Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

feat(algorithm): support biased second order random walk #280

Merged
merged 17 commits into from
Dec 4, 2023
Merged

feat(algorithm): support biased second order random walk #280

Changes from 1 commit
Commits
Show all changes
17 commits
Select commit Hold shift + click to select a range
9b7ec3e
chore: option rename
diaohancai Nov 5, 2023
7d760ee
feat: define second order random walk algorithm param
diaohancai Nov 5, 2023
5d8eb98
chore: code format
diaohancai Nov 5, 2023
3e4da8d
feat: add defaultWeight param
diaohancai Nov 5, 2023
a2b6efc
feat: implement second order random walk algorithm
diaohancai Nov 7, 2023
0c14381
feat: ListValue support getFirst
diaohancai Nov 7, 2023
5914bff
bug: null point
Nov 8, 2023
77d3752
chore: supplementary test parameters
Nov 8, 2023
70d8f0f
optimize: change weightThreshold parameter to Double type
Nov 8, 2023
548148e
optimize: weight threshold truncation
Nov 8, 2023
b4b94b1
optimize: method optimize
diaohancai Nov 13, 2023
7d6e8ca
chore: nullableKeys test
diaohancai Nov 13, 2023
18f1260
refactor: double value simplification
diaohancai Nov 15, 2023
9a93341
chore: add todo note
diaohancai Nov 15, 2023
37d2aa7
fix: can not load edge properties
diaohancai Nov 17, 2023
8cc1734
chore: remove TODO
diaohancai Nov 20, 2023
94bd35b
Merge branch 'master' into pr/280
imbajin Dec 4, 2023
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
53 changes: 20 additions & 33 deletions ...-algorithm/src/main/java/org/apache/hugegraph/computer/algorithm/sampling/RandomWalk.java
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -122,46 +122,36 @@ public void init(Config config) {
"actual got '%s'",
OPTION_WALK_PER_NODE, this.walkPerNode);
}
LOG.info("[RandomWalk] algorithm param, {}: {}", OPTION_WALK_PER_NODE, this.walkPerNode);

this.walkLength = config.getInt(OPTION_WALK_LENGTH, 3);
if (this.walkLength <= 0) {
throw new ComputerException("The param %s must be greater than 0, " +
"actual got '%s'",
OPTION_WALK_LENGTH, this.walkLength);
}
LOG.info("[RandomWalk] algorithm param, {}: {}", OPTION_WALK_LENGTH, this.walkLength);

this.weightProperty = config.getString(OPTION_WEIGHT_PROPERTY, "");
LOG.info("[RandomWalk] algorithm param, {}: {}",
OPTION_WEIGHT_PROPERTY, this.weightProperty);

this.defaultWeight = config.getDouble(OPTION_DEFAULT_WEIGHT, 1);
if (this.defaultWeight <= 0) {
throw new ComputerException("The param %s must be greater than 0, " +
"actual got '%s'",
OPTION_DEFAULT_WEIGHT, this.defaultWeight);
}
LOG.info("[RandomWalk] algorithm param, {}: {}",
OPTION_DEFAULT_WEIGHT, this.defaultWeight);

this.minWeightThreshold = config.getDouble(OPTION_MIN_WEIGHT_THRESHOLD, 0.0);
if (this.minWeightThreshold < 0) {
throw new ComputerException("The param %s must be greater than or equal 0, " +
"actual got '%s'",
OPTION_MIN_WEIGHT_THRESHOLD, this.minWeightThreshold);
}
LOG.info("[RandomWalk] algorithm param, {}: {}",
OPTION_MIN_WEIGHT_THRESHOLD, this.minWeightThreshold);

this.maxWeightThreshold = config.getDouble(OPTION_MAX_WEIGHT_THRESHOLD, Double.MAX_VALUE);
if (this.maxWeightThreshold < 0) {
throw new ComputerException("The param %s must be greater than or equal 0, " +
"actual got '%s'",
OPTION_MAX_WEIGHT_THRESHOLD, this.maxWeightThreshold);
}
LOG.info("[RandomWalk] algorithm param, {}: {}",
OPTION_MAX_WEIGHT_THRESHOLD, this.maxWeightThreshold);

if (this.minWeightThreshold > this.maxWeightThreshold) {
throw new ComputerException("%s must be greater than or equal %s, ",
Expand All @@ -174,17 +164,13 @@ public void init(Config config) {
"actual got '%s'",
OPTION_RETURN_FACTOR, this.returnFactor);
}
LOG.info("[RandomWalk] algorithm param, {}: {}",
OPTION_RETURN_FACTOR, this.returnFactor);

this.inOutFactor = config.getDouble(OPTION_INOUT_FACTOR, 1);
if (this.inOutFactor <= 0) {
throw new ComputerException("The param %s must be greater than 0, " +
"actual got '%s'",
OPTION_INOUT_FACTOR, this.inOutFactor);
}
LOG.info("[RandomWalk] algorithm param, {}: {}",
OPTION_INOUT_FACTOR, this.inOutFactor);
}

@Override
Expand Down Expand Up @@ -269,10 +255,10 @@ private Edge randomSelectEdge(Id preVertexId, IdList preVertexAdjacenceIdList, E
Iterator<Edge> iterator = edges.iterator();
while (iterator.hasNext()) {
Edge edge = iterator.next();
// calculate weight
Value weight = this.getWeight(edge);
Double finalWeight = this.calculateWeight(preVertexId, preVertexAdjacenceIdList,
edge.targetId(), weight);
// calculate edge weight
double weight = this.getEdgeWeight(edge);
Double finalWeight = this.calculateEdgeWeight(preVertexId, preVertexAdjacenceIdList,
edge.targetId(), weight);
weightList.add(finalWeight);
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

can we add a mark here: TODO: improve to avoid OOM

}

Expand All @@ -282,35 +268,36 @@ private Edge randomSelectEdge(Id preVertexId, IdList preVertexAdjacenceIdList, E
}

/**
* get edge weight by weight property
* get the weight of an edge by its weight property
*/
private Value getWeight(Edge edge) {
Value weight = edge.property(this.weightProperty);
if (weight == null) {
weight = new DoubleValue(this.defaultWeight);
private double getEdgeWeight(Edge edge) {
Value property = edge.property(this.weightProperty);
if (property == null) {
property = new DoubleValue(this.defaultWeight);
}

if (!weight.isNumber()) {
if (!property.isNumber()) {
throw new ComputerException("The value of %s must be a numeric value, " +
"actual got '%s'",
this.weightProperty, weight.string());
this.weightProperty, property.string());
}

// weight threshold truncation
if ((Double) weight.value() < this.minWeightThreshold) {
DoubleValue weight = (DoubleValue) property;
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

based on a double value like:

  1. double weight = this.defaultWeight;
  2. weight = property..doubleValue() if checked ok
  3. do truncation
  4. return weight

if (weight.doubleValue() < this.minWeightThreshold) {
weight = new DoubleValue(this.minWeightThreshold);
}
if ((Double) weight.value() > this.maxWeightThreshold) {
if (weight.doubleValue() > this.maxWeightThreshold) {
weight = new DoubleValue(this.maxWeightThreshold);
}
return weight;
return weight.doubleValue();
}

/**
* calculate edge weight
*/
private Double calculateWeight(Id preVertexId, IdList preVertexAdjacenceIdList,
Id nextVertexId, Value weight) {
private Double calculateEdgeWeight(Id preVertexId, IdList preVertexAdjacenceIdList,
Id nextVertexId, double weight) {
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

also keep double finalWeight and return double?

/*
* 3 types of vertices.
* 1. current vertex, called v
Expand All @@ -327,14 +314,14 @@ private Double calculateWeight(Id preVertexId, IdList preVertexAdjacenceIdList,
Double finalWeight = 0.0;
if (preVertexId != null && preVertexId.equals(nextVertexId)) {
// distance(t, x) = 0
finalWeight = 1.0 / this.returnFactor * (Double) weight.value();
finalWeight = 1.0 / this.returnFactor * weight;
} else if (preVertexAdjacenceIdList != null &&
preVertexAdjacenceIdList.contains(nextVertexId)) {
// distance(t, x) = 1
finalWeight = 1.0 * (Double) weight.value();
finalWeight = 1.0 * weight;
} else {
// distance(t, x) = 2
finalWeight = 1.0 / this.inOutFactor * (Double) weight.value();
finalWeight = 1.0 / this.inOutFactor * weight;
}
return finalWeight;
}
Expand Down