点邻居搜索 (VertexNeighborSearchBatchOp)

Java 类名：com.alibaba.alink.operator.batch.graph.VertexNeighborSearchBatchOp

Python 类名：VertexNeighborSearchBatchOp

功能介绍

节点k度邻居子图查询算法，根据用户输入的图数据和一组节点，在图中查找它们的k度邻居，然后导出子图。在子图较小时，还提供图可视化功能。

参数说明

该组件至少接入一个输入桩，表示图的边集；可选接入第二个输入桩，表示图的点集。组件输出对应于输入桩，分别表示子图的边集和点集，其中点集在第二个输入桩有输入时才有输出。

名称	中文名称	描述	类型	是否必须？	取值范围	默认值
edgeSourceCol	边表中起点所在列	边表中起点所在列	String	✓
edgeTargetCol	边表中终点所在列	边表中终点所在列	String	✓
sources	起始节点集合	起始节点集合	String[]	✓
asUndirectedGraph	是否为无向图	是否为无向图	Boolean			true
depth	深度	寻找邻居的深度	Integer			1
vertexIdCol	节点ID列	表示节点ID的列名	String			“id”

代码示例

Python 代码

from pyalink.alink import *

import pandas as pd

useLocalEnv(1)

df = pd.DataFrame([
    ["Alice", "Lisa", 1., "hello"],
		["Lisa", "Alice", 1., "hello"],
		["Lisa", "Karry", 1., "hello"],
		["Karry", "Lisa", 1., "213"],
		["Karry", "Bella", 1., "hello"],
		["Bella", "Karry", 1., "h123ello"],
		["Bella", "Lucy", 1., "hello"],
		["Lucy", "Bella", 1., "hello"],
		["Lucy", "Bob", 1., "123"],
		["Bob", "Lucy", 1., "hello"],
		["John", "Bob", 1., "hello"],
		["Bob", "John", 1., "hello"],
		["John", "Stella", 1., "123"],
		["Stella", "John", 1., "hello"],
		["Kate", "Stella", 1., "hello"],
		["Stella", "Kate", 1., "hello"],
		["Kate", "Jack", 1., "13"],
		["Jack", "Kate", 1., "hello"],
		["Jess", "Jack", 1., "13"],
		["Jack", "Jess", 1., "hello"],
		["Jess", "Jacob", 1., "hello"],
		["Jacob", "Jess", 1., "123"]]
)
data = BatchOperator.fromDataframe(df, schemaStr="start string, end string, value double, attr string")
op = VertexNeighborSearchBatchOp() \
    .setDepth(1) \
    .setSources(["John", "Lisa"]) \
    .setEdgeSourceCol("start") \
    .setEdgeTargetCol("end") \
    .setVertexIdCol("name") \
    .setAsUndirectedGraph(False)
op.linkFrom(data).print()

Java 代码

import org.apache.flink.types.Row;

import com.alibaba.alink.operator.batch.source.MemSourceBatchOp;
import org.junit.Test;

import java.util.Arrays;
import java.util.List;

public class VertexNeighborSearchBatchOpTest {
	@Test
	public void testDemo() throws Exception {
		List <Row> edgesRows = Arrays.asList(
			Row.of("Alice", "Lisa", 1., "hello"),
			Row.of("Lisa", "Alice", 1., "hello"),
			Row.of("Lisa", "Karry", 1., "hello"),
			Row.of("Karry", "Lisa", 1., "213"),
			Row.of("Karry", "Bella", 1., "hello"),
			Row.of("Bella", "Karry", 1., "h123ello"),
			Row.of("Bella", "Lucy", 1., "hello"),
			Row.of("Lucy", "Bella", 1., "hello"),
			Row.of("Lucy", "Bob", 1., "123"),
			Row.of("Bob", "Lucy", 1., "hello"),
			Row.of("John", "Bob", 1., "hello"),
			Row.of("Bob", "John", 1., "hello"),
			Row.of("John", "Stella", 1., "123"),
			Row.of("Stella", "John", 1., "hello"),
			Row.of("Kate", "Stella", 1., "hello"),
			Row.of("Stella", "Kate", 1., "hello"),
			Row.of("Kate", "Jack", 1., "13"),
			Row.of("Jack", "Kate", 1., "hello"),
			Row.of("Jess", "Jack", 1., "13"),
			Row.of("Jack", "Jess", 1., "hello"),
			Row.of("Jess", "Jacob", 1., "hello"),
			Row.of("Jacob", "Jess", 1., "123")
		);

		MemSourceBatchOp edgesSource = new MemSourceBatchOp(edgesRows,
			"start string, end string, value double, attr string");
		VertexNeighborSearchBatchOp op = new VertexNeighborSearchBatchOp()
			.setDepth(1)
			.setSources("John", "Lisa")
			.setEdgeSourceCol("start")
			.setEdgeTargetCol("end")
			.setVertexIdCol("name")
			.setAsUndirectedGraph(false);
		op.linkFrom(edgesSource).print();
	}
}

运行结果

start	end	value	attr
Alice	Lisa	1.0000	hello
Lisa	Karry	1.0000	hello
John	Bob	1.0000	hello
John	Stella	1.0000	123
Lisa	Alice	1.0000	hello
Karry	Lisa	1.0000	213
Bob	John	1.0000	hello
Stella	John	1.0000	hello