format (1)
above - the plugin will assume format (6) is used and will not realize
that format (1) was used.
If someone finds more examples where the author query fails please let
me know and I'll try to work around them.
Also, if people feel that using the Author field in the query restricts
the query too much, perhaps I could add a preference to ask which field
to use in the query.
Finally, the ISI database is far from perfect and I wish there were a
better solution. It has errors, and sometimes a query will fail simply
due to a typo in the database (examples I came across are "unshifting"
instead "upshifting," "held" instead of "field," etc.). Also, it
sometimes changes words in the title, for example "Second" becomes
"2nd," "alpha" becomes "a" and so on. Unfortunately, I am not aware of a
better solution at this point, though I'd be happy to hear about it if
anybody knows about one.
Yoav
-------------- next part --------------
A non-text attachment was scrubbed...
Name: isi-plugin.py
Type: text/x-python
Size: 20937 bytes
Desc: not available
URL: <http://icculus.org/pipermail/referencer/attachments/20080914/62626a4f/attachment.py>
Attached is the new version of the plugin. This version includes a
significant change as described below.
It turns out that all the previous versions of the plugin did not query
the ISI database for anything other than the title. This, in turn, was
quite fortunate since it also turns out that ISI requires a very strict
format for author query. The new version queries both author and year in
addition to title, and it uses some logic when to convert the current
author list to an acceptable ISI format. It also converts back to the
author name to a more human readable form in the referencer window (with
dots separating the author initials)
This function that converts authors to the restrictive ISI format,
however, is not perfect. It cannot cover all possible ways to write the
author name, so on some cases the query will return no results since the
author list that was generated by the plugin was incorrect. In these
cases the user may either clear the Authors field altogether or make
sure it is formatted in a sensible way. The authors list must be
separated by "and" as is the bibtex rule.
Sensible ways to write author name include:
(1) Firstname Surname and ...
(2) Surname, Firstname and ...
(3) A. B. Surname and ...
(4) Surname, A. B. and ...
(5) Surname, AB and ...
(6) Surname AB and ...
The plugin returns the author lists formated in format (3).
Known exceptions:
- Authors with more than three names (either due to a suffix such as Jr,
Sr, etc. or simply authors with 4 or more names). The ISI query system
does not support authors with more than 3 names (although the database
itself contains entries with such authors, you just cannot query them).
The plugin attempts to work around it, however, if the three initials
are joined together (e.g. "Surname, ABC" as in format (5) or (6)), the
plugin is not aware that these are initials and will treat it as a 3
letter name (format(1) or (2)).
- Author with a surname that has only two letters and uses format (1)
above - the plugin will assume format (6) is used and will not realize
that format (1) was used.
If someone finds more examples where the author query fails please let
me know and I'll try to work around them.
Also, if people feel that using the Author field in the query restricts
the query too much, perhaps I could add a preference to ask which field
to use in the query.
Finally, the ISI database is far from perfect and I wish there were a
better solution. It has errors, and sometimes a query will fail simply
due to a typo in the database (examples I came across are "unshifting"
instead "upshifting," "held" instead of "field," etc.). Also, it
sometimes changes words in the title, for example "Second" becomes
"2nd," "alpha" becomes "a" and so on. Unfortunately, I am not aware of a
better solution at this point, though I'd be happy to hear about it if
anybody knows about one.
Yoav
-------------- next part --------------
A non-text attachment was scrubbed...
Name: isi-plugin.py
Type: text/x-python
Size: 20937 bytes
Desc: not available
URL: <http://icculus.org/pipermail/referencer/attachments/20080914/62626a4f/attachment.py>