Honestly, almost all full-text search engines are perfectly fine.
Until you get to millions of documents there is minimal difference
in performance and effectiveness.
Categorisation tends to be seen as a good option due to its high
efficiency in locating information with medium numbers of
documents, as illustrated by the chart below:
This makes some pretty basic assumptions about user behaviour and
the crossover points in efficiency will change depending on those
assumptions but the basic idea holds.
However, what this chart does not capture is the cost of applying the categorisation in the first place. The benefit of search is that it is literally "set and forget" - the cost of capturing thousands of additional documents is minimal compared to hand-tagging each. Thus people tend to overvalue categorisation and undervalue search. Applying metadata for the sole purpose of location only makes sense when dealing with a small number of documents with a relatively frequent and valuable amount of reuse.
(Note of caution: Don't forget whole-of-lifecycle governance. If
you're going to end up with a pile of unowned documents in your
system, it can be costly to identify people responsible for their
later review. Ideally documents should be tagged with an
identifiable group owner at point of upload if you need to
delegate their subsequent management or disposal. Usernames are a
poor substitute but better than nothing.)
==================================== Stephen Bounds Executive, Information Management Cordelta E: stephen.bounds@... M: 0401 829 096 ====================================
On 26/05/2021 12:18 am, Jay Kreshel wrote: