Github user osma commented on the pull request:
https://github.com/apache/jena/pull/52#issuecomment-100212921
Configuring via the assembler is very good! And I see you have also added
some tests. Excellent!
For the tests, I think you could add a few more test cases (especially
TestDatasetWithLuceneMultilingualTextIndex seems incomplete). For example,
checking that a query for the German word "Gift" (meaning poison) doesn't match
the English word "gift", as it would with the standard analyzer. Also you could
check that at least some of the language-specific stemming rules work with the
multilingual index, just like you have already tested for book/books in
TestDatasetWithLocalizedAnalyzer.
About the implementation of multilingual indexes: You have now made a
separate Lucene index for each language. This is a valid solution, but have you
considered instead using a single index with an extra field "lang" or
"language" that would store the language of the indexed literal? Then at query
time you could simply add an extra parameter to all queries restricting the
lookup to a particular language. That might give simpler code and/or a more
compact index in some cases. The analyzer would still be selected based on the
language, of course.
---
If your project is set up for it, you can reply to this email and have your
reply appear on GitHub as well. If your project does not have this feature
enabled and wishes so, or if the feature is enabled but not working, please
contact infrastructure at [email protected] or file a JIRA ticket
with INFRA.
---
