jackrabbit-users mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Dirk Rudolph <dirk.rudo...@netcentric.biz>
Subject Using Lucene rep:suggest() in oak 1.2.2
Date Thu, 15 Oct 2015 18:17:21 GMT

I started trying some of the advanced query features of Oak (Running 1.2.2 in a AEM 6.1 instance).

First was the suggestion feature. I had a look at the unit test and it seems to be very simple
so I created an index, added some content (in that case using rich text, containing html tags):

text = <p>Lorem ipsum….</p>

Unfortunately the result did not satisfy my expectations. So I only get the suggestion, when
my query starts with the text in the text property:

SELECT [rep:suggest()] FROM … WHERE … SUGGEST(“<p>Lore”) => Suggestions works
SELECT [rep:suggest()] FROM … WHERE … SUGGEST(“ipsu”) => No suggestion

As a developer I would expect that those suggestions work on a per word or phrase base and
not as I was able to figure out on a per line base. So my question is if it is really suitable
to use a CRTokenizer for the Analyser. Wouldn’t it make sense to define a marker interface
users of oak can use to register multiple analysers and having then one suggester per analyser?
Those would then be ordered using service ranking and the first suitable suggester (then one
that returns more then 0 results) would be used for suggestions?

WDYT? So far suggestions are unfortunately not really useful for our costumers requirements.

Dirk Rudolph | Senior Software Engineer

Netcentric AG

M: +41 79 642 37 11
D: +49 174 966 84 34

dirk.rudolph@netcentric.biz <mailto:dirk.rudolph@netcentric.biz> | www.netcentric.biz

  • Unnamed multipart/alternative (inline, None, 0 bytes)
View raw message