Java API for Graph

graph

#1

Does the current Java ES 2.3 API support graph or we need to use something like JEST ?


Documentation Java API for Graph
(Mark Harwood) #2

Hi Sharad,
Below is a demo which obviously depends on core elasticsearch resources and the jar from the xpack plugin for graph:

package org.elasticsearch;

import java.net.InetAddress;
import java.util.Collection;

import org.elasticsearch.client.transport.TransportClient;
import org.elasticsearch.common.settings.Settings;
import org.elasticsearch.common.transport.InetSocketTransportAddress;
import org.elasticsearch.graph.GraphPlugin;
import org.elasticsearch.graph.action.GraphExploreAction;
import org.elasticsearch.graph.action.GraphExploreRequestBuilder;
import org.elasticsearch.graph.action.GraphExploreResponse;
import org.elasticsearch.graph.action.Hop;
import org.elasticsearch.graph.action.Vertex;
import org.elasticsearch.index.query.QueryBuilders;

/**
 * Demo graph client - find elastic meetup attendees and what other meetup groups they frequent
 */
public class GraphDemoClient {

	public static void main(String[] args) throws Exception {
	
		TransportClient client = TransportClient.builder()
				.settings(Settings.builder()
					.put("cluster.name", "elasticsearch")			    
					.build())
				.addPlugin(GraphPlugin.class)
				.build()
				.addTransportAddress(new InetSocketTransportAddress(InetAddress.getByName("localhost"), 9300));

	
		int requiredNumberOfSightings=1;
	
		GraphExploreRequestBuilder grb = new GraphExploreRequestBuilder(client, GraphExploreAction.INSTANCE).setIndices("meetuprsvps");
		Hop hop1 = grb.createNextHop(QueryBuilders.termQuery("_all", "elasticsearch"));
		// elasticsearch meetup attendees
		hop1.addVertexRequest("member_id").size(10).minDocCount(requiredNumberOfSightings); 
		// groups attended by elastic attendees
		grb.createNextHop(null).addVertexRequest("group_id").size(10).minDocCount(requiredNumberOfSightings); 
	
		GraphExploreResponse response = grb.get();
		Collection<Vertex> vertices = response.getVertices();
	
		System.out.println("==Members===");
		for (Vertex vertex : vertices) {
			if(vertex.getField().equals("member_id")){
				System.out.println(vertex.getTerm());        		
			}
		}
	
		System.out.println("==Groups===");
		for (Vertex vertex : vertices) {
			if(vertex.getField().equals("group_id")){
				System.out.println(vertex.getTerm());        		
			}
		}        

		client.close();

	}

}

Pom missing for 5.3.1 SNAPSHOT?
#3

Hi Mark,

Can you please help me with the Maven details for the graph plugin as I couldn't find it in the maven repo?

-Sharad


(Mark Harwood) #4

The graph jar you need is in the graph plugin zip. This zip should be unzipped into the "plugins" directory of elasticsearch as part of installation.


(stefani hubert) #5

if you add the following repository in your pom.xml you should be able to get the dependency properly managed :

<repositories>
    <repository>
        <id>elastic-plugins</id>
        <name>elastic plugins</name>
        <url>https://maven.elasticsearch.org/releases/</url>
    </repository>
</repositories>



<dependency>
    <groupId>org.elasticsearch.plugin</groupId>
    <artifactId>graph</artifactId>
    <version>2.3.3</version>
</dependency>

(David She) #6

Hi,
I have trawled through the elastic site and google looking for the Java API to use for ElasticSearch Graph. I notice your snippet above but would to have a link to the documentation or a tutorial.
Appreciated


(Ayush Rastogi) #7

Hello Sir,
I am new in elasticSearch. I use your program for creating the graph in elasticSearch and I want to see this graph in Kibana. Can you please help me for this. and how can I use queries on this graph like - shortest path query?


(Mark Harwood) #8

Please do not hijack this Java API question with a different topic (UI questions).
I'll answer your questions here but if you want to continue the discussion please open a new question topic.

If you want to view Graphs in Kibana install the x-pack plugin and here is an example of how to configure the UI.

That would be done outside of elasticsearch/kibana currently. We are working on an export-to-GraphML feature which would allow you to use tools like Gephi which offer various Graph algos e.g. shortest-path, centrality measures, community detection etc wrapped up in a UI. If you just want to write code that does graph analysis consider using tools like Python's NetworkX library - here is an example calling the Graph API and performing betweenness centrality measures using networkX.

If you want to discuss any of the above further please open a new topic so we can keep this one focused on the Java API.


(system) #9