欢迎您访问程序员文章站本站旨在为大家提供分享程序员计算机编程知识!
您现在的位置是: 首页

知识图谱3

程序员文章站 2022-06-12 17:05:08
...

一、引言

Neo4j 是当前较为主流和先进的原生图数据库之一,提供原生的图数据存储、检索和处理。它由 Neo Technology支持,从 2003 年开始开发,1.0 版本发布于 2010 年,2.0版本发布于 2013 年。经过十多年的发展,Neo4j 获得越来越高的关注度,它已经从一个 Java 领域内的图数据库逐渐发展成为适应多语言多框架的图数据库。Neo4j 支持ACID、集群、备份和故障转移,具有较高的可用性和稳定性;它具备非常好的直观性,通过图形化的界面表示节点和关系;同时它具备较高的可扩展性,能够承载上亿的节点、关系和属性,通过 REST 接口或者面向对象的 JAVA API进行访问。

二、Neo4j简介

(1)基本概念

Neo4j使用图相关的概念来描述数据模型,把数据保存为图中的节点以及节点之间的关系。数据主要由三部分构成:

  • 节点。节点表示对象实例,每个节点有唯一的ID区别其它节点,节点带有属性;
  • 关系。就是图里面的边,连接两个节点,另外这里的关系是有向的并带有属性;
  • 属性。key-value对,存在于节点和关系中,如图1所示。
    知识图谱3

(2)索引

  • 动机:Neo4j使用遍历操作进行查询。为了加速查询,Neo4j会建立索引,并根据索引找到遍历用的起始节点;

  • 介绍:默认情况下,相关的索引是由Apache Lucene提供的。但也能使用其他索引实现来提供。

  • 操作:用户可以创建任意数量的命名索引。每个索引控制节点或者关系,而每个索引都通过key/value/object三个参数来工作。其中object要么是一个节点,要么是一个关系,取决于索引类型。另外,Neo4j中有关于节点(关系)的索引,系统通过索引实现从属性到节点(关系)的映射。

  • 作用:

    1. 查找操作:系统通过设定访问条件比如,遍历的方向,使用深度优先或广度优先算法等条件对图进行遍历,从一个节点沿着关系到其他节点;
    2. 删除操作:Neo4j可以快速的插入删除节点和关系,并更新节点和关系中的属性。

(3)Neo4j的优势

  • 查询的高性能
  • 设计的灵活性
  • 开发的敏捷行
  • 与其他数据库的比较
  • 综合表现
    1. 闪电般的读/写速度,无与伦比的高性能表现;
    2. 非结构化数据存储方式,在数据库设计上具有很大的灵活性;
    3. 能很好地适应需求变化,并适合使用敏捷开发方法;
    4. 很容易使用,可以用嵌入式、服务器模式、分布式模式等方式来使用数据库;
    5. 使用简单框图就可以设计数据模型,方便建模;
    6. 图数据的结构特点可以提供更多更优秀的算法设计;
    7. 完全支持ACID完整的事务管理特性;
    8. 提供分布式高可用模式,可以支持大规模的数据增长;
    9. 数据库安全可靠,可以实时备份数据,很方便恢复数据;
    10. 图的数据结构直观而形象地表现了现实世界的应用场景。

三、主体类MedicalGraph介绍

class MedicalGraph:
    def __init__(self):
        pass
    
    # 读取文件,获得实体,实体关系
    def read_file(self):
        psss
    # 创建节点
    def create_node(self, label, nodes):
        pass
    # 创建疾病节点的属性
    def create_diseases_nodes(self, disease_info):
        pass
    # 创建知识图谱实体
    def create_graphNodes(self):
        pass
    # 创建实体关系边
    def create_graphRels(self):
        pass
    # 创建实体关系边
    def create_relationship(self, start_node, end_node, edges, rel_type, rel_name):
        pass
  1. 读取文件
def read_file(self):
        """
        读取文件,获得实体,实体关系
        :return:
        """
        # cols = ["name", "alias", "part", "age", "infection", "insurance", "department", "checklist", "symptom",
        #         "complication", "treatment", "drug", "period", "rate", "money"]
        # 实体
        diseases = []  # 疾病
        aliases = []  # 别名
        symptoms = []  # 症状
        parts = []  # 部位
        departments = []  # 科室
        complications = []  # 并发症
        drugs = []  # 药品

        # 疾病的属性:age, infection, insurance, checklist, treatment, period, rate, money
        diseases_infos = []
        # 关系
        disease_to_symptom = []  # 疾病与症状关系
        disease_to_alias = []  # 疾病与别名关系
        diseases_to_part = []  # 疾病与部位关系
        disease_to_department = []  # 疾病与科室关系
        disease_to_complication = []  # 疾病与并发症关系
        disease_to_drug = []  # 疾病与药品关系

        all_data = pd.read_csv(self.data_path, encoding='gb18030').loc[:, :].values
        for data in all_data:
            disease_dict = {}  # 疾病信息
            # 疾病
            disease = str(data[0]).replace("...", " ").strip()
            disease_dict["name"] = disease
            # 别名
            line = re.sub("[,、;,.;]", " ", str(data[1])) if str(data[1]) else "未知"
            for alias in line.strip().split():
                aliases.append(alias)
                disease_to_alias.append([disease, alias])
            # 部位
            part_list = str(data[2]).strip().split() if str(data[2]) else "未知"
            for part in part_list:
                parts.append(part)
                diseases_to_part.append([disease, part])
            # 年龄
            age = str(data[3]).strip()
            disease_dict["age"] = age
            # 传染性
            infect = str(data[4]).strip()
            disease_dict["infection"] = infect
            # 医保
            insurance = str(data[5]).strip()
            disease_dict["insurance"] = insurance
            # 科室
            department_list = str(data[6]).strip().split()
            for department in department_list:
                departments.append(department)
                disease_to_department.append([disease, department])
            # 检查项
            check = str(data[7]).strip()
            disease_dict["checklist"] = check
            # 症状
            symptom_list = str(data[8]).replace("...", " ").strip().split()[:-1]
            for symptom in symptom_list:
                symptoms.append(symptom)
                disease_to_symptom.append([disease, symptom])
            # 并发症
            complication_list = str(data[9]).strip().split()[:-1] if str(data[9]) else "未知"
            for complication in complication_list:
                complications.append(complication)
                disease_to_complication.append([disease, complication])
            # 治疗方法
            treat = str(data[10]).strip()[:-4]
            disease_dict["treatment"] = treat
            # 药品
            drug_string = str(data[11]).replace("...", " ").strip()
            for drug in drug_string.split()[:-1]:
                drugs.append(drug)
                disease_to_drug.append([disease, drug])
            # 治愈周期
            period = str(data[12]).strip()
            disease_dict["period"] = period
            # 治愈率
            rate = str(data[13]).strip()
            disease_dict["rate"] = rate
            # 费用
            money = str(data[14]).strip() if str(data[14]) else "未知"
            disease_dict["money"] = money

            diseases_infos.append(disease_dict)

        return set(diseases), set(symptoms), set(aliases), set(parts), set(departments), set(complications), \
                set(drugs), disease_to_alias, disease_to_symptom, diseases_to_part, disease_to_department, \
                disease_to_complication, disease_to_drug, diseases_infos
  1. 创建节点
def create_node(self, label, nodes):
        """
        创建节点
        :param label: 标签
        :param nodes: 节点
        :return:
        """
        count = 0
        for node_name in nodes:
            node = Node(label, name=node_name)
            self.graph.create(node)
            count += 1
            print(count, len(nodes))
        return
  1. 创建带有属性节点
def create_diseases_nodes(self, disease_info):
        """
        创建疾病节点的属性
        :param disease_info: list(Dict)
        :return:
        """
        count = 0
        for disease_dict in disease_info:
            node = Node("Disease", name=disease_dict['name'], age=disease_dict['age'],
                        infection=disease_dict['infection'], insurance=disease_dict['insurance'],
                        treatment=disease_dict['treatment'], checklist=disease_dict['checklist'],
                        period=disease_dict['period'], rate=disease_dict['rate'],
                        money=disease_dict['money'])
            self.graph.create(node)
            count += 1
            print(count)
        return
  1. 创建知识图谱实体
def create_graphNodes(self):
        """
        创建知识图谱实体
        :return:
        """
        disease, symptom, alias, part, department, complication, drug, rel_alias, rel_symptom, rel_part, \
        rel_department, rel_complication, rel_drug, rel_infos = self.read_file()
        self.create_diseases_nodes(rel_infos)
        self.create_node("Symptom", symptom)
        self.create_node("Alias", alias)
        self.create_node("Part", part)
        self.create_node("Department", department)
        self.create_node("Complication", complication)
        self.create_node("Drug", drug)

        return
  1. 创建知识图谱关系
def create_graphRels(self):
        disease, symptom, alias, part, department, complication, drug, rel_alias, rel_symptom, rel_part, \
        rel_department, rel_complication, rel_drug, rel_infos = self.read_file()

        self.create_relationship("Disease", "Alias", rel_alias, "ALIAS_IS", "别名")
        self.create_relationship("Disease", "Symptom", rel_symptom, "HAS_SYMPTOM", "症状")
        self.create_relationship("Disease", "Part", rel_part, "PART_IS", "发病部位")
        self.create_relationship("Disease", "Department", rel_department, "DEPARTMENT_IS", "所属科室")
        self.create_relationship("Disease", "Complication", rel_complication, "HAS_COMPLICATION", "并发症")
        self.create_relationship("Disease", "Drug", rel_drug, "HAS_DRUG", "药品")
  1. 创建实体关系边
def create_relationship(self, start_node, end_node, edges, rel_type, rel_name):
        """
        创建实体关系边
        :param start_node:
        :param end_node:
        :param edges:
        :param rel_type:
        :param rel_name:
        :return:
        """
        count = 0
        # 去重处理
        set_edges = []
        for edge in edges:
            set_edges.append('###'.join(edge))
        all = len(set(set_edges))
        for edge in set(set_edges):
            edge = edge.split('###')
            p = edge[0]
            q = edge[1]
            query = "match(p:%s),(q:%s) where p.name='%s'and q.name='%s' create (p)-[rel:%s{name:'%s'}]->(q)" % (
                start_node, end_node, p, q, rel_type, rel_name)
            try:
                self.graph.run(query)
                count += 1
                print(rel_type, count, all)
            except Exception as e:
                print(e)
        return
  • 参考资料
  1. QASystemOnMedicalGraph
  2. 韩浩明.图数据库系统研究综述[J].计算机光盘软件与应用,2014,17(23):14-