Does FullText indexes support SpanNearQuery

SimonThordal · January 21, 2021, 10:13am

Say you want to search for something like "safety program" but also want to account for "safety programmes" and stemming isn't an option as the indexed documents are in many different languages. The way to do this in Lucene would be with a SpanNearQuery with a slop of 0 and a wildcard at the the end of the 2nd term. When trying to do something like "safety program"* in the fulltext search however it seems very slow. Is this because it isn't supported in the Lucene version Neo4j is using or is the query wrong somehow?

clem · January 22, 2021, 6:23pm

My guess is that there is no glue API in APOC to this.

call apoc.help("SpanNearQuery")

returns nothing. call apoc.help("Query") returns a lot of query related APOC calls but none for what you want.

I think you need to create an issue for APOC lucene to support this call:

michael.hunger · January 23, 2021, 2:35pm

if there is a lucense query syntax for the SpanNearQuery you should just be able to use that.

Neo4j 4.2 uses Lucene 8.2.0

SimonThordal · January 28, 2021, 7:54am

Thank you Michael - do you happen to know the Lucene version 3.5 is on?

michael.hunger · January 29, 2021, 10:12pm

You can check yourself

github.com

neo4j/neo4j/blob/3.5/pom.xml#L68


      
              <name>Nexus Release Repository</name>
              <url>https://oss.sonatype.org/service/local/staging/deploy/maven2/</url>
            </repository>
          </distributionManagement>
          
          
<properties>
            <project.build.sourceEncoding>UTF-8</project.build.sourceEncoding>
            <maven.compiler.source>1.8</maven.compiler.source>
            <maven.compiler.target>1.8</maven.compiler.target>
            <licensing.prepend.text>licensing/notice-agpl-prefix.txt</licensing.prepend.text>
            <lucene.version>5.5.5</lucene.version>
            <bouncycastle.version>1.68</bouncycastle.version>
            <generate-config-docs-phase>prepare-package</generate-config-docs-phase>
            <java.exports/>
            <test.runner.jvm/>
            <test.runner.jvm.settings.additional/>
            <test.runner.jvm.settings>-Xmx2G -XX:+UseG1GC -XX:-OmitStackTraceInFastThrow -XX:+HeapDumpOnOutOfMemoryError
              -XX:HeapDumpPath=target/test-data -Dorg.neo4j.unsafe.impl.internal.dragons.UnsafeUtil.DIRTY_MEMORY=true
              -Dorg.neo4j.unsafe.impl.internal.dragons.UnsafeUtil.CHECK_NATIVE_ACCESS=true
              -Dorg.neo4j.io.pagecache.impl.muninn.usePreciseCursorErrorStackTraces=true
              -Dorg.neo4j.kernel.impl.api.KernelStatement.trackStatements=true

5.5.5

Topic		Replies	Views
Full-text search skip and limit Neo4j Graph Platform search , migrated , lucene , full-text-search	2	453	December 21, 2022
Legacy indexing - Lucene Store closes spontaneously? Neo4j Graph Platform apoc	4	1080	October 25, 2018
Fulltext Search with apoc.periodic iterate Procedures & APOC	0	281	November 3, 2020
Lucene’s MoreLikeThis feature Cypher cypher	4	726	May 31, 2019
Fulltext search on keyword analyzer not working as expected Newbie Questions	4	720	February 3, 2020

Does FullText indexes support SpanNearQuery

Related topics