Hi,

I see that @Field has a default of TermVector.NO.
Now I'm new to search algorithms, so excuse me if I got it wrong... but isn't the TermVector essential for ranking?
Don't we need frequency & positions?
For example:
- Don't we need number of occurrences (so that when searching for "Math", we'd get high rank for the document "math math math", and lower rank for "math is nice")
- Don't we need positions, to tell if words appear close together (when searching for the PhraseQuery "java programmer" with a slop, I'd expect high rank for "java programmer", and low rank for "java is nice, but I'm a Perl programmer").

If it's really necessary for ranking, can I globally configure Hibernate Search to store TermVector for all fields (instead of explicitly requesting it on each field)?

Thanks :)

Hi,

the term vector is only an additional and optional per document inverted index which can be helpful for use cases like match highlighting or "find similar searches". For basic ranking it is not necessary since it is based on the overall index.
Term vectors have a quite special usecase and I recommend you read up on it in the Lucene documentation. Since storing the term vector increases the total index size it is per default disabled and should really only be enabled for the fields for which this additional feature is needed.

--Hardy