Citeology: Visualizing the Relationships between Research Publications
From the post:
Justin Matejka at Autodesk Research has recently released the sophisticated visualization “Citeology: Visualizing Paper Genealogy” [autodeskresearch.com]. The visualization shows the 3,502 unique academic research papers that were published at the CHI and UIST, two of the most renowned human-computer interaction (HCI) conferences, between the years 1982 and 2010.
All the articles are listed by year and sorted with the most cited papers in the middle, whereas the 11,699 citations that connect the articles to one another are represented by curved lines. Selecting a single paper reveals colors the papers from the past that the paper referenced in blue, in addition to the future articles which referenced it, in brownish-red. Titles, The resulting graphs can be explored as a low-rez interactive screen, or as a high-rez, static PDF graph.
Interesting visualization but what does it mean for one paper to cite another?
I was spoiled by the granularity of legal decision indexing, at least for United States decisions, that broke cases down by issues. So that you could separate out a case being cited for a jurisdictional issue versus the same case being cited on a damages issue. I realize it took a large number of very clever editors (now I assume assisted by computers) to create such an index but it made use of the vast archives of legal decisions possible.
I suppose my question is: Why does one paper cite another? To agree with some fact finding or to disagree? If either, which fact(s)? To extend, supprt or correct some technique? If so, which one? For exampe, so I could trace papers that extend the Patricia trie as opposed to those that cite it in passing. It would certainly make research in any number of areas much easier and possibly more effective.