Re: Solr Highlighting Question

2010-09-24 Thread Koji Sekiguchi

 (10/09/25 8:07), Jed Glazner wrote:

Hi Koji,

I'm trying to get the FVH to work per your suggestion, but I think I must have
something misconfigured...

Here is the field def in my schema.xml:

fieldType name=nGram class=solr.TextField
analyzer
tokenizer class=solr.LowerCaseTokenizerFactory/
filter class=solr.EdgeNGramFilterFactory minGramSize=2 maxGramSize=15
side=front/
/analyzer
/fieldType
field name=grams type=nGram indexed=true stored=false
multiValued=false termVectors=true termPositions=true termOffsets=true/

Then here is my request handler in solrconfig.xml:

requestHandler name=beyond class=solr.SearchHandler default=true
lst name=defaults
str name=defTypeedismax/str
str name=qfname_title^3 plain^6 grams^1 soundsLike^1/str
!-- automatic phrase! --
!-- uncomment this to enable phrase query. it's disabled now because i saw a
thread that said FVH didn't work with MPQ --
!--str name=pfname_title^6 plain^12/str--
str name=tie0.1/str
/lst
/requestHandler

But when I try the following query:

http://192.168.1.75:8983/solr/music/select?q=beyoqt=beyondhl=truefl.hl=gramsf.grams.hl.useFastVectorHighlighter=true

There are hits, and results, but the highligher section is empty:
lst name=highlighting
lst name=artist.artist.241/
lst name=artist.artist.1616/
lst name=track.track.7022/
lst name=album.album.1775/
lst name=album.album.3128/
lst name=album.album.4590/
lst name=album.album.6951/
lst name=album.album.9626/
lst name=artist.artist.10498/
lst name=artist.artist.29631/
/lst

Any Ideas?
On 09/08/2010 07:54 PM, Koji Sekiguchi wrote:
 (10/09/09 2:26), Jed Glazner wrote:

  Thanks for taking time to read through this.  I'm using a checkout from

  the solr 3.x branch

  My problem is with the highlighter and wildcards

  I can get the highlighter to work with wild cards just fine, the problem
  is that  solr is returning the term matched, when what I want it to do
  is highlight the chars in the term that were matched.


  Example:

  
http://192.168.1.75:8983/solr/music/select?indent=onq=name_title:wel*qt=beyondhl=truehl.fl=name_titlef.name_title.hl.usePhraseHighlighter=truef.name_title.hl.highlightMultiTerm=true


  The results that come back look like this:

  emWelcome/em to the Jungle

  What I want them to look like is this:
  emWel/emcome to the Jungle

  From what I gathered by searching the archives is that solr 1.1 used to
  do this... Is there a way to get that functionality?

  Thanks!



  Try to use FastVectorHighlighter on n-gram field for highlighting problem...
  But FVH cannot process wildcard query. So you should query wel instead of
  wel*. Then this makes you got unwanted hit like voemwel/em.
  I don't think there is a solution for both of them with OOTB today.

  There is a JIRA issue, but no patches there:

  https://issues.apache.org/jira/browse/SOLR-1926

  Koji




Hi Jed,

Unfortunately, FVH doesn't support variable-length-gram tokens.
For example, this should work:

tokenizer class=solr.NGramTokenizerFactory minGramSize=2 maxGramSize=2/

But this may not work:

tokenizer class=solr.NGramTokenizerFactory minGramSize=2 maxGramSize=4/

Can you try constant-length-gram tokens?

Thanks,

Koji

--
http://www.rondhuit.com/en/



Solr Highlighting Question

2010-09-08 Thread Jed Glazner

Thanks for taking time to read through this.  I'm using a checkout from

the solr 3.x branch

My problem is with the highlighter and wildcards

I can get the highlighter to work with wild cards just fine, the problem
is that  solr is returning the term matched, when what I want it to do
is highlight the chars in the term that were matched.


Example:

http://192.168.1.75:8983/solr/music/select?indent=onq=name_title:wel*qt=beyondhl=truehl.fl=name_titlef.name_title.hl.usePhraseHighlighter=truef.name_title.hl.highlightMultiTerm=true

The results that come back look like this:

emWelcome/em   to the Jungle

What I want them to look like is this:
emWel/emcome to the Jungle

  From what I gathered by searching the archives is that solr 1.1 used to
do this... Is there a way to get that functionality?

Thanks!



Re: Solr Highlighting Question

2010-09-08 Thread Jed Glazner




Anybody?

On 09/08/2010 11:26 AM, Jed Glazner wrote:

  Thanks for taking time to read through this.  I'm using a checkout from

the solr 3.x branch

My problem is with the highlighter and wildcards

I can get the highlighter to work with wild cards just fine, the problem
is that  solr is returning the term matched, when what I want it to do
is highlight the chars in the term that were matched.


Example:

http://192.168.1.75:8983/solr/music/select?indent=onq=name_title:wel*qt=beyondhl=truehl.fl=name_titlef.name_title.hl.usePhraseHighlighter=truef.name_title.hl.highlightMultiTerm=true

The results that come back look like this:

emWelcome/em   to the Jungle

What I want them to look like is this:
emWel/emcome to the Jungle

   From what I gathered by searching the archives is that solr 1.1 used to
do this... Is there a way to get that functionality?

Thanks!

  



-- 

This email and its attachments (if any) are for the sole use of the
intended recipient, and may contain private, confidential, and
privileged material. Any review, copying, or distribution of this
email, its attachments or the information contained herein is strictly
prohibited. If you are not the intended recipient, please contact the
sender immediately and permanently delete the original and any copies
of this email and any attachments.






Re: Solr Highlighting Question

2010-09-08 Thread Koji Sekiguchi

 (10/09/09 2:26), Jed Glazner wrote:

Thanks for taking time to read through this.  I'm using a checkout from

the solr 3.x branch

My problem is with the highlighter and wildcards

I can get the highlighter to work with wild cards just fine, the problem
is that  solr is returning the term matched, when what I want it to do
is highlight the chars in the term that were matched.


Example:

http://192.168.1.75:8983/solr/music/select?indent=onq=name_title:wel*qt=beyondhl=truehl.fl=name_titlef.name_title.hl.usePhraseHighlighter=truef.name_title.hl.highlightMultiTerm=true 



The results that come back look like this:

emWelcome/em   to the Jungle

What I want them to look like is this:
emWel/emcome to the Jungle

  From what I gathered by searching the archives is that solr 1.1 used to
do this... Is there a way to get that functionality?

Thanks!



Try to use FastVectorHighlighter on n-gram field for highlighting problem...
But FVH cannot process wildcard query. So you should query wel instead of
wel*. Then this makes you got unwanted hit like voemwel/em.
I don't think there is a solution for both of them with OOTB today.

There is a JIRA issue, but no patches there:

https://issues.apache.org/jira/browse/SOLR-1926

Koji

--
http://www.rondhuit.com/en/