Details
-
Improvement
-
Resolution: Obsolete
-
High
-
None
-
None
Description
Current Elasticsearch Search Engine implementation index Content field data in nested documents, one per language.
One problem with this approach is that it is not in line with how relevancy is calculated. As website typically shows Content in a single translation, having all languages indexed in the same index will result in incorrect search statistics, meaning scoring will not be correct for the content displayable on the website.
Another problem is that nested documents were implemented for very narrow purpose, and using them in this way might have negative performance and scaling impact.
Using multiple indexes to index Content per translation would enable removing usage of nested documents to index field data per translation. It would also enable correct relevancy calculation per translation.
Attachments
Issue Links
- relates to
-
EZP-24300 Search engines workshop in Cologne
- Closed