WO2020056984A1 - 最短路径查询方法、***、计算机设备和存储介质 - Google Patents

最短路径查询方法、***、计算机设备和存储介质 Download PDF

Info

Publication number
WO2020056984A1
WO2020056984A1 PCT/CN2018/124237 CN2018124237W WO2020056984A1 WO 2020056984 A1 WO2020056984 A1 WO 2020056984A1 CN 2018124237 W CN2018124237 W CN 2018124237W WO 2020056984 A1 WO2020056984 A1 WO 2020056984A1
Authority
WO
WIPO (PCT)
Prior art keywords
shortest path
fund
entities
knowledge
metabase
Prior art date
Application number
PCT/CN2018/124237
Other languages
English (en)
French (fr)
Inventor
陈泽晖
胡逸凡
谢云
黄鸿顺
Original Assignee
平安科技(深圳)有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 平安科技(深圳)有限公司 filed Critical 平安科技(深圳)有限公司
Publication of WO2020056984A1 publication Critical patent/WO2020056984A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06QINFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q40/00Finance; Insurance; Tax strategies; Processing of corporate or income taxes
    • G06Q40/06Asset management; Financial planning or analysis
    • YGENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02DCLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D10/00Energy efficient computing, e.g. low power processors, power management or thermal management

Definitions

  • the present application relates to the field of graph theory and network analysis technology, and in particular, to a shortest path query method, system, computer design, and storage medium.
  • Knowledge graph is essentially a semantic network, which is a graph-based data structure composed of nodes and edges.
  • each node represents the "entity” existing in the real world, and each edge is the “relationship” between the entity and the entity.
  • the knowledge graph is the most effective way to represent relationships.
  • the knowledge graph is a network of relationships that connects all different kinds of information together.
  • Knowledge graphs provide the ability to analyze problems from the perspective of "relationships”.
  • Knowledge reasoning can be understood as "link prediction", that is, new relationships or links are derived from existing relational graphs.
  • Common reasoning algorithms include based on Logical reasoning and reasoning based on distributed representation methods.
  • Knowledge graphs can be combined with multiple data sources to analyze the relationships between entities to better understand user behavior.
  • many data are unprocessed unstructured data, such as text, pictures , Audio, video, etc.
  • unprocessed unstructured data such as text, pictures , Audio, video, etc.
  • a shortest path query method includes: establishing a fund knowledge metabase after acquiring fund knowledge in a fund information source, the fund knowledge metabase is a full map, and the full map includes multiple submaps, the full map and each The sub-map includes an entity; merging the fund knowledge in the fund knowledge metabase, and storing the fused fund knowledge metabase in a database; periodically calculating the full picture and each of the The shortest path between all entities in the sub-map, and establish a shortest path matrix according to the calculation result of the shortest path; obtain a query request, and search for the corresponding shortest path result from the shortest path matrix according to the query request; when When there is no corresponding result in the query request in the shortest path matrix, the corresponding sub-graph is selected from the fund knowledge metabase according to the query request;
  • the two entities use a preset algorithm to calculate the shortest path.
  • a shortest path query system includes an acquisition unit configured to acquire fund knowledge in a fund information source and establish a fund knowledge metabase.
  • the fund knowledge metabase is a full map, and the full map includes multiple submaps.
  • the full picture and each of the sub-pictures include an entity; a fusion unit configured to fuse the fund knowledge in the fund knowledge metabase and store the fused fund knowledge metabase in a database;
  • the maintenance unit is configured to periodically calculate the shortest path between all entities in the full graph and each of the sub-graphs, and establish a shortest path matrix according to the calculation result of the shortest path;
  • the shortest path query unit is configured to obtain a query A request to search a shortest path result corresponding to the query request from the shortest path matrix according to the query request; a sub-graph query unit is set when the query request does not have a corresponding result in the shortest path matrix When the corresponding sub-graph is selected from the fund knowledge metabase according to the query request; the operation unit is set to the sub-graph
  • a computer device includes a memory and a processor.
  • the memory stores computer-readable instructions.
  • the processor causes the processor to execute the steps of the shortest path query method.
  • a storage medium storing computer-readable instructions, when the computer-readable instructions are executed by one or more processors, cause one or more of the processors to execute the steps of the shortest path query method described above.
  • the above shortest path query method, system, computer equipment, and storage medium establish a fund knowledge metabase by obtaining fund knowledge from a fund information source; merge the fund knowledge in the fund knowledge metabase and store it In the database; periodically calculating the shortest path between all the entities in the full graph and each of the sub-graphs, and establishing a shortest path matrix according to the calculation result of the shortest path; obtaining a query request, and from the shortest path matrix Searching the shortest path result corresponding to the query request; when the query request does not have a corresponding result in the shortest path matrix, the corresponding short message is filtered out from the fund knowledge metabase according to the query request
  • the sub-graph calculating a shortest path by using a preset algorithm for a specific pair of entities in the sub-graph. Through the establishment and maintenance of full graphs and sub-graphs, real-time query and full-scale query functions of the graph are realized. Further calculation of the path of each entity in the graph provides a basis for the further reasoning of the relationship between the entities in the
  • FIG. 1 is a flowchart of a shortest path query method in an embodiment of the present application
  • FIG. 2 is a flowchart of establishing a fund knowledge metabase in one embodiment of the present application
  • FIG. 3 is a flowchart of establishing a shortest path matrix in an embodiment of the present application.
  • FIG. 4 is a functional framework diagram of a shortest path query system in an embodiment of the present application.
  • FIG. 1 is a flowchart of a shortest path query method in an embodiment of the present application.
  • the flowchart includes: S1.
  • a fund knowledge metabase is established after obtaining fund knowledge from a fund information source, and the fund knowledge metabase is established.
  • the library is a full map, which contains multiple sub-maps, and the full map and each of the sub-maps include entities; in this step, the knowledge in the fund information source is identified, understood, screened, and summarized. It is extracted to establish a fund knowledge metabase, which includes, in addition to entities, relationships between entities and attributes of the entities.
  • Request a judgment to determine whether it has additional query conditions. If there are no additional query conditions, search the result of the corresponding shortest path in the shortest path matrix, and display the result. S5.
  • the corresponding subgraph is filtered out from the fund knowledge metabase according to the query request; in the query condition in this step, The distance between two fund managers also includes additional conditions, so there is no corresponding result in the shortest path matrix.
  • the additional conditions such as the shortest path between fund managers A and B do not pass through the shortest path of fund company C, the fund manager Between A and B, only the shortest path connected by others is used to match the corresponding subgraph in the fund knowledge metabase according to the keywords in the query request. S6.
  • a predetermined algorithm is used to calculate the shortest path for specific two or two entities in the subgraph.
  • the subgraph corresponding to the query request containing additional query conditions is received, and the Dijkstra algorithm is used for the specific two entities in the subgraph.
  • the shortest path for the entity such as calculating the shortest path for fund manager A, B does not pass through fund company C.
  • FIG. 2 is a flowchart of establishing a fund knowledge metabase in an embodiment of the present application.
  • the flowchart includes: S101. Identifying fund knowledge in an information source and identifying a data type of the fund knowledge. And data source; in this step, the knowledge in the fund information source is identified according to its data type and data source. For example, the data in the internal database of the enterprise is structured data, and the chart data in websites such as Tiantian Fund Network is semi-structured data. The entire text data of fund research reports, fund manager resumes, and snowball community reviews are unstructured data. S102.
  • the merging the fund knowledge in the fund knowledge metabase and storing the fused fund knowledge metabase in a database includes: S201.
  • Each entity in the library is identified by ID, and the fund knowledge metabase also includes the relationship between the entities and the attributes of each entity; in this step, the knowledge data in the knowledge metabase is fused according to a preset fusion rule.
  • all entities are identified by ID. For example, fund entities and stock entities use market transaction codes as ID identifiers.
  • S202. The entities in the fund knowledge metabase are judged according to the ID identifier, and those with a unified ID identifier are the same entity, and the same entity is merged according to the relationship and the attribute to complete the fund.
  • the fusion of knowledge does not need to be merged if it does not have a unified ID.
  • the fusion of data in this step includes the fusion of new data replacing the old data, and also includes the evaluation of the quality of knowledge and fusion with weights according to preset fusion rules.
  • the preset fusion rule is to identify the entities in the fund knowledge metabase by ID, fuse the relationships and attributes of entities identified by the same ID, and fuse the similar attributes of entities that do not have the same ID; S203.
  • the fused fund knowledge metabase is stored in a database; in this step, entities in the fund knowledge metabase are fused according to different relationships and attributes and stored in a database, the database includes a relational database and an RDF database , Graph database or any combination of databases.
  • ID identification is performed on the entities in the fund knowledge metabase, and then the ID identification entities are fused according to a preset fusion rule, and the fused fund knowledge metabase is stored in a database.
  • the orderly integration of the knowledge in the knowledge metabase is a subsequent subgraph that can quickly match the corresponding fund entity in the fund knowledge metabase.
  • FIG. 3 is a flowchart of establishing the shortest path matrix in an embodiment of the present application.
  • the flowchart includes: S301. Set the weights of the edges in the full graph to 1, and use the FLOYD algorithm to perform the full graph. The calculation of the weighted multi-source shortest path; in this step, the weights of the edges between all the entities in the whole graph are set to 1, and then the FLOYD algorithm is used to calculate the shortest path of the two entities for all entities in the whole graph.
  • the shortest path calculated in the whole picture is the shortest path with equal weight and multiple sources.
  • the FLOYD algorithm is as follows: S30101. Starting from any one-sided path, the distance between all two points is the weight of the edge.
  • the calculation result of the shortest path with multiple sources with weights calculated in the figure is established based on the calculation result.
  • S304. Periodically check the fund knowledge metabase, and if there are changes in entities and relationships, update the entity and relationship in the fund knowledge metabase and the shortest path matrix; periodically check the fund knowledge metabase in this step Whether there are changes in the entities and relationships in the database. If there are changes, update the changed entities or relationships to the full graph and the corresponding subgraphs, and calculate the equal-weighted multi-source shortest path and weighted multi-source shortest path respectively.
  • the calculation result is updated in the shortest path matrix.
  • the shortest path is calculated for each entity in the full graph and each sub-graph by a preset algorithm, and a shortest path matrix is established, which provides a basis for subsequent real-time query and full query of the shortest path.
  • the obtaining a query request, and searching the shortest path result corresponding to the query request from the shortest path matrix according to the query request includes: S401. Obtaining two queries included in the query request. The keywords of the entity are searched in the shortest path matrix; the entities in this step include fund companies, funds, fund managers, etc. In this step, the default user query is the shortest path with equal weight. For Different types of relationships are considered edges of equal length. S402. A shortest path result corresponding to the query request is searched in the shortest path matrix, and the result is displayed by using d3js technology. In this embodiment, the shortest path with equal weight can be directly obtained from the shortest path matrix, which reflects that the graph has the function of real-time query.
  • the specific two or two entities in the sub-graph are calculated using the preset algorithm to calculate the shortest path, and the calculation results are displayed, including the Dijkstra algorithm is used to short the specific two points in the sub-graph.
  • the calculation of the path, the specific two points refer to the two entities included in the query request, and the calculation result is displayed in json data format using d3js technology to display.
  • the calculation of the Dijkstra algorithm is as follows: S601 The length from the starting point A to itself is 0, and the straight line distance between A and other points is the weight of the edge. If there is no edge connection between the two points, it is infinite; S602. Put A into a set M and find the set.
  • the point C with the smallest distance between the other points is put into the set; S603. Since the newly added C may affect the distance from other points in the set M to the starting point A, the distance between each vertex in the set M and the starting point A is updated; S604. Repeat the steps S602 and step S603, until all vertices are traversed.
  • the shortest path results that cannot be obtained quickly in the shortest path matrix can also be obtained through a fast calculation process. It shows that the map has the function of full query.
  • the step of periodically calculating the shortest path between all the entities in the full graph and each sub-graph, and after establishing the shortest path matrix according to the calculation result of the shortest path the method further includes: The path length between two entities in the shortest path matrix is compared, and the path length is compared with a preset threshold. If the path length is lower than the threshold, the two entities corresponding to the path length are identified. The entities are output to other platforms, which perform deep relationship mining on the identified fund entities.
  • the path length between two entities in the shortest path matrix is compared with a preset threshold, and the entity corresponding to the path length below the threshold is identified, which is more conducive to mining funds with similar styles. Managers and their potential impact relationships.
  • a shortest path query system includes: an obtaining unit configured to obtain fund knowledge in a fund information source and establish a fund knowledge metabase, where the fund knowledge metabase is a full map, The full picture includes a plurality of sub-pictures, and the full picture and each of the sub-pictures include entities; a fusion unit is configured to fuse the fund knowledge in the fund knowledge metabase, and the fused The fund knowledge metabase is stored in a database; a maintenance unit is configured to periodically calculate the shortest path between all entities in the full graph and each of the sub-graphs, and establish the shortest path according to the calculation result of the shortest path Matrix; a shortest path query unit configured to obtain a query request, and search the shortest path matrix corresponding to the query request from the shortest path matrix according to the query request; a subgraph query unit configured to be set when the query request is in When there is no corresponding result in the shortest path matrix, the corresponding subgraph is selected from the fund knowledge metabase according to the query
  • the fusion unit includes an identification module configured to identify each entity in the fund knowledge metabase, and the fund knowledge metabase further includes a relationship between entities and an entity's Attributes; a merging module configured to judge each entity in the fund knowledge metabase based on the ID identifier, and those with a unified ID identifier are the same entity, and merge the same entity according to the relationship and the attribute After the fusion of the fund knowledge is completed, if there is no unified ID identification, then the merger is not required; the storage module is set to store the fused fund knowledge metabase in a database.
  • the maintenance unit includes: a first-level operation module configured to use a FLOYD algorithm to calculate the equal-weighted multi-source shortest path for the full graph, wherein the weight of the edges of the full graph is set to 1; the second-level computation module , Set to use the FLOYD algorithm to calculate the multi-source shortest path with weights for each subgraph; establish a matrix module and set to calculate the multi-source shortest path with weights and the result of the multi-source shortest path with weights To establish a shortest path matrix; an inspection module configured to periodically check the fund knowledge metabase, and if there are changes in entities and relationships, update the entities and relationships in the fund knowledge metabase and the shortest path matrix.
  • a first-level operation module configured to use a FLOYD algorithm to calculate the equal-weighted multi-source shortest path for the full graph, wherein the weight of the edges of the full graph is set to 1
  • the second-level computation module Set to use the FLOYD algorithm to calculate the multi-source shortest path
  • the shortest path query unit includes: a search module configured to obtain keywords of two entities included in the query request, and perform a keyword search in the shortest path matrix; a display module, set In order to search for the shortest path result corresponding to the query request in the shortest path matrix, the result is displayed through d3js technology.
  • the operation unit includes: an operation module configured to calculate a shortest path to specific two points in the sub-graph by using Dijkstra algorithm, where the specific two points refer to two points included in the query request. Entities, and return the calculation results in json data format to display using d3js technology.
  • the maintenance unit further includes: a mining module configured to mine a potential relationship between two entities.
  • the mining module includes: an identification entity module configured to obtain a path length between two entities in the shortest path matrix, and comparing the path length with a preset threshold, if the path length is low At the threshold, two entities corresponding to the path length are identified; a mining relationship module is configured to output the identified entities to other platforms, and the other platforms perform deep relationship mining on the identified entities .
  • a computer device in one embodiment, includes a memory and a processor.
  • the memory stores computer-readable instructions.
  • the processor is caused to execute the foregoing embodiments. Steps in the shortest path query method.
  • a storage medium storing computer-readable instructions, and when the computer-readable instructions are executed by one or more processors, one or more of the processors execute the foregoing embodiments. Steps in the shortest path query method.
  • the storage medium may be a non-volatile storage medium.
  • the program may be stored in a computer-readable storage medium.
  • the storage medium may include: Read-only database (ROM, Read Only Memory), random access database (RAM, Random Access Memory), magnetic disk or optical disk, etc.

Landscapes

  • Engineering & Computer Science (AREA)
  • Business, Economics & Management (AREA)
  • Finance (AREA)
  • Accounting & Taxation (AREA)
  • Development Economics (AREA)
  • Operations Research (AREA)
  • Technology Law (AREA)
  • Human Resources & Organizations (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Economics (AREA)
  • Marketing (AREA)
  • Strategic Management (AREA)
  • Game Theory and Decision Science (AREA)
  • Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

本申请涉及图论和网络分析技术领域,特别涉及一种最短路径查询方法、***、计算机设备和存储介质。一种最短路径方法,包括获取信息源中的基金知识后建立基金知识元库;对基金知识进行融合并存储于数据库中;定期计算全图和各子图中所有实体两两间最短路径,并建立最短路径矩阵;获取查询请求,并从最短路径矩阵中搜索对应的结果;当最短路径矩阵中不存在对应的结果时,到基金知识元库中筛选出对应的子图;计算子图中特定的两两实体最短路径。通过对全图和子图的建立与维护,实现了图谱的实时查询和全量查询功能,通过进一步对图谱中各实体间的路径的计算,为图谱中各实体关系的进一步推理提供了基础。

Description

最短路径查询方法、***、计算机设备和存储介质
本申请要求于2018年09月19日提交中国专利局、申请号为201811093461.2、发明名称为“最短路径查询方法、***、计算机设备和存储介质”的中国专利申请的优先权,其全部内容通过引用结合在本申请中。
技术领域
本申请涉及图论和网络分析技术领域,特别是涉及一种最短路径查询方法、***、计算机设别和存储介质。
背景技术
知识图谱本质上是语义网络,是一种基于图的数据结构,由节点和边组成。在知识图谱里,每个节点表示现实世界中存在的“实体”,每条边为实体与实体之间的“关系”。知识图谱是关系的最有效的表示方式,通俗地讲,知识图谱就是把所有不同种类的信息连接在一起而得到的一个关系网络。知识图谱提供了从“关系”的角度去分析问题的能力,知识的推理可以理解成“链接预测”,也就是从已有的关系图谱里推导出新的关系或链接,常用的推理算法包括基于逻辑的推理和基于分布式表示方法的推理。
知识图谱可以结合多种数据源去分析实体之间的关系,从而对用户的行为有更好的理解,在大数据时代,很多数据都是未经处理过的非结构化数据,比如文本、图片、音频、视频等。特别在互联网金融行业里,往往需要根据大量的文本数据,并从这些非结构化数据里提取出有价值的信息。
而现有的知识图谱中对于各实体的分析不全面,且对于各实体的查询也比较简单,不能实现实时查询和全量查询的功能。
发明内容
基于此,有必要针对现有知识图谱中对于各实体的分析不全面且不能对各实体进行实时查询和全量查询的问题,提供一种最短路径查询方法、***、计算机设备和存储介质。
一种最短路径查询方法,包括:获取基金信息源中的基金知识后建立基金知识元库,所述基金知识元库为全图,所述全图中包含多个子图,所述全图和各所述子图中包括实体;将所述基金知识元库中的所述基金知识进行融合,并将经过融合的所述基金知识元库存储于数据库中;定期计算所述全图和各所述子图中所有实体两两之间的最短路径,根据所述最短路径的计算结果建立最短路径矩阵;获取查询请求,根据所述查询请求从所述最短路径矩阵中搜索对应的最短路径结果;当所述查询请求在所述最短路径矩阵中不存在对应的结果时,则根据所述查询请求到所述基金知识元库中筛选出对应的所述子图;对所述子图中特定的两两实体采用预设算法计算最短路径。
一种最短路径查询***,包括:获取单元,设置为获取基金信息源中的基金知识后建立基金知识元库,所述基金知识元库为全图,所述全图中包含多个子图,所述全图和各所述子图中包括实体;融合单元,设置为将所述基金知识元库中的所述基金知识进行融合,并将经过融合的所述基金知识元库存储于数据库中;维护单元,设置为定期计算所述全图和各所述子图中所有实体两两之间的最短路径,根据所述最短路径的计算结果建立最短路径矩阵;最短路径查询单元,设置为获取查询请求,根据所述查询请求从所述最短路径矩阵中搜索与所述查询请求对应的最短路径结果;子图查询单元,设置为当所述查询请求在所述最短路径矩阵中不存在对应的结果时,则根据所述查询请 求到所述基金知识元库中筛选出对应的所述子图;运算单元,设置为对所述子图中特定的两两实体采用预设算法计算最短路径。
一种计算机设备,包括存储器和处理器,所述存储器中存储有计算机可读指令,所述计算机可读指令被所述处理器执行时,使得所述处理器执行上述最短路径查询方法的步骤。
一种存储有计算机可读指令的存储介质,所述计算机可读指令被一个或多个处理器执行时,使得一个或多个所述处理器执行上述最短路径查询方法的步骤。
上述最短路径查询方法、***、计算机设备和存储介质,通过获取基金信息源中的基金知识后建立基金知识元库;将所述基金知识元库中的所述基金知识进行融合,并将其存储于数据库中;定期计算所述全图和各所述子图中所有实体两两之间的最短路径,根据所述最短路径的计算结果建立最短路径矩阵;获取查询请求,从所述最短路径矩阵中搜索与所述查询请求对应的最短路径结果;当所述查询请求在所述最短路径矩阵中不存在对应的结果时,则根据所述查询请求到所述基金知识元库中筛选出对应的所述子图;对所述子图中特定的两两实体采用预设算法计算最短路径。通过对全图和子图的建立与维护,实现了图谱的实时查询和全量查询功能,经过进一步的对图谱中各实体的路径计算,为图谱中各实体关系的进一步推理提供了基础。
附图说明
通过阅读下文优选实施方式的详细描述,各种其他的优点和益处对于本领域普通技术人员将变得清楚明了。附图仅用于示出优选实施方式的目的,而并不认为是对本申请的限制。
图1为本申请在一个实施例中最短路径查询方法的流程图;
图2为本申请在一个实施例中建立基金知识元库的流程图;
图3为本申请在一个实施例中建立最短路径矩阵的流程图;
图4为本申请在一个实施例中最短路径查询***的功能框架图。
具体实施方式
为了使本申请的目的、技术方案及优点更加清楚明白,以下结合附图及实施例,对本申请进行进一步详细说明。应当理解,此处所描述的具体实施例仅仅用以解释本申请,并不用于限定本申请。
图1为本申请在一个实施例中最短路径查询方法的流程图,如图所示,该流程图包括:S1、获取基金信息源中的基金知识后建立基金知识元库,所述基金知识元库为全图,所述全图中包含多个子图,所述全图和各所述子图中包括实体;本步骤中通过把基金信息源中的知识经过识别、理解、筛选、归纳等过程抽取出来,建立基金知识元库,所述知识元库除了包括实体外,还包括实体之间的关系和实体的属性。S2、将所述基金知识元库中的所述基金知识进行融合,并将经过融合的所述基金知识元库存储于数据库中;本步骤中通过使来自不同基金知识源的知识在同一框架规范下进行数据整合,并对所述知识元库中的实体进行ID标识,该融合过程中包括新数据替换旧数据的融合,还包括根据预设的融合规则对知识的质量进行评估和带权重的融合。S3、定期计算所述全图和各所述子图中所有实体两两之间的最短路径,根据所述最短路径的计算结果建立最短路径矩阵;本步骤中根据预设的权重规则,对全图和各子图中所有的实体两两计算其最短路径,并根据计算的最短路径结果建立最短路径矩阵,并且定期检查所述基金知识元库中的图谱,若有实体或关系的更新,则在对应的图谱和最短路径矩阵中进行更新。S4、获取查询请求,根据所述查询请求 从所述最短路径矩阵中搜索与所述查询请求对应的最短路径结果;本步骤中获取查询请求为查询两个基金经理之间的距离,对该查询请求进行判断,判断其是否有额外查询条件,若没有额外查询条件,则到所述最短路径矩阵中搜索其对应的最短路径的结果,并将该结果进行展示。S5、当所述查询请求在所述最短路径矩阵中不存在对应的结果时,则根据所述查询请求到所述基金知识元库中筛选出对应的所述子图;本步骤中查询条件中两个基金经理之间的距离还包含额外条件,故在所述最短路径矩阵中不存在其对应的结果,其额外条件比如基金经理A,B之间不通过基金公司C的最短路径,基金经理A,B之间只通过其他人相连的最短路径等,根据查询请求中的关键词到所述基金知识元库中匹配到其对应的子图。S6、对所述子图中特定的两两实体采用预设算法计算最短路径;本步骤中接收包含额外查询条件的查询请求对应的子图,并采用Dijkstra算法对该子图中特定的两个实体的最短路径,比如计算基金经理A,B不通过基金公司C的最短路径。本申请通过上述步骤方法,实现了对图谱中两两实体的最短路径计算的功能,并通过计算得到的最短路径建立了最短路径矩阵,实现了图谱可实时查询、全量查询的功能。
图2为本申请在一个实施例中建立基金知识元库的流程图,如图2所示,该流程图包括:S101、对信息源中的基金知识进行识别,识别所述基金知识的数据类型和数据来源;本步骤中通过对基金信息源中的知识根据其数据类型和数据来源进行识别,例如企业内部数据库的数据为结构化数据,天天基金网等网站中的图表数据为半结构化数据,基金研报、基金经理简历、雪球社区评论等整篇文本数据为非结构化数据。S102、根据所述基金知识的数据类型和数据来源进行筛选 与归纳,筛选出具有相同所述数据类型和相同所述数据来源的所述基金知识并归纳为一类;本步骤中将具有同一数据类型和同一数据来源的知识数据归纳为同一类,并且根据其不同的数据类型采取不同的抽取方法,例如对于结构化数据,通过人工设定规则来进行数据抽取,对于半结构化数据,通过爬虫或正规表达式匹配来进行数据抽取,对于非结构化数据,通过自然语言处理来进行数据抽取。S103、根据归纳整理后的所述基金知识,建立基金知识元库;本实施例中通过对基金信息源中的数据进行抽取并建立基金知识元库,为后续对所述基金知识元库中的数据进行进一步的整合提供了基础。
在一个实施例中,所述将所述基金知识元库中的所述基金知识进行融合,并将经过融合的所述基金知识元库存储于数据库中,包括:S201、对所述基金知识元库中的各实体进行ID标识,所述基金知识元库中还包括各实体间的关系和各实体的属性;本步骤中在对所述知识元库中的知识数据根据预设融合规则进行融合之前,先对所有的实体进行ID标识,例如,基金实体和股票实体以市场交易代码作为ID标识。S202、根据所述ID标识对所述基金知识元库中的各实体进行判断,具有统一ID标识的为同一实体,将所述同一实体根据所述关系和所述属性进行合并后完成所述基金知识的融合,不具有统一ID标识的,则无需进行合并;本步骤中对数据的融合包括新数据替换旧数据的融合,还包括根据预设融合规则对知识的质量进行评估和带权重的融合,该预设融合规则即将所述基金知识元库中的实体进行ID标识,对于同一ID标识的实体进行关系与属性的融合,对于不具有同一ID标识的实体进行相似属性的融合;S203、将经过融合的所述基金知识元库存储于数据库中;本步骤中将所述基金知识元库中的实 体按照关系和属性的不同进行融合后存储于数据库中,所述数据库包括关系数据库,RDF数据库,图数据库或任意数据库的结合。本实施例中通过对所述基金知识元库中的实体进行ID标识,再将进行ID标识的实体根据预设融合规则进行融合,并将融合后的所述基金知识元库存储于数据库中,将所述知识元库中的知识进行有序的整合,为后续能够快速的匹配基金实体在所述基金知识元库中相对应的子图。
图3为本申请在一个实施例中建立最短路径矩阵的流程图,如图3所示,该流程图包括:S301、将全图中边的权重设置为1,运用FLOYD算法对全图进行等权重多源最短路径的计算;本步骤中将全图中所有实体两两间的边的权重均设置为1,再运用FLOYD算法对全图中所有的实体进行两两实体的最短路径的计算,全图中计算的最短路径为等权重多源最短路径,其中FLOYD算法如下所示:S30101、从任意一条单边路径开始,所有两点之间的距离为边的权重,如果两点之间无边相连为无穷大。S30102、对于每一对顶点i和j,看看是否存在一个顶点k使得从i到k再到j比己知的路径更短,如果是更新它。Floyd算法是一个动态规划算法,它的递推公式如下:d[i][j]=min(d[i][j],d[i][k]+d[k][j])。S302、设置各子图中的边的权重,运用FLOYD算法对各子图进行带权重多源最短路径的计算;本步骤中通过对各子图中的实体间的边设置权重,运用FLOYD算法计算两两实体间的带权重多源最短路径,比如两个基金经理之间的关系若为亲属、同导师、共同管理过同一个基金的,其权重设置为1;如果两个基金经理之间的关系为同所学校毕业的、同公司或同学的,则权重设置为2。S303、根据所述等权重多源最短路径的计算结果和所述带权重多源最短路径的计算结果,建立最短路径矩阵;本步骤中获取全 图中计算的等权重多源最短路径与各子图中计算的带权重多源在最短路径的计算结果,根据该计算结果建立最短路径矩阵。S304、定期检查所述基金知识元库,若有实体和关系的变化,在所述基金知识元库和所述最短路径矩阵中更新该实体和关系;本步骤中定期检查所述基金知识元库中的实体和关系是否发生变化,若有变化,则将变化后的实体或关系更新到全图和对应的子图中,并分别计算等权重多源最短路径和带权重多源最短路径,将计算结果更新于所述最短路径矩阵中。本实施例中通过预设算法分别对全图和各子图中的实体两两计算其最短路径,并建立最短路径矩阵,为后续能够实时查询和全量查询最短路径提供基础。
在一个实施例中,所述获取查询请求,根据所述查询请求从所述最短路径矩阵中搜索与所述查询请求对应的最短路径结果,包括:S401、获取所述查询请求中包含的两个实体的关键词,到所述最短路径矩阵中进行关键词搜索;本步骤中所述实体包括基金公司、基金、基金经理等,此步骤中,默认用户查询的就是等权的一条最短路径,对于不同类型的关系都看做是等长度的边。S402、在所述最短路径矩阵中搜索到所述查询请求对应的最短路径结果,则将该结果通过d3js技术进行展示。本实施例中对于等权重最短路径可到所述最短路径矩阵中直接获取,体现了图谱具备实时查询的功能。
在一个实施例中,所述对所述子图中特定的两两实体采用预设算法计算最短路径,并将计算结果进行展示,包括采用Dijkstra算法对所述子图中特定的两点进行最短路径的计算,其中特定的两点指所述查询请求中包含的两个实体,并将该计算结果以json数据格式返回后采用d3js技术进行展示,其中,Dijkstra算法的计算如下步骤所示: S601、起点A到自身的长度为0,A到别的点之间的直线距离为边的权重,如果两点之间无边相连为无穷大;S602、将A放入一个集合M中,并且找出集合外其它点距离最小的点C放入集合中;S603、由于新加进来的C可能影响集合M中其它点到起点A的距离,更新集合M中各个顶点到起点A的距离;S604、重复步骤S602、步骤S603,直到遍历完所有顶点。本实施例中对带有额外查询条件的查询请求,通过先匹配子图再进行计算的过程,对于不能在所述最短路径矩阵中快速获取的最短路径结果,也能通过快速的计算过程获取,体现了图谱具备全量查询的功能。
在一个实施例中,所述定期计算所述全图和所述各子图中所有实体两两之间的最短路径,根据所述最短路径的计算结果建立最短路径矩阵之后,还包括:获取所述最短路径矩阵中两两实体间的路径长度,将该路径长度与预设阈值进行比对,若该路径长度低于所述阈值,标识该路径长度所对应的两个实体;将标识后的所述实体输出至其它平台中,所述其它平台对所述标识后的基金实体进行深度关系挖掘。本实施例中通过将所述最短路径矩阵中两两实体间的路径长度与预设阈值进行比对,将低于所述阈值的路径长度对应的实体进行标识,更加有利于挖掘风格类似的基金经理及他们之间潜在的影响关系。
如图4所示,在一个实施例中,一种最短路径查询***包括:获取单元,设置为获取基金信息源中的基金知识后建立基金知识元库,所述基金知识元库为全图,所述全图中包含多个子图,所述全图和各所述子图中包括实体;融合单元,设置为将所述基金知识元库中的所述基金知识进行融合,并将经过融合的所述基金知识元库存储于数据库中;维护单元,设置为定期计算所述全图和各所述子图中所有实体 两两之间的最短路径,根据所述最短路径的计算结果建立最短路径矩阵;最短路径查询单元,设置为获取查询请求,根据所述查询请求从所述最短路径矩阵中搜索与所述查询请求对应的最短路径结果;子图查询单元,设置为当所述查询请求在所述最短路径矩阵中不存在对应的结果时,则根据所述查询请求到所述基金知识元库中筛选出对应的所述子图;运算单元,设置为对所述子图中特定的两两实体采用预设算法计算最短路径。
在一个实施例中,所述融合单元包括:标识模块,设置为对所述基金知识元库中的各实体进行ID标识,所述基金知识元库中还包括各实体间的关系和各实体的属性;合并模块,设置为根据所述ID标识对所述基金知识元库中的各实体进行判断,具有统一ID标识的为同一实体,将所述同一实体根据所述关系和所述属性进行合并后完成所述基金知识的融合,不具有统一ID标识的,则无需进行合并;存储模块,设置为将经过融合的所述基金知识元库存储于数据库中。
在一个实施例中,所述述维护单元包括:一级运算模块,设置为运用FLOYD算法对全图进行等权重多源最短路径的计算,其中全图边的权重设置为1;二级计算模块,设置为运用FLOYD算法对各子图进行带权重多源最短路径的计算;建立矩阵模块,设置为根据所述等权重多源最短路径的计算结果和所述带权重多源最短路径的计算结果,建立最短路径矩阵;检查模块,设置为定期检查所述基金知识元库,若有实体和关系的变化,在所述基金知识元库和所述最短路径矩阵中更新该实体和关系。
在一个实施例中,所述最短路径查询单元包括:搜索模块,设置为获取所述查询请求中包含的两个实体的关键词,到所述最短路径矩 阵中进行关键词搜索;展示模块,设置为在所述最短路径矩阵中搜索到所述查询请求对应的最短路径结果,则将该结果通过d3js技术进行展示。
在一个实施例中,所述运算单元包括:运算模块,设置为采用Dijkstra算法对所述子图中特定的两点进行最短路径的计算,其中特定的两点指所述查询请求中包含的两个实体,并将该计算结果以json数据格式返回后采用d3js技术进行展示。
在一个实施例中,所述维护单元还包括:挖掘模块,设置为挖掘两两实体间的潜在关系。
在一个实施例中,所述挖掘模块包括:标识实体模块,设置为获取所述最短路径矩阵中两两实体间的路径长度,将该路径长度与预设阈值进行比对,若该路径长度低于所述阈值,标识该路径长度所对应的两个实体;挖掘关系模块,设置为将标识后的所述实体输出至其它平台中,所述其它平台对所述标识后的实体进行深度关系挖掘。
在一个实施例中,提出了一种计算机设备,所述计算机设备包括存储器和处理器,存储器中存储有计算机可读指令,计算机可读指令被处理器执行时,使得处理器执行上述各实施例中的所述最短路径查询方法的步骤。
在一个实施例中,提出了一种存储有计算机可读指令的存储介质,所述计算机可读指令被一个或多个处理器执行时,使得一个或多个所述处理器执行上述各实施例中的所述最短路径查询方法的步骤。其中,所述存储介质可以为非易失性存储介质。
本领域普通技术人员可以理解上述实施例的各种方法中的全部或部分步骤是可以通过程序来指令相关的硬件来完成,该程序可以存 储于一计算机可读存储介质中,存储介质可以包括:只读数据库(ROM,Read Only Memory)、随机存取数据库(RAM,Random Access Memory)、磁盘或光盘等。
以上所述实施例的各技术特征可以进行任意的组合,为使描述简洁,未对上述实施例中的各个技术特征所有可能的组合都进行描述,然而,只要这些技术特征的组合不存在矛盾,都应当认为是本申请的说明书记载的范围。
以上所述实施例仅表达了本申请一些示例性实施例,其描述较为具体和详细,但并不能因此而理解为对本申请专利所要求保护范围的限制。应当指出的是,对于本领域的普通技术人员来说,在不脱离本申请构思的前提下,还可以做出若干变形和改进,这些变形和改进都属于本申请所要求的保护范围。因此,本申请专利的保护范围应以所附权利要求为准。

Claims (20)

  1. 一种最短路径查询方法,包括:
    获取基金信息源中的基金知识后建立基金知识元库,所述基金知识元库为全图,所述全图中包含多个子图,所述全图和各所述子图中包括实体;
    将所述基金知识元库中的所述基金知识进行融合,并将经过融合的所述基金知识元库存储于数据库中;
    定期计算所述全图和各所述子图中所有实体两两之间的最短路径,根据所述最短路径的计算结果建立最短路径矩阵;
    获取查询请求,根据所述查询请求从所述最短路径矩阵中搜索对应的最短路径结果;
    当所述查询请求在所述最短路径矩阵中不存在对应的结果时,则根据所述查询请求到所述基金知识元库中筛选出对应的所述子图;
    对所述子图中特定的两两实体采用预设算法计算最短路径。
  2. 如权利要求1所述的一种最短路径查询方法,其中,所述获取基金信息源中的基金知识后建立基金知识元库,包括:
    对信息源中的基金知识进行识别,识别所述基金知识的数据类型和数据来源;
    根据所述基金知识的数据类型和数据来源进行筛选与归纳,筛选出具有相同所述数据类型和相同所述数据来源的所述基金知识并归纳为一类;
    根据归纳整理后的所述基金知识,建立基金知识元库。
  3. 如权利要求1所述的一种最短路径查询方法,其中,所述将所述基金知识元库中的所述基金知识进行融合,并将经过融合的所述基金知识元库存储于数据库中,包括:
    对所述基金知识元库中的各实体进行ID标识,所述基金知识元库中还包括各实体间的关系和各实体的属性;
    根据所述ID标识对所述基金知识元库中的各实体进行判断,具有统一ID标识的为同一实体,将所述同一实体根据所述关系和所述属性进行合并后完成所述基金知识的融合,不具有统一ID标识的,则无需进行合并;
    将经过融合的所述基金知识元库存储于数据库中。
  4. 如权利要求1所述的一种最短路径查询方法,其中,所述定期计算所述全图和各所述子图中所有实体两两之间的最短路径,根据所述最短路径的计算结果建立最短路径矩阵,包括:
    将全图中边的权重设置为1,运用FLOYD算法对全图进行等权重多源最短路径的计算;
    设置各子图中的边的权重,运用FLOYD算法对各子图进行带权重多源最短路径的计算;
    根据所述等权重多源最短路径的计算结果和所述带权重多源最短路径的计算结果,建立最短路径矩阵;
    定期检查所述基金知识元库,若有实体和关系的变化,在所述基金知识元库和所述最短路径矩阵中更新该实体和关系。
  5. 如权利要求1所述的一种最短路径查询方法,其中,所述获取查询请求,根据所述查询请求从所述最短路径矩阵中搜索与所述查询请求对应的最短路径结果,包括:
    获取所述查询请求中包含的两个实体的关键词,到所述最短路径矩阵中进行关键词搜索;
    在所述最短路径矩阵中搜索到所述查询请求对应的最短路径结 果,则将该结果通过d3js技术进行展示。
  6. 如权利要求1所述的一种最短路径查询方法,其中,所述对所述子图中特定的两两实体采用预设算法计算最短路径,包括:
    采用Dijkstra算法对所述子图中特定的两点进行最短路径的计算,其中特定的两点指所述查询请求中包含的两个实体,并将该计算结果以json数据格式返回后采用d3js技术进行展示。
  7. 如权利要求1所述的一种最短路径查询方法,其中,所述定期计算所述全图和所述各子图中所有实体两两之间的最短路径,根据所述最短路径的计算结果建立最短路径矩阵之后,还包括:
    获取所述最短路径矩阵中两两实体间的路径长度,将该路径长度与预设阈值进行比对,若该路径长度低于所述阈值,标识该路径长度所对应的两个实体;
    将标识后的所述实体输出至其它平台中,所述其它平台对所述标识后的实体进行深度关系挖掘。
  8. 一种最短路径查询***,包括:
    获取单元,设置为获取基金信息源中的基金知识后建立基金知识元库,所述基金知识元库为全图,所述全图中包含多个子图,所述全图和各所述子图中包括实体;
    融合单元,设置为将所述基金知识元库中的所述基金知识进行融合,并将经过融合的所述基金知识元库存储于数据库中;
    维护单元,设置为定期计算所述全图和各所述子图中所有实体两两之间的最短路径,根据所述最短路径的计算结果建立最短路径矩阵;
    最短路径查询单元,设置为获取查询请求,根据所述查询请求从 所述最短路径矩阵中搜索与所述查询请求对应的最短路径结果;
    子图查询单元,设置为当所述查询请求在所述最短路径矩阵中不存在对应的结果时,则根据所述查询请求到所述基金知识元库中筛选出对应的所述子图;
    运算单元,设置为对所述子图中特定的两两实体采用预设算法计算最短路径。
  9. 如权利要求8所述的最短路径查询***,其中,所述获取单元包括:
    识别模块,设置为对信息源中的基金知识进行识别,识别所述基金知识的数据类型和数据来源;
    筛选模块,设置为根据所述基金知识的数据类型和数据来源进行筛选与归纳,筛选出具有相同所述数据类型和相同所述数据来源的所述基金知识并归纳为一类;
    建立模块,设置为根据归纳整理后的所述基金知识,建立基金知识元库。
  10. 如权利要求8所述的最短路径查询***,其中,所述融合单元包括:
    标识模块,设置为对所述基金知识元库中的各实体进行ID标识,所述基金知识元库中还包括各实体间的关系和各实体的属性;
    合并模块,设置为根据所述ID标识对所述基金知识元库中的各实体进行判断,具有统一ID标识的为同一实体,将所述同一实体根据所述关系和所述属性进行合并后完成所述基金知识的融合,不具有统一ID标识的,则无需进行合并;
    存储模块,设置为将经过融合的所述基金知识元库存储于数据库 中。
  11. 如权利要求8所述的最短路径查询***,其中,所述维护单元包括:
    一级运算模块,设置为运用FLOYD算法对全图进行等权重多源最短路径的计算,其中全图边的权重设置为1;
    二级计算模块,设置为运用FLOYD算法对各子图进行带权重多源最短路径的计算;
    建立矩阵模块,设置为根据所述等权重多源最短路径的计算结果和所述带权重多源最短路径的计算结果,建立最短路径矩阵;
    检查模块,设置为定期检查所述基金知识元库,若有实体和关系的变化,在所述基金知识元库和所述最短路径矩阵中更新该实体和关系。
  12. 如权利要求8所述的最短路径查询***,其中,所述最短路径查询单元包括:
    搜索模块,设置为获取所述查询请求中包含的两个实体的关键词,到所述最短路径矩阵中进行关键词搜索;
    展示模块,设置为在所述最短路径矩阵中搜索到所述查询请求对应的最短路径结果,则将该结果通过d3js技术进行展示。
  13. 如权利要求8所述的最短路径查询***,其中,所述运算单元包括:
    运算模块,设置为采用Dijkstra算法对所述子图中特定的两点进行最短路径的计算,其中特定的两点指所述查询请求中包含的两个实体,并将该计算结果以json数据格式返回后采用d3js技术进行展示。
  14. 如权利要求8所述的最短路径查询***,其中,所述维护单 元还包括:
    挖掘模块,设置为挖掘两两实体间的潜在关系。
  15. 如权利要求14所述的最短路径查询***,其中,所述挖掘模块包括:
    标识实体模块,设置为获取所述最短路径矩阵中两两实体间的路径长度,将该路径长度与预设阈值进行比对,若该路径长度低于所述阈值,标识该路径长度所对应的两个实体;
    挖掘关系模块,设置为将标识后的所述实体输出至其它平台中,所述其它平台对所述标识后的实体进行深度关系挖掘。
  16. 一种计算机设备,包括存储器和处理器,所述存储器中存储有计算机可读指令,所述计算机可读指令被所述处理器执行时,使得所述处理器执行以下步骤:获取基金信息源中的基金知识后建立基金知识元库,所述基金知识元库为全图,所述全图中包含多个子图,所述全图和各所述子图中包括实体;将所述基金知识元库中的所述基金知识进行融合,并将经过融合的所述基金知识元库存储于数据库中;定期计算所述全图和各所述子图中所有实体两两之间的最短路径,根据所述最短路径的计算结果建立最短路径矩阵;获取查询请求,根据所述查询请求从所述最短路径矩阵中搜索对应的最短路径结果;当所述查询请求在所述最短路径矩阵中不存在对应的结果时,则根据所述查询请求到所述基金知识元库中筛选出对应的所述子图;对所述子图中特定的两两实体采用预设算法计算最短路径。
  17. 如权利要求16所述的一种计算机设备,其中,所述所述获取基金信息源中的基金知识后建立基金知识元库时,使得所述处理器执行以下步骤:对信息源中的基金知识进行识别,识别所述基金知识 的数据类型和数据来源;根据所述基金知识的数据类型和数据来源进行筛选与归纳,筛选出具有相同所述数据类型和相同所述数据来源的所述基金知识并归纳为一类;根据归纳整理后的所述基金知识,建立基金知识元库。
  18. 如权利要求16所述的一种计算机设备,其中,所述对所述子图中特定的两两实体采用预设算法计算最短路径时,使得所述处理器执行以下步骤:采用Dijkstra算法对所述子图中特定的两点进行最短路径的计算,其中特定的两点指所述查询请求中包含的两个实体,并将该计算结果以json数据格式返回后采用d3js技术进行展示。
  19. 一种存储有计算机可读指令的存储介质,所述计算机可读指令被一个或多个处理器执行时,使得一个或多个所述处理器执行以下步骤:获取基金信息源中的基金知识后建立基金知识元库,所述基金知识元库为全图,所述全图中包含多个子图,所述全图和各所述子图中包括实体;将所述基金知识元库中的所述基金知识进行融合,并将经过融合的所述基金知识元库存储于数据库中;定期计算所述全图和各所述子图中所有实体两两之间的最短路径,根据所述最短路径的计算结果建立最短路径矩阵;获取查询请求,根据所述查询请求从所述最短路径矩阵中搜索对应的最短路径结果;当所述查询请求在所述最短路径矩阵中不存在对应的结果时,则根据所述查询请求到所述基金知识元库中筛选出对应的所述子图;对所述子图中特定的两两实体采用预设算法计算最短路径。
  20. 如权利要求19所述的一种存储有计算机可读指令的存储介质,其中,所述对所述子图中特定的两两实体采用预设算法计算最短路径时,使得所述一个或多个处理器执行以下步骤:采用Dijkstra算 法对所述子图中特定的两点进行最短路径的计算,其中特定的两点指所述查询请求中包含的两个实体,并将该计算结果以json数据格式返回后采用d3js技术进行展示。
PCT/CN2018/124237 2018-09-19 2018-12-27 最短路径查询方法、***、计算机设备和存储介质 WO2020056984A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201811093461.2A CN109492111B (zh) 2018-09-19 2018-09-19 最短路径查询方法、***、计算机设备和存储介质
CN201811093461.2 2018-09-19

Publications (1)

Publication Number Publication Date
WO2020056984A1 true WO2020056984A1 (zh) 2020-03-26

Family

ID=65690485

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2018/124237 WO2020056984A1 (zh) 2018-09-19 2018-12-27 最短路径查询方法、***、计算机设备和存储介质

Country Status (2)

Country Link
CN (1) CN109492111B (zh)
WO (1) WO2020056984A1 (zh)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112231350A (zh) * 2020-10-13 2021-01-15 汉唐信通(北京)科技有限公司 一种基于知识图谱的企业商机挖掘方法和装置
CN112396184A (zh) * 2020-12-01 2021-02-23 中山大学 一种基于图结构数据的关系挖掘方法及***
CN112949745A (zh) * 2021-03-23 2021-06-11 中国检验检疫科学研究院 多源数据的融合处理方法、装置、电子设备及存储介质
CN113568324A (zh) * 2021-06-29 2021-10-29 之江实验室 一种基于仿真演绎的知识图谱修正方法
CN113723047A (zh) * 2021-07-27 2021-11-30 山东旗帜信息有限公司 一种基于法律文件的图谱构建方法、设备及介质
CN114543796A (zh) * 2022-02-14 2022-05-27 国网电力科学研究院有限公司 一种室外定位方法、***及存储介质
CN115878713A (zh) * 2022-10-27 2023-03-31 浙江大学 一种复杂大规模sdn网络实体快速查询方法及平台

Families Citing this family (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110347810B (zh) * 2019-05-30 2022-08-19 重庆金融资产交易所有限责任公司 对话式检索回答方法、装置、计算机设备及存储介质
CN110377628A (zh) * 2019-07-23 2019-10-25 京东方科技集团股份有限公司 一种信息获取方法、装置及电子设备
CN110717076B (zh) * 2019-09-06 2024-05-28 平安科技(深圳)有限公司 节点管理方法、装置、计算机设备及存储介质
CN110837550B (zh) * 2019-11-11 2023-01-17 中山大学 基于知识图谱的问答方法、装置、电子设备及存储介质
CN111309944B (zh) * 2020-01-20 2023-07-14 同方知网数字出版技术股份有限公司 一种基于图数据库的数字人文搜索方法
CN111640218B (zh) * 2020-05-28 2022-06-21 广东电网有限责任公司 一种无人机巡检路线规划方法、装置、终端及存储介质
CN112749212A (zh) * 2021-01-20 2021-05-04 青岛以萨数据技术有限公司 一种基于六度空间理论的人员库***平台及构建方法
CN113569030A (zh) * 2021-07-29 2021-10-29 北京三快在线科技有限公司 信息查询方法、装置、设备及存储介质

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN106445988A (zh) * 2016-06-01 2017-02-22 上海坤士合生信息科技有限公司 一种大数据的智能处理方法和***
CN106897273A (zh) * 2017-04-12 2017-06-27 福州大学 一种基于知识图谱的网络安全动态预警方法
WO2017193685A1 (zh) * 2016-05-11 2017-11-16 华为技术有限公司 社交网络中数据的处理方法和装置
CN107368468A (zh) * 2017-06-06 2017-11-21 广东广业开元科技有限公司 一种运维知识图谱的生成方法及***

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102332009B (zh) * 2011-09-02 2013-09-04 北京大学 一种大规模数据集上的关系查询方法
GB201118332D0 (en) * 2011-10-24 2011-12-07 Skype Ltd Processing search queries in a network of interconnected nodes
CN107016459A (zh) * 2016-03-23 2017-08-04 西安电子科技大学 一种基于网络社区信息的点到点最短路径计算方法
CN107451210B (zh) * 2017-07-13 2020-11-20 北京航空航天大学 一种基于查询松弛结果增强的图匹配查询方法
CN108073711B (zh) * 2017-12-21 2022-01-11 北京大学深圳研究生院 一种基于知识图谱的关系抽取方法和***
CN108491556A (zh) * 2018-04-18 2018-09-04 武汉轻工大学 基于大数据降维处理的公交线路查询方法和查询设备

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2017193685A1 (zh) * 2016-05-11 2017-11-16 华为技术有限公司 社交网络中数据的处理方法和装置
CN106445988A (zh) * 2016-06-01 2017-02-22 上海坤士合生信息科技有限公司 一种大数据的智能处理方法和***
CN106897273A (zh) * 2017-04-12 2017-06-27 福州大学 一种基于知识图谱的网络安全动态预警方法
CN107368468A (zh) * 2017-06-06 2017-11-21 广东广业开元科技有限公司 一种运维知识图谱的生成方法及***

Cited By (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112231350B (zh) * 2020-10-13 2022-04-12 汉唐信通(北京)科技有限公司 一种基于知识图谱的企业商机挖掘方法和装置
CN112231350A (zh) * 2020-10-13 2021-01-15 汉唐信通(北京)科技有限公司 一种基于知识图谱的企业商机挖掘方法和装置
CN112396184A (zh) * 2020-12-01 2021-02-23 中山大学 一种基于图结构数据的关系挖掘方法及***
CN112396184B (zh) * 2020-12-01 2023-09-05 中山大学 一种基于图结构数据的关系挖掘方法及***
CN112949745A (zh) * 2021-03-23 2021-06-11 中国检验检疫科学研究院 多源数据的融合处理方法、装置、电子设备及存储介质
CN112949745B (zh) * 2021-03-23 2024-04-19 中国检验检疫科学研究院 多源数据的融合处理方法、装置、电子设备及存储介质
CN113568324B (zh) * 2021-06-29 2023-10-20 之江实验室 一种基于仿真演绎的知识图谱修正方法
CN113568324A (zh) * 2021-06-29 2021-10-29 之江实验室 一种基于仿真演绎的知识图谱修正方法
CN113723047A (zh) * 2021-07-27 2021-11-30 山东旗帜信息有限公司 一种基于法律文件的图谱构建方法、设备及介质
CN114543796A (zh) * 2022-02-14 2022-05-27 国网电力科学研究院有限公司 一种室外定位方法、***及存储介质
CN114543796B (zh) * 2022-02-14 2023-09-08 国网电力科学研究院有限公司 一种室外定位方法、***及存储介质
CN115878713B (zh) * 2022-10-27 2023-10-20 浙江大学 一种复杂大规模sdn网络实体快速查询方法及平台
CN115878713A (zh) * 2022-10-27 2023-03-31 浙江大学 一种复杂大规模sdn网络实体快速查询方法及平台

Also Published As

Publication number Publication date
CN109492111B (zh) 2023-05-30
CN109492111A (zh) 2019-03-19

Similar Documents

Publication Publication Date Title
WO2020056984A1 (zh) 最短路径查询方法、***、计算机设备和存储介质
WO2020048058A1 (zh) 基金知识推理方法、***、计算机设备和存储介质
US11010359B2 (en) Multi-entity normalization
US9916350B2 (en) Automated creation of join graphs for unrelated data sets among relational databases
US10475045B2 (en) Database management methods, systems, and devices for identifying related customer support tickets
CN110291517A (zh) 图数据库中的查询语言互操作性
US20170193116A1 (en) Indirect Filtering in Blended Data Operations
US10140325B2 (en) Data source identification mapping in blended data operations
US8775226B2 (en) Computing and managing conflicting functional data requirements using ontologies
US20180253653A1 (en) Rich entities for knowledge bases
US20240031379A1 (en) Security appliance to monitor networked computing environment
KR20150080533A (ko) 데이터 저장 시스템에서 데이터 소스 특성화
US10067965B2 (en) Hierarchic model and natural language analyzer
US11144549B2 (en) Dynamic generation of join statements for database operations
US8620941B2 (en) Apparatus and method of searching and visualizing instance path
WO2016029230A1 (en) Automated creation of join graphs for unrelated data sets among relational databases
US11100141B2 (en) Monitoring organization-wide state and classification of data stored in disparate data sources of an organization
US11423036B2 (en) Systems and methods for selecting datasets
US11615150B2 (en) Neighborhood-based entity disambiguation system and method
US10089409B2 (en) Event-triggered data quality verification
US20180150486A1 (en) Linking datasets
Athanasiou et al. Big POI data integration with Linked Data technologies.
Rashid et al. A quality assessment approach for evolving knowledge bases
US10901987B2 (en) Dynamic automatic generation of database views
CN110737779A (zh) 知识图谱的构建方法、装置、存储介质和电子设备

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 18934450

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 18934450

Country of ref document: EP

Kind code of ref document: A1