feat: label propagation algorithm

__tests__/unit/label-propagation.spec.ts

@@ -0,0 +1,54 @@
+import { Graph } from "@antv/graphlib";
+import { labelPropagation } from '../../packages/graph/src';
+import { dataTransformer } from "../utils/data";
+import labelPropagationTestData from '../data/label-propagation-test-data.json';
+
+
+describe('label propagation', () => {
+    it('simple label propagation', () => {
+        const oldData = {
+            nodes: [
+                { id: '0' }, { id: '1' }, { id: '2' }, { id: '3' }, { id: '4' },
+                { id: '5' }, { id: '6' }, { id: '7' }, { id: '8' }, { id: '9' },
+                { id: '10' }, { id: '11' }, { id: '12' }, { id: '13' }, { id: '14' },
+            ],
+            edges: [
+                { source: '0', target: '1' }, { source: '0', target: '2' }, { source: '0', target: '3' }, { source: '0', target: '4' },
+                { source: '1', target: '2' }, { source: '1', target: '3' }, { source: '1', target: '4' },
+                { source: '2', target: '3' }, { source: '2', target: '4' },
+                { source: '3', target: '4' },
+                { source: '0', target: '0' },
+                { source: '0', target: '0' },
+                { source: '0', target: '0' },
+
+                { source: '5', target: '6', weight: 5 }, { source: '5', target: '7' }, { source: '5', target: '8' }, { source: '5', target: '9' },
+                { source: '6', target: '7' }, { source: '6', target: '8' }, { source: '6', target: '9' },
+                { source: '7', target: '8' }, { source: '7', target: '9' },
+                { source: '8', target: '9' },
+
+                { source: '10', target: '11' }, { source: '10', target: '12' }, { source: '10', target: '13' }, { source: '10', target: '14' },
+                { source: '11', target: '12' }, { source: '11', target: '13' }, { source: '11', target: '14' },
+                { source: '12', target: '13' }, { source: '12', target: '14' },
+                { source: '13', target: '14', weight: 5 },
+
+                { source: '0', target: '5' },
+                { source: '5', target: '10' },
+                { source: '10', target: '0' },
+                { source: '10', target: '0' },
+            ]
+        };
+        const data = dataTransformer(oldData);
+        const graph = new Graph(data);
+        const clusteredData = labelPropagation(graph, false, 'weight');
+        expect(clusteredData.clusters.length).not.toBe(0);
+        expect(clusteredData.clusterEdges.length).not.toBe(0);
+    });
+
+    it('label propagation with large graph', () => {
+        const data = dataTransformer(labelPropagationTestData);
+        const graph = new Graph(data);
+        const clusteredData = labelPropagation(graph, false, 'weight');
+        expect(clusteredData.clusters.length).not.toBe(0);
+        expect(clusteredData.clusterEdges.length).not.toBe(0);
+    }
+});

__tests__/utils/data.ts

@@ -6,8 +6,8 @@ import { INode, IEdge } from '../../packages/graph/src/types';
  * @return {{nodes:INode[],edges:IEdge[]}} new data
  */
 export const dataTransformer = (data: {
-  nodes: { id: ID; [key: string]: any }[];
-  edges: { source: ID; target: ID; [key: string]: any }[];
+  nodes: { id: ID;[key: string]: any }[];
+  edges: { source: ID; target: ID;[key: string]: any }[];
 }): { nodes: INode[]; edges: IEdge[] } => {
   const { nodes, edges } = data;
   return {
@@ -22,31 +22,31 @@ export const dataTransformer = (data: {
   };
 };
 
-export const dataPropertiesTransformer = (data: { nodes: { id: NodeID, [key: string]: any }[], edges: { source: NodeID, target: NodeID, [key: string]: any }[] }): { nodes: INode[], edges: IEdge[] } => {
-    const { nodes, edges } = data;
-    return {
-        nodes: nodes.map((n) => {
-            const { id, properties, ...rest } = n;
-            return { id, data: { ...properties, ...rest } };
-        }),
-        edges: edges.map((e, i) => {
-            const { id, source, target, ...rest } = e;
-            return { id: id ? id : `edge-${i}`, target, source, data: rest };
-        }),
-    };
+export const dataPropertiesTransformer = (data: { nodes: { id: ID, [key: string]: any }[], edges: { source: ID, target: ID, [key: string]: any }[] }): { nodes: INode[], edges: IEdge[] } => {
+  const { nodes, edges } = data;
+  return {
+    nodes: nodes.map((n) => {
+      const { id, properties, ...rest } = n;
+      return { id, data: { ...properties, ...rest } };
+    }),
+    edges: edges.map((e, i) => {
+      const { id, source, target, ...rest } = e;
+      return { id: id ? id : `edge-${i}`, target, source, data: rest };
+    }),
+  };
 };
 
 
-export const dataLabelDataTransformer = (data: { nodes: { id: NodeID, [key: string]: any }[], edges: { source: NodeID, target: NodeID, [key: string]: any }[] }): { nodes: INode[], edges: IEdge[] } => {
-    const { nodes, edges } = data;
-    return {
-        nodes: nodes.map((n) => {
-            const { id, label, data } = n;
-            return { id, data: { label, ...data } };
-        }),
-        edges: edges.map((e, i) => {
-            const { id, source, target, ...rest } = e;
-            return { id: id ? id : `edge-${i}`, target, source, data: rest };
-        }),
-    };
+export const dataLabelDataTransformer = (data: { nodes: { id: ID, [key: string]: any }[], edges: { source: ID, target: ID, [key: string]: any }[] }): { nodes: INode[], edges: IEdge[] } => {
+  const { nodes, edges } = data;
+  return {
+    nodes: nodes.map((n) => {
+      const { id, label, data } = n;
+      return { id, data: { label, ...data } };
+    }),
+    edges: edges.map((e, i) => {
+      const { id, source, target, ...rest } = e;
+      return { id: id ? id : `edge-${i}`, target, source, data: rest };
+    }),
+  };
 };

package.json

@@ -21,7 +21,7 @@
     "build:ci": "pnpm -r run build:ci",
     "prepare": "husky install",
     "test": "jest",
-    "test_one": "jest ./__tests__/unit/k-means.spec.ts",
+    "test_one": "jest ./__tests__/unit/label-propagation.spec.ts",
     "coverage": "jest --coverage",
     "build:site": "vite build",
     "deploy": "gh-pages -d site/dist",

packages/graph/src/detect-cycle.ts

@@ -62,7 +62,7 @@ export const detectDirectedCycle = (
       return true;
     },
   };
-  for (let key of Object.keys(unvisitedSet)) {
+  for (const key of Object.keys(unvisitedSet)) {
     depthFirstSearch(graph, key, callbacks, true, false);
   }
   return cycle;
@@ -193,8 +193,9 @@ export const detectAllDirectedCycle = (
     adjList: { [key: ID]: number[] }
   ) => {
     let closed = false; // whether a path is closed
-    if (nodeIds && include === false && nodeIds.indexOf(node.id) > -1)
+    if (nodeIds && !include && nodeIds.indexOf(node.id) > -1) {
       return closed;
+    }
     path.push(node);
     blocked.add(node);
     const neighbors = adjList[node.id];
@@ -277,7 +278,7 @@ export const detectAllDirectedCycle = (
         // 对自环情况 (点连向自身) 特殊处理：记录自环，但不加入adjList
         if (
           neighbor === node.id &&
-          !(include === false && nodeIds.indexOf(node.id) > -1)
+          !(!include && nodeIds.indexOf(node.id) > -1)
         ) {
           allCycles.push({ [node.id]: node });
         } else {
@@ -306,8 +307,9 @@ export const detectAllDirectedCycle = (
       });
       const startNode = idx2Node[minIdx];
       // StartNode is not in the specified node to include. End the search ahead of time.
-      if (nodeIds && include && nodeIds.indexOf(startNode.id) === -1)
+      if (nodeIds && include && nodeIds.indexOf(startNode.id) === -1) {
         return allCycles;
+      }
       circuit(startNode, startNode, adjList);
       nodeIdx = minIdx + 1;
     } else {

packages/graph/src/index.ts

@@ -13,3 +13,4 @@ export * from './connected-component';
 export * from './mst';
 export * from './k-means';
 export * from './detect-cycle';
+export * from './label-propagation';

packages/graph/src/label-propagation.ts

@@ -0,0 +1,197 @@
+import { uniqueId } from "./utils";
+import { ClusterData, INode, IEdge, Graph, Matrix } from "./types";
+import { ID } from "@antv/graphlib";
+
+function getAdjMatrix(graph: Graph, directed: boolean) {
+  const nodes = graph.getAllNodes();
+  const matrix: Matrix[] = [];
+  // map node with index in data.nodes
+  const nodeMap = new Map<string | number, number>();
+
+  if (!nodes) {
+    throw new Error("invalid nodes data!");
+  }
+
+  if (nodes) {
+    nodes.forEach((node, i) => {
+      nodeMap.set(node.id, i);
+      const row: number[] = [];
+      matrix.push(row);
+    });
+  }
+
+  const edges = graph.getAllEdges();
+  if (edges) {
+    edges.forEach((edge) => {
+      const { source, target } = edge;
+      const sIndex = nodeMap.get(source);
+      const tIndex = nodeMap.get(target);
+      if ((!sIndex && sIndex !== 0) || (!tIndex && tIndex !== 0)) return;
+      matrix[sIndex][tIndex] = 1;
+      if (!directed) {
+        matrix[tIndex][sIndex] = 1;
+      }
+    });
+  }
+  return matrix;
+}
+
+/**
+ * 标签传播算法
+ * @param graphData 图数据
+ * @param directed 是否有向图，默认为 false
+ * @param weightPropertyName 权重的属性字段
+ * @param maxIteration 最大迭代次数
+ */
+export const labelPropagation = (
+  graph: Graph,
+  directed: boolean = false,
+  weightPropertyName: string = "weight",
+  maxIteration: number = 1000
+): ClusterData => {
+  // the origin data
+  const nodes = graph.getAllNodes();
+  const edges = graph.getAllEdges();
+
+  const clusters: { [key: string]: { id: string; nodes: INode[] } } = {};
+  const nodeMap: { [key: ID]: { node: INode; idx: number } } = {};
+  const nodeToCluster = new Map<ID, string>();
+  // init the clusters and nodeMap
+  nodes.forEach((node, i) => {
+    const cid: string = uniqueId();
+    nodeToCluster.set(node.id, cid);
+    clusters[cid] = {
+      id: cid,
+      nodes: [node],
+    };
+    nodeMap[node.id] = {
+      node,
+      idx: i,
+    };
+  });
+
+  // the adjacent matrix of calNodes inside clusters
+  const adjMatrix = getAdjMatrix(graph, directed);
+  // the sum of each row in adjacent matrix
+  const ks = [];
+  /**
+   * neighbor nodes (id for key and weight for value) for each node
+   * neighbors = {
+   *  id(node_id): { id(neighbor_1_id): weight(weight of the edge), id(neighbor_2_id): weight(weight of the edge), ... },
+   *  ...
+   * }
+   */
+  const neighbors: { [key: ID]: { [key: ID]: number } } = {};
+  adjMatrix.forEach((row, i) => {
+    let k = 0;
+    const iid = nodes[i].id;
+    neighbors[iid] = {};
+    row.forEach((entry, j) => {
+      if (!entry) return;
+      k += entry;
+      const jid = nodes[j].id;
+      neighbors[iid][jid] = entry;
+    });
+    ks.push(k);
+  });
+
+  let iter = 0;
+
+  while (iter < maxIteration) {
+    let changed = false;
+    nodes.forEach((node) => {
+      const neighborClusters: { [key: string]: number } = {};
+      Object.keys(neighbors[node.id]).forEach((neighborId) => {
+        const neighborWeight = neighbors[node.id][neighborId];
+        const neighborNode = nodeMap[neighborId].node;
+
+        const neighborClusterId = nodeToCluster.get(neighborNode.id);
+        if (!neighborClusters[neighborClusterId]) {
+          neighborClusters[neighborClusterId] = 0;
+        }
+        neighborClusters[neighborClusterId] += neighborWeight;
+      });
+      // find the cluster with max weight
+      let maxWeight = -Infinity;
+      let bestClusterIds: string[] = [];
+      Object.keys(neighborClusters).forEach((clusterId) => {
+        if (maxWeight < neighborClusters[clusterId]) {
+          maxWeight = neighborClusters[clusterId];
+          bestClusterIds = [clusterId];
+        } else if (maxWeight === neighborClusters[clusterId]) {
+          bestClusterIds.push(clusterId);
+        }
+      });
+      if (
+        bestClusterIds.length === 1 &&
+        bestClusterIds[0] === nodeToCluster.get(node.id)
+      ) {
+        return;
+      }
+      const selfClusterIdx = bestClusterIds.indexOf(nodeToCluster.get(node.id));
+      if (selfClusterIdx >= 0) bestClusterIds.splice(selfClusterIdx, 1);
+      if (bestClusterIds && bestClusterIds.length) {
+        changed = true;
+
+        // remove from origin cluster
+        const selfCluster = clusters[nodeToCluster.get(node.id)];
+        const nodeInSelfClusterIdx = selfCluster.nodes.indexOf(node);
+        selfCluster.nodes.splice(nodeInSelfClusterIdx, 1);
+
+        // move the node to the best cluster
+        const randomIdx = Math.floor(Math.random() * bestClusterIds.length);
+        const bestCluster = clusters[bestClusterIds[randomIdx]];
+        bestCluster.nodes.push(node);
+        nodeToCluster.set(node.id, bestCluster.id);
+      }
+    });
+    if (!changed) break;
+    iter++;
+  }
+
+  // delete the empty clusters
+  Object.keys(clusters).forEach((clusterId) => {
+    const cluster = clusters[clusterId];
+    if (!cluster.nodes || !cluster.nodes.length) {
+      delete clusters[clusterId];
+    }
+  });
+
+  // get the cluster edges
+  const clusterEdges: IEdge[] = [];
+  const clusterEdgeMap: { [key: string]: IEdge } = {};
+  edges.forEach((edge) => {
+    let i = 0;
+    const { source, target } = edge;
+    const weight = (edge.data[weightPropertyName] || 1) as number;
+    const sourceClusterId = nodeToCluster.get(nodeMap[source].node.id);
+    const targetClusterId = nodeToCluster.get(nodeMap[target].node.id);
+    const newEdgeId = `${sourceClusterId}---${targetClusterId}`;
+    if (clusterEdgeMap[newEdgeId]) {
+      clusterEdgeMap[newEdgeId].data.weight += weight;
+      (clusterEdgeMap[newEdgeId].data.count as number)++;
+    } else {
+      const newEdge = {
+        id: i++,
+        source: sourceClusterId,
+        target: targetClusterId,
+        data: {
+          weight,
+          count: 1,
+        },
+      };
+      clusterEdgeMap[newEdgeId] = newEdge;
+      clusterEdges.push(newEdge);
+    }
+  });
+
+  const clustersArray: { id: string; nodes: INode[] }[] = [];
+  Object.keys(clusters).forEach((clusterId) => {
+    clustersArray.push(clusters[clusterId]);
+  });
+  return {
+    clusters: clustersArray,
+    clusterEdges,
+    nodeToCluster,
+  };
+};

packages/graph/src/utils.ts

@@ -1,5 +1,4 @@
 import { Node, PlainObject } from "@antv/graphlib";
-import { Vector } from "./vector";
 import { DistanceType, KeyValueMap, NodeData } from "./types";
 import { uniq } from "@antv/util";
 
@@ -101,3 +100,10 @@ function euclideanDistance(source: number[], target: number[]) {
   return Math.sqrt(res);
 }
 
+export const uniqueId = (index: number = 0) => {
+  const random1 = `${Math.random()}`.split('.')[1].slice(0, 5);
+  const random2 = `${Math.random()}`.split('.')[1].slice(0, 5);
+  return `${index}-${random1}${random2}`;
+};
+
+