search.md

title	layout
Search	default

Note: This page is actively under development. At present, it's more of a collection of thoughts than actual documentation.

##solr_home directory

The /solr_home/ directory can be kept anywhere. There's no functional need for it to be available within the scope of the website. By default, it lives in parallel to /htdocs/ and /includes/. Its location is only relevant when starting Solr, which is when the path is provided, e.g. -Dsolr.solr.home=/var/www/example.com/solr_home/.

Tag weighting

In /solr_home/statedecoded/conf/solrconfig.xml, the request handler for standard searches (<requestHandler name="/search" class="solr.SearchHandler">) is configured to give tags the same weight as all other indexed fields. Depending on the source of your tags, you may be well served by adjusting this number.

For instance, if your tags are well written, accurate, and not over-broad, then you could increase its weight from the (implied) default of 1.0, and take it up to 2.0, 5.0, even 10.0, for Solr to value tags, in determining the relevance of a given law, that multiple of times above all other fields.

On the other hand, if your tags are good-not-great, then you'd want to decrease the weight of tags, down to .5 or .2 or even .1, to have Solr value other fields beyond tags, but still include tags in its index.

This is done by appending ^5 (to weight tags at 500% that of other fields) or ^.1 (to weight tags at just 10% of other fields) after tags within the <str name="…"> tag, e.g.:

<str name="qf">
                catch_line
                      tags^5
                      text
                 structure
          section_ancestor^10
         section_descendent^10
       refers_to_descendent
         refers_to_ancestor
  referred_to_by_descendent
    referred_to_by_ancestor
                       term
                 definition
</str>

Section number tokenizer

Within /solr_home/statedecoded/conf/schema.xml there is a tokenizer that indexes cross-references mentions of other laws within a law. The job of this tokenizer is to identify the pattern of a cross-reference, treat it as a single unit of data (instead of inadvertently breaking it up), and index it.

Here's what the tokenizer looks like:

   <fieldType name="descendent_facet_hierarchical" class="solr.TextField">
     <analyzer type="index">
       <charFilter class="solr.PatternReplaceCharFilterFactory" pattern="[\.-]" replacement="~"/>
       <!-- remove () or [] that somebody might use to identify a chapter -->
       <charFilter class="solr.PatternReplaceCharFilterFactory" pattern="\(.*?\)$" replacement=""/>
       <charFilter class="solr.PatternReplaceCharFilterFactory" pattern="\[.*?\]$" replacement=""/>
       <tokenizer class="solr.PathHierarchyTokenizerFactory" delimiter="~" />
       <filter class="solr.LowerCaseFilterFactory" />
     </analyzer>
     <analyzer type="query">
       <charFilter class="solr.PatternReplaceCharFilterFactory" pattern="[\.-]" replacement="~"/>
       <tokenizer class="solr.KeywordTokenizerFactory" />
       <filter class="solr.LowerCaseFilterFactory" />
     </analyzer>
   </fieldType>

As of this writing, this tokenizer is decoupled from the configuration performed in /includes/config-sample.inc.php, which is to say that it may require manual configuration. As provided, it should work for most legal codes.