Please comment on this use-case for accessing the Lucene API directly from the context of a Neo4j User Defined Procedure -
i build in-memory lucene index collection of node properties , search against index.
these search transactions happening in parallel, need able construct separate search index each transaction. seems not possible using native (manual) neo4j indexes, since "global", hence use of memory-based search index, mistaken?
this working great our use case. couple of notes:
- use memoryindex rather ramdirectory.
- be sure specify "provided" in dependencies lucene.
- you can use 5.5.0, used natively neo4j.
Comments
Post a Comment