Not documented: how to get fulltext indexing not to drop hyphens (dashes)

[clem]

Full text indexing (Indexes for full-text search - Neo4j Cypher Manual ) works well for me except the default analyzer strips hyphens from indexing.

Consider that the search for "Emergency" should not return "Non-Emergency". Product codes often have dashes in them too.

This is actually a lucene question/issue. I did find this:

stackoverflow.com

Lucene Index problems with "-" character

java, lucene, indexing, escaping

asked by Zteve on 07:05AM - 17 Apr 12 UTC

So this works:

CALL db.index.fulltext.createNodeIndex("IndexName",["Label"],["property"], 
{ analyzer: "classic"}) // does not throw out hyphens, e.g. Non-Emergency is one word

This doesn't seem to be documented in the Neo4J documentation.

[clem]

Oops... I didn't actually try this out until just now. The 'classic' Analyzer option doesn't do what the stackoverflow page claims.

'whitespace' sort of works except it won't split on other punctation marks (e.g. comma, period, etc."

Rat