UKCRC Web Site and Strategy Document

[Stevan Harnad]

On Thu, 16 Dec 2004, Rob Pooley wrote:

> > That is rare enough not worry about. We are talking about averages, probability,
> > predictions here, and the predominant reason for citing is positive, not negative.
>
> I rest my case.
>
> Of course the major reason some articles get cited is that the student
> who made the reference list has seen them cited in all previous papers,
> often not having actually read them before including them in the current
> paper.

Among the many helpful algorithms for detecting and correcting this are:

(1) relevance measures
(2) co-citation measures (who cited with/by whom? student or cited master?)
(3) CiteRank measures (are the cited items much cited items? authors)
(4) Self-citation index
(5) Hubs/authorities (citing much-cited or much-citing articles, authors?)
(6) Chronometrics (timeliness, recency)

The way to capitalize on all these possibilities is to analyze them shrewdly, not
to reject them out of hand. (And reject them in favour of
what? re-exercise of individual
judgment in every single case?)

Stevan Harnad

> ("goto considered Harmful" and Plotkin's Structured Operational
> Semantics notes are two very common examples in fields where I read and
> review. Both important, but neither often actually read!)
>
>