search speed, full text case insensitive indexing

[Krishna Shetty]

I am working search feature. I need to return all nodes with matching string in 'name' property. 

I am using the query as suggested here:

match <mypattern> where n.name =~ '.*SUBSTRING.*' return n.Name;

I have also added index on name property.

CREATE INDEX ON :MyLabel1(name);

But the query is not fast.

I have read about full text case insensitive indexing, please suggest how to do this through cypher query.

Or is there any other way to improve search speed.

Thanks

[Michael Hunger]

See my blog post

http://jexp.de/blog/2014/03/full-text-indexing-fts-in-neo4j-2-0/

Sent from mobile device

--
You received this message because you are subscribed to the Google Groups "Neo4j" group.
To unsubscribe from this group and stop receiving emails from it, send an email to neo4j+un...@googlegroups.com.
For more options, visit https://groups.google.com/d/optout.

[Krishna Shetty]

Thank you.

Did the following:

>sudo vim conf/neo4j-server.properties

Added below two lines at the bottom(these lines were not there already)

node_auto_indexing=true

node_keys_indexable=name

Restarted neo4j.

And ran below cypher:

>MATCH (n) WHERE has(n.name) SET n.name=n.name;

+-------------------+

| No data returned. |

+-------------------+

Properties set: 94701

19652 ms

It used to take around 4sec, I think same search query now takes around 2 seconds

I am using neo4j-community-2.0.1 on AWS EC2 medium instance.

I think it still takes long time return search result.

Is neo4j suitable for these kind of searches?

Thanks

[Michael Hunger]

Wrong config file

It's neo4j.properties

Also misses the prop-keys

Sent from mobile device

[Krishna Shetty]

Updated in conf/neo4j.properties. i.e.,

# Enable auto-indexing for nodes, default is false

node_auto_indexing=true

# The node property keys to be auto-indexed, if enabled

node_keys_indexable=name

I have not done the step 'Setup Node Auto-Index as Fulltext-Index'. This has REST POST calls. Is this setting mandatory? I haven't tried REST calls, is it possible to make this setting from 'neo4j-shell'?

Also, I am not using ':node_auto_index' in my query. I need to find-out how to use ':node_auto_index' in 'match' as I don't use 'start' in my query.

Query still takes 1.6 seconds. 

Thanks,

Krishna

--
You received this message because you are subscribed to a topic in the Google Groups "Neo4j" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/neo4j/mPBSwxhEfA4/unsubscribe.
To unsubscribe from this group and all its topics, send an email to neo4j+un...@googlegroups.com.

[Michael Hunger]

Sent from mobile device

Am 26.04.2014 um 19:42 schrieb Krishna Shetty <ksh...@livingtree.com>:

Updated in conf/neo4j.properties. i.e.,

# Enable auto-indexing for nodes, default is false

node_auto_indexing=true

# The node property keys to be auto-indexed, if enabled

node_keys_indexable=name

I have not done the step 'Setup Node Auto-Index as Fulltext-Index'. This has REST POST calls. Is this setting mandatory? I haven't tried REST calls, is it possible to make this setting from 'neo4j-shell'?

You have to configure the index correctly _upfront_

Also, I am not using ':node_auto_index' in my query. I need to find-out how to use ':node_auto_index' in 'match' as I don't use 'start' in my query

You _have_ to use start for fts

[Krishna Shetty]

Thank you.

It takes around 300ms now.

I used to curl to make Auto-Index as Fulltext-Index REST call as below.

curl -H Accept:application/json -H charset=UTF-8 -H Content-Type:application/json -X POST -d '{"name":"node_auto_index", "config":{"type":"fulltext","provider":"lucene"}}' http://localhost:7474/db/data/index/node

Thanks,

Krishna

[Michael Hunger]

Make sure to do that as first thing on a blank db

Sent from mobile device