As of Neo4j 3.5 we have a fulltext search index. Specifically it’s Lucene 5.5.5. In Lucene 5.5.4 a magical feature was added to Lucene called MoreLikeThis. Given a document (translates to a node with properties containing text strings) this feature will find all other documents that are most like it in comparison to the entire collection of documents in question.
I would like to experiment with this Lucene feature. How can I call that function from Cypher? Is it present out of the box with Neo4j? Do I have to import it?
That functionality is not available via Cypher. Technically it should be possible by using private APIs to get a reference to the IndexReader instance and then use a stored procedure implementing code inspired by Lucene 'MoreLikeThis' Example Code - DZone.
I consider this an interesting feature for apoc. Do you mind creating a github issue for this at Issues · neo4j-contrib/neo4j-apoc-procedures · GitHub ?
Love the enthusiasm Stefan!
1 Like
I'll investigate on it but cannot commit on a timeline yet - being "slightly" busy.
We're happy to assist with testing.