org.neo4j.gds.embeddings.node2vec.Node2VecRandomWalkTask Maven / Gradle / Ivy
Go to download
Show more of this group Show more artifacts with this name
Show all versions of algo Show documentation
Show all versions of algo Show documentation
Neo4j Graph Data Science :: Algorithms
The newest version!
/*
* Copyright (c) "Neo4j"
* Neo4j Sweden AB [http://neo4j.com]
*
* This file is part of Neo4j.
*
* Neo4j is free software: you can redistribute it and/or modify
* it under the terms of the GNU General Public License as published by
* the Free Software Foundation, either version 3 of the License, or
* (at your option) any later version.
*
* This program is distributed in the hope that it will be useful,
* but WITHOUT ANY WARRANTY; without even the implied warranty of
* MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
* GNU General Public License for more details.
*
* You should have received a copy of the GNU General Public License
* along with this program. If not, see .
*/
package org.neo4j.gds.embeddings.node2vec;
import org.neo4j.gds.api.Graph;
import org.neo4j.gds.termination.TerminationFlag;
import org.neo4j.gds.core.utils.progress.tasks.ProgressTracker;
import org.neo4j.gds.ml.core.samplers.RandomWalkSampler;
import org.neo4j.gds.traversal.NextNodeSupplier;
import java.util.concurrent.atomic.AtomicLong;
final class Node2VecRandomWalkTask implements Runnable {
private final Graph graph;
private final NextNodeSupplier nextNodeSupplier;
private final int walksPerNode;
private final ProgressTracker progressTracker;
private final TerminationFlag terminationFlag;
private final AtomicLong walkIndex;
private final CompressedRandomWalks compressedRandomWalks;
private final RandomWalkProbabilities.Builder randomWalkProbabilitiesBuilder;
private final RandomWalkSampler sampler;
private final int walkBufferSize;
private int walks;
private int maxWalkLength;
private long maxIndex;
Node2VecRandomWalkTask(
Graph graph,
NextNodeSupplier nextNodeSupplier,
int walksPerNode,
RandomWalkSampler.CumulativeWeightSupplier cumulativeWeightSupplier,
ProgressTracker progressTracker,
TerminationFlag terminationFlag,
AtomicLong walkIndex,
CompressedRandomWalks compressedRandomWalks,
RandomWalkProbabilities.Builder randomWalkProbabilitiesBuilder,
int walkBufferSize,
long randomSeed,
int walkLength,
double returnFactor,
double inOutFactor
) {
this.graph = graph;
this.nextNodeSupplier = nextNodeSupplier;
this.walksPerNode = walksPerNode;
this.progressTracker = progressTracker;
this.terminationFlag = terminationFlag;
this.walkIndex = walkIndex;
this.compressedRandomWalks = compressedRandomWalks;
this.randomWalkProbabilitiesBuilder = randomWalkProbabilitiesBuilder;
this.walkBufferSize = walkBufferSize;
this.sampler = RandomWalkSampler.create(
graph,
cumulativeWeightSupplier,
walkLength,
returnFactor,
inOutFactor,
randomSeed
);
this.walks = 0;
this.maxWalkLength = 0;
this.maxIndex = 0;
}
private boolean consumePath(long[] path) {
var index = walkIndex.getAndIncrement(); //perhaps we can also use a buffer to minimize walkIndex atomic operations
maxIndex = index;
randomWalkProbabilitiesBuilder.registerWalk(path);
compressedRandomWalks.add(index, path);
maxWalkLength = Math.max(path.length, maxWalkLength);
if (walks++ == walkBufferSize) {
walks = 0;
return this.terminationFlag.running();
}
return true;
}
int maxWalkLength() {
return maxWalkLength;
}
long maxIndex() {
return maxIndex;
}
@Override
public void run() {
long nodeId;
while (true) {
nodeId = nextNodeSupplier.nextNode();
if (nodeId == NextNodeSupplier.NO_MORE_NODES) break;
if (graph.degree(nodeId) == 0) {
progressTracker.logProgress();
continue;
}
sampler.prepareForNewNode(nodeId);
for (int walkIndex = 0; walkIndex < walksPerNode; walkIndex++) {
var path = sampler.walk(nodeId);
boolean shouldContinue = consumePath(path);
if (!shouldContinue) {
break;
}
}
progressTracker.logProgress();
}
}
}
© 2015 - 2025 Weber Informatics LLC | Privacy Policy