lucene-solr-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From kenf_nc <ken.fos...@realestate.com>
Subject Single value vs multi value setting in tokenized field
Date Sun, 16 Jan 2011 20:47:56 GMT

I have to support both general searches (free form text) and directed
searches (field:val field2:val). To do the general search I have a field
defined as:
   <field name="content" type="text" indexed="true" stored="false"
termVectors="true" multiValued="true" />
and several copyField commands like:
  <copyField source="description" dest="content" />
  <copyField source="title" dest="content" />
  <copyField source="tags" dest="content" />
  <copyField source="features" dest="content" />
Note that tags and features are multi-value themselves. So after indexing I
have a 'general text' bucket with numerous (usually in the 20 to 30 range)
rows of strings. 

My question is would it be better, for indexing speed and search
speed/quality, to concatenate all the text into a single string and store it
in "content" as one value? What are the implications on search results? If
Description is say a couple paragraphs of text and tags are
"Cuisine","Italian","Romantic" would the tags get lost in the muck of the
bigger text?

One thing to keep in mind. I'm sure some of you are going to say 'Dismax'
and in some situations I will, but my index has numerous document types that
have vastly different schemas. Another document may not have "title" and
"features" but might have "recommendations" and "location". In a general
query it wouldn't make sense to include every possible field in a dismax
query, I don't even know what all the fields are, new ones are added all the
time.

Has anyone got advice, suggestions on this topic (blending directed search
with general search)? 
Thanks in advance,
Ken
-- 
View this message in context: http://lucene.472066.n3.nabble.com/Single-value-vs-multi-value-setting-in-tokenized-field-tp2268635p2268635.html
Sent from the Solr - User mailing list archive at Nabble.com.

Mime
View raw message