apache
Expand Full Text Search Capabilities
Our FTS capabilities are pretty limited right now -- REGEXP_MATCH does a primitive regex match and TEXT_MATCHES is simply a wrapper around basic Lucene level searching (does not expose fuzzy queries or the option to rank).
This issue states a proposal for expanding our FTS capabilities with the following initial ideas:
- Support LIKE operator (https://www.w3schools.com/sql/sql_like.asp).
- Improve TEXT_MATCH (see #7215 , #7216 and #7217 ).
- Allow ranked searches (something on the lines of https://www.postgresql.org/docs/9.5/textsearch-controls.html#TEXTSEARCH-RANKING)
this is an relatively old issue, and some things are already implemented.
Coming from the ecommerce field where good text search is an absolute must-have, the question is: why shouldn't search be from the same quality in real-time analytics??
here are some screenshots of the backend settings of a great lucene integration to a shop software (magento addon by extendware) yes they are a little aged and don't include the latest things like using embeddings for similarity search, but they are still a perfect baseline.. => Maybe some of these settings and their functionality are a good inspiration for advancing full text search in pinot further:
