練習Neo4j

之前都是自己偶而玩玩Neo4j，趁假日把上班遇到的問題改用Neo4j來解決。
先做一個簡單的資料

CREATE (ACME:DeptGroup:OrgGroup {id: 'Ga', name: 'ACME'})
CREATE (ACME_Member:UserGroup {id: 'Gb', name: 'allmember'})
CREATE (Keanu:Person {id: 'Ua', name: 'Keanu Reeves'})

CREATE (Dev1:DeptGroup:OrgGroup {id: 'Gc', name: 'Development 1'})
CREATE (Dev1_Mmember:UserGroup {id: 'Ge', name: 'allmember'})

CREATE (Project_Lineage:ProjectGroup:OrgGroup {id:'Gg', name: 'Project Lineage'})
CREATE (Lineage_Mmember:UserGroup {id: 'Gj', name: 'allmember'})

CREATE (Unity_ResearchGroup:ProjectGroup:OrgGroup {id:'Gh', name: 'Unity Research Group'})
CREATE (Unity_Mmember:UserGroup {id: 'Gk', name: 'allmember'})

CREATE (Dev2:DeptGroup:OrgGroup {id: 'Gd', name: 'Development 2'})
CREATE (Dev2_Mmember:UserGroup {id: 'Gl', name: 'allmember'})

CREATE (Project_X:ProjectGroup:OrgGroup {id: 'Gi', name: 'Social Game Project X'})
CREATE (X_Mmember:UserGroup {id: 'Gm', name: 'allmember'})

CREATE
  (Dev1)-[:SUPERVISED_BY]->(ACME),
  (Dev2)-[:SUPERVISED_BY]->(ACME),
  (Project_Lineage)-[:SUPERVISED_BY]->(Dev1),
  (Unity_ResearchGroup)-[:SUPERVISED_BY]->(Dev1),
  (Project_X)-[:SUPERVISED_BY]->(Dev2),
  (ACME_Member)-[:BELONG_TO]->(ACME),
  (Dev1_Member)-[:BELONG_TO]->(Dev1),
  (Dev2_Member)-[:BELONG_TO]->(Dev2),
  (Lineage_Mmember)-[:BELONE_TO]->(Project_Lineage),
  (Unity_Mmember)-[:BELONE_TO]->(Unity_ResearchGroup),
  (X_Mmember)-[:BELONE_TO]->(Project_X),
  (Dev1_Mmember)-[:SHARE_TO]->(ACME_Member),
  (Dev2_Mmember)-[:SHARE_TO]->(ACME_Member),
  (Lineage_Mmember)-[:SHARE_TO]->(Dev1_Mmember),
  (Unity_Mmember)-[:SHARE_TO]->(Dev1_Mmember),
  (X_Mmember)-[:SHARE_TO]->(Dev2_Mmember),
  (Lineage_Mmember)-[:SHARE_TO]->(Dev1_Mmember),
  (Unity_Mmember)-[:SHARE_TO]->(Dev1_Mmember),
  (X_Mmember)-[:SHARE_TO]->(Dev2_Mmember),
  (Keanu)-[:MEMBER_OF]->(ACME_Member)

目的透過成員查詢到所屬UserGroup的同時，把有設定[:SHARE_TO]的relation的多層關係的UserGroups都查詢出來。

match (p:Person {id:'Ua'})-[:MEMBER_OF]->(u:UserGroup)
return u.id as gid
union
match (udeep:UserGroup)-[:SHARE_TO*]->(u)
return udeep.id as gid

之前是自己實作adjacency list解決速度的問題，看來也可以輕鬆地改用Neo4j搞定。