Add fulltext Lucene analyzers for faster search in text data

[jarasch]

We are preparing Cypher queries for users that want to query data either via Cypher (Neo4j-Browser) or Neo4j-Bloom.

Therefore we need to build text analyzers on the text properties on the following lables/properties:

• Fragment.text
• Paper.title
• GeneSymbol.sid
• Gene.name
• Protein.name
• PatentClaim.text
• PatentTitle.text
• PatentAbstract.text
• Entity.name

[jarasch]

CALL db.index.fulltext.createNodeIndex("textOfPapersAndPatents",["Fragment", "Abstract", "Paper", "Patent", "PatentTitle", "PatentClaim","PatentAbstract"],["title", "text"])

[jarasch]

// Fulltext index on GeneSymbol where the gene name is stored in property sid
CALL db.index.fulltext.createNodeIndex("GeneSymbolFullTextIndex",["GeneSymbol"],["sid"])

[jarasch]

// Fulltext index on author names CALL db.index.fulltext.createNodeIndex("AuthorFullTextIndex",["Author"],["first", "middle","last"])

[jarasch]

// Fulltext index on entity names like company names CALL db.index.fulltext.createNodeIndex("EntityFullTextIndex",["Entity"],["name"])

[motey]

A dedicated loader to create all needed text indexes would make sense. this loader can be mounted into the motherlode pipeline.

Can one create text indexes on nodes that are not existing yet?

if yes we can collect all text indexes (including these from other loaders that are allready existing) at one place and create them at the beginning of the pipeline.

[motey]

https://github.com/covidgraph/graph-processing_fulltext-indexes

Will run this against DEV today

[motey]

Indexes are on DEV and PRD