我有一个图表,其中“User”类型的节点可以使用“SIMILAR”关系连接(或不连接)到其他“User”节点。
我想找到按节点数排序的所有子图(互连的“用户”节点)。
实施例: 想象一下,我有这些节点(关系总是类型为“SIMILAR”)
A -> B,
A -> C,
D -> B,
E -> F,
Z
我想得到:
[A,B,C,D];
[E,F];
[Z];
使用Cypher或遍历。
答案 0 :(得分:4)
在密码中它会很昂贵。
像这样的查询:
MATCH m WITH collect(m) as all
MATCH n
RETURN distinct [x in all WHERE (n)-[*0..]-(x) | x.name] as cluster
这也可以,但可能同样昂贵:)
MATCH n-[*0..]-m
WITH n, m
ORDER BY id(m)
WITH n, collect(m.name) as cluster
RETURN distinct cluster
在Java中它看起来像这样:
@Test
public void testSimpleCluster() throws Exception {
createData();
try (Transaction tx = db.beginTx()) {
TraversalDescription traversal = db.traversalDescription().depthFirst().uniqueness(Uniqueness.NODE_GLOBAL);
Map<Node, Set<Node>> clusters = new HashMap<>();
GlobalGraphOperations ops = GlobalGraphOperations.at(db);
for (Node node : ops.getAllNodes()) {
if (inCluster(node, clusters) != null) continue;
clusters.put(node, IteratorUtil.addToCollection(traversal.traverse(node).nodes(), new HashSet<Node>()));
}
System.out.println("clusters = " + clusters.values());
tx.success();
}
}
private Node inCluster(Node node, Map<Node, Set<Node>> clusters) {
for (Map.Entry<Node, Set<Node>> entry : clusters.entrySet()) {
if (entry.getValue().contains(node)) return entry.getKey();
}
return null;
}
private void createData() {
try (Transaction tx = db.beginTx()) {
Node a = node("a");
Node b = node("b");
Node c = node("c");
Node d = node("d");
Node e = node("e");
Node f = node("f");
Node z = node("z");
connect(a, b);
connect(a, c);
connect(d, b);
connect(e, f);
tx.success();
}
}
private void connect(Node a, Node b) {
a.createRelationshipTo(b, SIMILAR);
}
private Node node(String name) {
Node node = db.createNode();
node.setProperty("name", name);
return node;
}