This chapter covers
- Debugging relevancy scores
- Boosting relevancy based upon fields, terms, and payloads
- Crafting function queries to improve relevancy scores
- Using Solr for personalized search and recommendations
- Running and measuring relevancy experiments
The overview of Solr’s default relevancy scoring algorithm in chapter 3 introduced how Solr calculates the similarity score (by default) between a query and its matching documents. Since then, most of this book has focused upon using additional search features to scale Solr, improving the user experience, and configuring text and query analysis in order to find the right set of documents which match any given query.
Once that right set of documents is found, it’s usually just as important to order those documents based upon relevancy to ensure your application’s users find the best matches at the very top of their search results. Although Solr’s default similarity calculation works well on generic text, you can usually significantly improve the relevancy of your search results by passing along additional information to Solr about your content.