Re: [Jabref-users] Contribute a plugin to visualize reference relatinship(citation network) among papers

[ZHANG L. <zl...@ho...>]

Hi Vicent,

I am glad that you like the plug-in: )

By adding graphical modeling capacity to the plugin, the "revise" feature 
can be implemented. But adding the capacity requires much effort which I can 
not afford currently, and the revising function will damage the validity of 
the reference text due to the essential design idea of the plugin, 
sub-string containing test. The plugin can only add new reference entry at 
the end of the reference text rather than correcting the incorrect one 
because it can not locate the incorrect reference entry.

Another solution is let the plugin to first find out the potential 
misspelled reference entry before visualizing and give users an opportunity 
to adjust the input. It also need considerable effort and can not be 
implemented by myself in the coming months.

A current acceptable solution may be directly revising the reference text in 
reference field of the paper. Although the solution requires more user 
effort such as locating the misspelling reference entry in un-well-formed 
reference text, it maintains the validity of the reference text.

I will increase the accuracy and usability of the plugin whenever I get 
time: )

Best regards,
Liang ZHANG

From: Vicent
Sent: Thursday, July 30, 2009 2:42 AM
To: ZHANG Liang
Cc: jab...@li...
Subject: Re: [Jabref-users] Contribute a plugin to visualize reference 
relatinship(citation network) among papers


Liang,

Thanks a lot for your contribution. I think many people were looking for 
something like that within JabRef.

One question/suggestion:

Is it possible to "revise" the reference map/network which is produced by 
the pluggin? I mean, is it possible to "change" (delete or add) the rows in 
the reference picture? It would be useful if someone detects a relationship 
which has not been detected by the plugin (maybe, a misspelling issue, 
etc.).

Thank you again.

--
Vicent





On Wed, Jul 29, 2009 at 20:32, ZHANG Liang <zl...@ho...> wrote:

Hi Cody,

Thanks a lot for your help: ) About the stripping strategy, I have not set
the algorithm to tripping out these characters yet. I am a newcomer of the
academic world and thus I does not have rich knowledge and experience about
the subtitles and even reference written rules, so I do not know how to
process these characters correctly. As I learn more, I will increasingly
refine the plug-in: )

Best regards,
Liang ZHANG


> Hi Liang,
>
> Did you also try stripping out other characters (:;'"().?)? Some
> publishers use different techniques for subtitles.
>
> It would be interesting to do an study of how well your matching
> approach did with large datasets, and whether the error rate was low
> enough to limit erroneous conclusions by network analysts. Often times
> if the names are the same, it is a journal paper that build on a
> conference paper by the same authors, which wouldn't affect the network
> analysis much.
>
> Our graph drawing page should have most of our work that is applicable
> to this:
> http://www.cs.umd.edu/hcil/graphvis/
> My (poor) graph drawing bibliography has other papers that might
> interest you:
> http://www.cs.umd.edu/~cdunne/projs/graphbib.html
>
> Best wishes,
> Cody
>
>
> ZHANG Liang wrote:
>> Hi Cody,
>>
>> Thanks for your compliment: )I am a graduate student majoring in
>> software engineering with some knowledge about graphical modeling, so I
>> have limited knowledge about information visualization: )
>>
>> About matching the reference entry in references text with other paper's
>> title, I just used a simple sub-string containing test approach:
>>
>> First, remove all the blanks and line delimiters in the references text
>> of the master paper to generate a modified references text.
>>
>> Second, remove all the blanks in the referred papers' titles to generate
>> modified papers' titles.
>>
>> Finally, for each paper's modified title, test whether it is contained
>> in the modified references text.
>>
>> The advantage of the approach is that it can correct some typing errors
>> such as more blanks and facilitate text editing effort of users. Since
>> it is impossible for the modified references to generate new char
>> sequence matching paper's title which are not contained in the original
>> references text and the possibility of paper title confliction is low,
>> the approach is acceptable for small set of papers to generate useful
>> information.
>>
>> Your research topic is very interesting, now I am exploring your and
>> your lab's website: )
>>
>> Best regards,
>> Liang ZHANG
>> --------------------------------------------------
>> From: "Cody Dunne" <cd...@cs...>
>> Sent: Wednesday, July 29, 2009 8:33 PM
>> To: "ZHANG Liang" <zl...@ho...>
>> Cc: <jab...@li...>
>> Subject: Re: [Jabref-users] Contribute a plugin to visualize
>> referencerelatinship(citationnetwork) among papers
>>
>>> Hi Zhang,
>>>
>>> ZHANG Liang wrote:
>>>> Hello everyone,
>>>>
>>>> I have just developed a visualization plug-in to help myself
>>>> understanding the reference relationship(citation network) between some
>>>> papers. The project is hosted at
>>>> http://sourceforge.net/projects/jabrefprrvp/ and the plug-in, source
>>>> file, demo etc. can be download at
>>>> http://code.google.com/p/jabrefprrvp/downloads/list since currently the
>>>> sourceforge file manager does not work correctly.
>>>>
>>>> Hope it can benefit you too:)
>>>>
>>>> Best regards,
>>>> Liang ZHANG
>>>
>>> That's quite interesting. How does it match the free text references to
>>> the other entries? Also, have you pursued other visualizations for many
>>> papers like force-directed or semantic substrates?
>>>
>>> I'm working on something similar that combines JabRef with a network
>>> analysis tool and multi-document summarization techniques. I have a
>>> preliminary screenshot here:
>>> http://www.cs.umd.edu/~cdunne/files/iopener_workbench.png
>>> It doesn't do the free text link conversion yours does, though. I'm
>>> using the clean dataset from the ACL Anthology Network for my research:
>>> http://belobog.si.umich.edu/clair/anthology/index.cgi
>>> If you can output the links you generate in a standard tab-delimited
>>> format (http://www.cs.umd.edu/hcil/nvss/netFormat.shtml) you can load it
>>> in lots of different visualization tools like SocialAction or NodeXL
>>> automatically. I'm involved with both if you'd like help with that.
>>>
>>> It's great so see other people working on citation network
>>> visualization!
>>>
>>> Cody
>>>
>>
>

------------------------------------------------------------------------------
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
trial. Simplify your report design, integration and deployment - and focus 
on
what you do best, core application coding. Discover what's new with
Crystal Reports now.  http://p.sf.net/sfu/bobj-july
_______________________________________________
Jabref-users mailing list
Jab...@li...
https://lists.sourceforge.net/lists/listinfo/jabref-users