Solr同义词实现难题：需原搜索词结果优先于同义词结果

Hey there, I’ve run into this exact synonym prioritization issue with Solr before—let’s break down how to fix it so your original search phrase results always come first, even when using bidirectional synonyms.

First, the core problem here is that Solr’s default synonym treatment treats both terms as semantically equivalent, so their match scores end up identical. To get the original phrase to rank higher, we need to explicitly boost matches on the user’s exact search term over its synonyms.

Here are a few actionable solutions tailored to your Solr 6.3.0 setup:

1. Explicit Boosting in Query Construction #

The most straightforward fix is to modify your query to give the original search phrase a higher boost than its synonym. For example, when a user searches for "campaign framework", adjust your query to prioritize matches on that exact phrase across all your fields, while still including the synonym with a lower boost.

Your updated query would look something like this (formatted for readability):

(title_t:"campaign framework")^3 OR (title_t:"integrated campaign")^1 OR
  (promotional_keywords_t:"campaign framework")^3 OR (promotional_keywords_t:"integrated campaign")^1 OR
  (promotional_description_t:"campaign framework")^3 OR (promotional_description_t:"integrated campaign")^1 OR
  (abstract_t:"campaign framework")^3 OR (abstract_t:"integrated campaign")^1 OR
  (body_t:"campaign framework")^3 OR (body_t:"integrated campaign")^1
)```
The `^3` gives the original phrase 3x the weight of the synonym, ensuring those results rank first. You’ll need to dynamically generate this query in your application layer—first resolve the user’s search term to its synonyms, then build the boosted clauses for each field.

### 2. Use SynonymGraphFilter with Boost Queries
Solr 6.3.0 includes `SynonymGraphFilterFactory`, which handles multi-word synonyms far better than the older `SynonymFilterFactory` by preserving position data (critical for accurate phrase matching). Combine this with Solr’s `bq` (Boost Query) parameter to add extra weight to exact matches of the original phrase.

First, update your field type to use `SynonymGraphFilterFactory` in both index and query analyzers:
```xml
<fieldType name="text_synonym" class="solr.TextField">
  <analyzer type="index">
    <tokenizer class="solr.StandardTokenizerFactory"/>
    <filter class="solr.LowerCaseFilterFactory"/>
    <filter class="solr.SynonymGraphFilterFactory" synonyms="synonyms.txt" expand="true"/>
  </analyzer>
  <analyzer type="query">
    <tokenizer class="solr.StandardTokenizerFactory"/>
    <filter class="solr.LowerCaseFilterFactory"/>
    <filter class="solr.SynonymGraphFilterFactory" synonyms="synonyms.txt" expand="true"/>
  </analyzer>
</fieldType>

Then, when constructing your query, use the bq parameter to boost exact matches of the original phrase across all relevant fields:

q=has_presentation_b:(True) AND (-exclude_from_search_b:(True)) AND (title_t:"campaign framework" OR promotional_keywords_t:"campaign framework" OR promotional_description_t:"campaign framework" OR abstract_t:"campaign framework" OR body_t:"campaign framework")
&bq=title_t:"campaign framework"^2 promotional_keywords_t:"campaign framework"^2 promotional_description_t:"campaign framework"^2 abstract_t:"campaign framework"^2 body_t:"campaign framework"^2

The bq parameter adds a score boost to any document matching the exact phrase, pushing those results to the top while still including synonym matches.

3. Tuning the Third-Party Synonym Plugin #

If you want to stick with the multi-word synonym plugin you tried, double-check its configuration for settings that prioritize original search terms. Many such plugins include options to assign higher weights to the user’s exact query phrase—look for parameters like preferOriginal or adjust the synonym mapping syntax to explicitly weight the original term higher.

Just make sure the plugin is fully compatible with Solr 6.3.0, as older plugins might have issues with the newer query and analysis pipelines.

Whichever approach you choose, test with both search terms ("integrated campaign" and "campaign framework") to verify that the original phrase results always rank above synonym matches.

内容的提问来源于stack exchange，提问作者lalit S. joshi