Details
-
Bug
-
Resolution: Won't Fix
-
Major
-
4.5.0
-
Untriaged
-
Unknown
Description
Beta feedback. This customer currently uses ES and the ES Plugin and has evaluated 4.5 Beta Full Text Search.
- At index build time, they want to be able to run a process on ingestion or call an API and add the results of that API call to the index but not to the original document.
- In this use case, they do NLP / entity extraction and add the canonical form to a stored field, which may or may not be searchable but certainly can be retrieved with the search results.
- Saving the changes to the document before ingestion is not a good option because they display the documents to their users today directly, so any fields added just for ingestion would be shown to customers.
- They have some natural language fields, want to do entity extraction, turn those into canonical phrases. They don't today use n-gram to help with auto-completion / type ahead, but they plan to do so.
- Search query completion- typing in someone's name, the full name will be pulled out as a single token (e.g. "lin" --> "Lindsey Vonn" pops out)