Basically correct. However, there is a long-standing convention
within scholarly literature that the author's affiliation at the
time that the work was done is a significant part of the metadata
of a document. This _may_ be due to the limitations of the old
recording media (eg pages in a scholarly journal) but I think
that there is a bit more to it than that.
In our system at AGCRC we are doing the correct thing using an
external authority for the agent information. In order to preserve
the contextual affiliation we record *two* agents for each "report"
- the author and the "sponsor" - ie the author's employer at the
time of publication. This allows us to change the information
about the author in _that_ database without compromising the need
described above.
Alex Satrapa wrote:
>
> Asking a question like "Who was John Smith working for when he wrote the
> article Brown's Cows Eat Meat?" ... leads to a series of discoveries.
> The metadata for the article will tell us who the creator was (that's
> how we found the article, along with its title), who the contributor
> was, and when it was published. From there, we check John Smith's
> employment history - not the document's metadata.
--
Best Simon
|