图数据库架构

属性图模型

属性图由节点、边和属性组成，适合表达实体间的复杂关系。

(Alice)-[:FRIEND]->(Bob)-[:WORKS_AT]->(Company)
(Alice)-[:LIKES]->(Product)<-[:SELLS]-(Company)

@Configuration
public class Neo4jConfig {

    @Bean
    public Driver neo4jDriver() {
        return GraphDatabase.driver(
            "bolt://localhost:7687",
            AuthTokens.basic("neo4j", "password"));
    }
}

Cypher查询语言

Cypher是Neo4j的声明式查询语言，用模式匹配表达图查询。

@Repository
public interface PersonRepository extends Neo4jRepository<Person, Long> {

    @Query("MATCH (p:Person)-[:FRIEND]->(f:Person) WHERE p.name = $name RETURN f")
    List<Person> findFriendsByName(@Param("name") String name);

    @Query("MATCH (p:Person)-[:FRIEND*1..3]->(f:Person) WHERE p.name = $name RETURN DISTINCT f")
    List<Person> findFriendsWithinThreeHops(@Param("name") String name);
}

// 创建节点和关系
CREATE (alice:Person {name: 'Alice', age: 30})
CREATE (bob:Person {name: 'Bob', age: 25})
CREATE (alice)-[:FRIEND {since: 2020}]->(bob)

// 查找最短路径
MATCH path = shortestPath(
  (a:Person {name: 'Alice'})-[*]-(b:Person {name: 'Charlie'})
)
RETURN path

// 社区发现算法
CALL gds.louvain.stream('social')
YIELD nodeId, communityId
RETURN gds.util.asNode(nodeId).name AS name, communityId
ORDER BY communityId

图算法

Neo4j提供丰富的图算法库，包括中心性、社区发现、路径查找等。

@Service
public class GraphAlgorithmService {

    @Autowired
    private Driver driver;

    // PageRank算法
    public List<Map<String, Object>> pageRank() {
        try (Session session = driver.session()) {
            return session.run(
                "CALL gds.pageRank.stream('social') " +
                "YIELD nodeId, score " +
                "RETURN gds.util.asNode(nodeId).name AS name, score " +
                "ORDER BY score DESC LIMIT 10"
            ).list(record -> Map.of(
                "name", record.get("name").asString(),
                "score", record.get("score").asDouble()
            ));
        }
    }
}

知识图谱应用

构建知识图谱实现智能问答、推荐系统和风控分析。

@Service
public class KnowledgeGraphService {

    public List<String> findRelatedEntities(String entity) {
        try (Session session = driver.session()) {
            return session.run(
                "MATCH (e:Entity {name: $entity})-[r]-(related) " +
                "RETURN type(r) AS relation, labels(related)[0] AS type, related.name AS name",
                Parameters.parameters("entity", entity)
            ).list(record -> String.format("%s --[%s]--> %s(%s)",
                entity,
                record.get("relation").asString(),
                record.get("name").asString(),
                record.get("type").asString()
            ));
        }
    }
}