You are viewing a plain text version of this content. The canonical link for it is here.
Posted to solr-user@lucene.apache.org by ppunet <pp...@gmail.com> on 2019/06/19 18:58:55 UTC

How to have Suggester to return part of the content instead of the entire content in Solr version 7.6.0?

Here is my problem statement and I would really appreciate for your feedback.

Solr version 7.6.0

1. There are 1000's of pdf's with large amount of content are indexed to
Solr.
2. Using AnalyzingInfixSuggester for the suggestions.

Q. As the SuggeterComponent provides the 'entire content' of the field in
the suggestions. How is it possible to have Suggester to return only part of
the content of the field, instead of the entire content, which in my
scenario quite long?


Thanks in advance.

PD



--
Sent from: http://lucene.472066.n3.nabble.com/Solr-User-f472068.html

Re: How to have Suggester to return part of the content instead of the entire content in Solr version 7.6.0?

Posted by ppunet <pp...@gmail.com>.
What would you recommend to use for the search field autocomplete
functionality?
Considering I have the whole pdf content, and the user can start searching
for any term from the pdf.



--
Sent from: http://lucene.472066.n3.nabble.com/Solr-User-f472068.html

Re: How to have Suggester to return part of the content instead of the entire content in Solr version 7.6.0?

Posted by Erick Erickson <er...@gmail.com>.
you're mis-using suggester. It’s purpose is to return the entire contents of the “document”, to handle, specifically, multi-word suggestions, typically just a few words, often 2-4. Putting a large text field instead is outside the design.

Also remember that the suggester is looking for the _indexed_ values to return best match. So stemming etc. are in the part that matches, and the stored portion (which is a big, unanalyzed bunch of text) is all that’s available to be returned.

How would suggester know which parts of a large text field to return?

Perhaps you can extract “important” short phrases from the text and put those in the suggester as separate documents, but I admit that’s very hand-wavy. This is where Named Entity Recognition etc. are sometimes used, but that’s not something you get OOB, usually it’s done during the ETL process.

You really have to re-think your expectations about what OOB suggester is good for.

Best,
Erick

> On Jun 19, 2019, at 11:58 AM, ppunet <pp...@gmail.com> wrote:
> 
> Here is my problem statement and I would really appreciate for your feedback.
> 
> Solr version 7.6.0
> 
> 1. There are 1000's of pdf's with large amount of content are indexed to
> Solr.
> 2. Using AnalyzingInfixSuggester for the suggestions.
> 
> Q. As the SuggeterComponent provides the 'entire content' of the field in
> the suggestions. How is it possible to have Suggester to return only part of
> the content of the field, instead of the entire content, which in my
> scenario quite long?
> 
> 
> Thanks in advance.
> 
> PD
> 
> 
> 
> --
> Sent from: http://lucene.472066.n3.nabble.com/Solr-User-f472068.html