Does any of you have experience with using the Virtuoso RDF store
for autocompletion, i.e. giving relevant suggestions based on a number
of characters typed in?
I am currently working on such a thing but see a number of problems.
It would be nice if someone has tackled autocompletion in Virtuoso
before and could help.
My current worries:
- Performance and functionality:
* bif:contains:Text indexing seems necessary for performance,
implying use of bif:contains I suppose.
- Issues:
+ Four characters minimum input required before results get
returned seems too much.
(For example, if we are looking for the occurrence of a three
letter word bif:contains would probably not work.)
I would need to find out whether the four character limit
could be changed.
+ autocompletion would need to find keywords containing non-utf8
characters based on their simplified ASCII representation.
I would need to find out whether the indexing mechanism is
accessible
under the hood so as to simplify UTF-8 Literals into ASCII keywords.
+ Preferably I would like to be able to enter whole strings
containing spaces and first look for exact matches (this could be
emulated by bif:contains + a regex filter ).
- Performance:
* Our store is expected to contain millions of triples, I haven't
finetuned performance yet, but am worried whether < 1 second response
can be accomplished.
- Learning curve:
Virtuoso looks great, lots of useful,interesting functionality. But
this comes at the cost of having to learn a lot (Virtuoso/PL,
..sparql extensions, administration, finetuning,..)
Lourens van der Meij
