Information Technology Reference
In-Depth Information
Search
When journals first found themselves on the web, publishers invested
serious money in their content delivery platforms, pursuing the holy
grail of making their websites 'sticky' - ensuring that visitors stayed on
the site and were encouraged to come back regularly. I worked for a
major physics publisher at that time and the hope was to build a
website that would be the first port of call each day for any physicists
worth their salt. To do this, we built great search functionality, we
invented (I believe) reference linking and we came up with a plethora
of personalized functions. But our readers and Google had other ideas.
Researchers, in the main, just did not want to work that way. They
understandably wanted a one-stop shop, and Google gave them the
facility to search a broad swathe of the literature. Almost overnight the
focus changed. Instead of personalized functions, we were working on
search-engine optimization. Instead of focusing on journal home pages,
we were thinking about article abstract pages, because those were
where users searching on Google were landing. So abstract pages
suddenly contained better navigation and links to news and
announcements and to other search and discovery tools, and we paid
much more attention to branding at the article level; the unit was no
longer the journal issue, let alone the volume - it was the article.
Publishers have to expose the full text of their articles to search
engines for indexing. Not to do so is to miss out on an important
avenue of traffic, although still today I hear of organizations (not
established publishers) trying to build those sticky websites and
refusing to allow Google to index them because they want direct traffic
and not referrals from search engines. In most if not all cases, this is
folly and doomed to failure; it is not what users want.
However, search engines index so much content that, even with
sophisticated search behaviour, finding what you want can be like
looking for the proverbial needle in a haystack, particularly if you only
have incomplete information on what you are looking for. The trouble
with text-based searching is that if you are searching, for example, for
information using the word 'pipe', you will also find something that
says 'this is not a pipe' (with apologies to René Magritte). The search
engine does not understand the context of the words.
