Question

I am using Solr Grouping result. But it behaves wrong.

I grouped on the base of email field. In my database i have 2 rows for the email address “[e-mail]” but SOLR showing 7147 numFound which is of course wrong, it should be 2.

When i try to search with particular email address like “[e-mail]” with grouping then it shows perfect result and it shows 2 in the numFound attribute which is correct.

I believe that, it is because of field type in my Solr schema, i was using text field type in start but i have changed it to my own field type now. I have defined my own field type and using my own field type now.

After using my own field type, i am facing same issue.

Answer 1

If you look at your grouping response, you will see it matched 7147 documents because the group consists of all emails that have “hotmail” in their address <str name="groupValue">hotmail</str> and not the entire email address. This behavior is because you are expecting the values in the fields to be indexed as complete strings, like “[e-mail]”, but from the definition of your email fieldType you are doing some tokenizing of the field values. That is resulting in multiple indexed values for that field. Specifically the StandardTokenizerFactory that splits a value on all non alphanumeric characters. So that same email address is being indexed as three separate values, “ashhaf63”, “hotmail” and “com”.

Because of this, I would recommend creating a new field that uses a simple string fieldType like the following:

<!-- The StrField type is not analyzed, but indexed/stored verbatim. -->
<fieldType name="string" class="solr.StrField" sortMissingLast="true" />

Then create a new field like this:

<field name="emailaddress" type="string" stored="true" indexed="true"/>

Then perform your grouping on this new emailaddress field, which will group on the entire email address value.