如何自定义Solr SpellCheck Component返回的freq字段为frequency字段？

Absolutely, you can replace the default freq value (which counts document occurrences, hence 1 for your single-record terms) with the custom frequency field from your dictionary core. Here's a step-by-step solution using Solr's DirectSolrSpellChecker:

Step 1: Update SpellCheck Component Configuration #

Modify your solrconfig.xml to use DirectSolrSpellChecker and explicitly specify your frequency field as the source for spell check frequencies. This checker supports using indexed numeric fields as weight values instead of just document counts.

<searchComponent name="spellcheck" class="solr.SpellCheckComponent">
  <lst name="spellchecker">
    <str name="name">default</str>
    <!-- Use DirectSolrSpellChecker to leverage custom frequency fields -->
    <str name="classname">solr.DirectSolrSpellChecker</str>
    <str name="field">keyword</str> <!-- Your vocabulary field -->
    <str name="frequencyField">frequency</str> <!-- Map to your custom frequency field -->
    <str name="buildOnCommit">false</str> <!-- Match your existing build setting -->
    <!-- Optional: Adjust other parameters like similarity threshold if needed -->
  </lst>
</searchComponent>

Step 2: Verify Field Configuration #

Ensure your dictionary core's schema (either schema.xml or managed-schema) defines the frequency field as indexed (required for the spell checker to read its values) and as a numeric type:

<field name="frequency" type="int" indexed="true" stored="true"/>

Step 3: Rebuild the Spell Check Dictionary #

Even though you use spellcheck.build=false in requests, you'll need to build the dictionary once to populate it with your frequency values. Run this request against your dictionary core:

spell?omitHeader=true&spellcheck.build=true

Step 4: Test Your Spell Check Request #

Now when you run your original request:

spell?omitHeader=true&wt=xml&json.nl=flat&spellcheck=true&spellcheck.q=faceboek&spellcheck.build=false

The response should return your custom frequency value instead of 1:

<lst> 
  <str name="word">facebook</str> 
  <int name="freq">89504</int> 
</lst>

Key Notes #

• Spell Checker Type: Only DirectSolrSpellChecker supports the frequencyField parameter—other implementations like IndexBasedSpellChecker don't have this capability.
• Field Data Type: Your frequency field must be a numeric type (int, long, float, etc.) to be used as a frequency weight.

内容的提问来源于stack exchange，提问作者Simon Brink