Knowledge graph construction method and device, computing device, and storage medium

Abstract

The application discloses a knowledge graph construction method, comprising the following steps: receiving an information extraction instruction, wherein the information extraction instruction is used for indicating an information extraction strategy for performing information extraction on source data of a knowledge graph; then, performing information extraction on the source data by using the information extraction strategy indicated by the information extraction instruction, and obtaining a plurality of multivariate group data, wherein each multivariate group data comprises information of an entity type used for indicating an entity, information of an entity attribute, and information of a correlation relationship; and finally, constructing a knowledge graph according to the plurality of multivariate group data, wherein the knowledge graph records entities included in the source data and the relationships between different entities. The application guarantees the application range of the knowledge graph construction method, and improves the flexibility of the knowledge graph construction.

Description

[0001] The application claims priority to the Chinese Patent Application No. 201910792526.0, filed on August 26, 2019, and entitled "A knowledge graph construction method and device", the whole content of which is incorporated herein by reference. TECHNICAL FIELD

[0002] The present application relates to the technical field of cloud computing, in particular to a knowledge graph construction method and device, a computing device and a storage medium. BACKGROUND

[0003] More and more enterprises have realized the importance of knowledge to business, and it is urgent to sort out the knowledge system in business to improve work efficiency and effect. As a representation form of knowledge organization and knowledge representation, using a knowledge graph to represent a knowledge system has become a development trend.

[0004] In the related art, when constructing a knowledge graph, a knowledge graph ontology model needs to be designed in combination with domain knowledge of a domain to which a business belongs, information extraction is then performed on data related to the business to obtain information for indicating entities in the data, and the extracted information is filled into the knowledge graph ontology to obtain the knowledge graph.

[0005] The construction process of the knowledge graph is usually implemented by a customized module, which is customized according to domain requirements of the domain to which the business belongs. However, since requirements of different domains are different, the customized module is difficult to be used for constructing knowledge graphs of different domains, resulting in poor applicability. SUMMARY

[0006] The present application provides a knowledge graph construction method and device, a computing device and a storage medium, which can solve the problem of poor applicability of the method for constructing a knowledge graph in the related art.

[0007] In a first aspect, the present application provides a knowledge graph construction method, which includes: receiving an information extraction instruction, the information extraction instruction being used to indicate an information extraction strategy for performing information extraction on source data for constructing a knowledge graph; performing information extraction on the source data by using the information extraction strategy indicated by the information extraction instruction to obtain a plurality of multi-tuple data, each multi-tuple data including information for indicating an entity type of an entity, information of an entity attribute and information of an association relationship; and constructing a knowledge graph according to the plurality of multi-tuple data, the knowledge graph recording entities included in the source data and relationships between different entities.

[0008] The knowledge graph construction method provided by the embodiments of the present application can receive an information extraction instruction, determine an information extraction strategy used for information extraction on source data for constructing a knowledge graph, perform information extraction on the source data by using the information extraction strategy to obtain a plurality of tuple data, and then construct the knowledge graph according to the plurality of tuple data. Compared with the related art, the information extraction strategy can be configured according to business requirements, and different information extraction strategies can be used for source data in different fields, so that the knowledge graph can be constructed according to the source data in different fields, the application range of the knowledge graph construction method is ensured, and the flexibility of constructing the knowledge graph is improved.

[0009] Optionally, before the knowledge graph is constructed according to the plurality of tuple data, the method can further include: obtaining a knowledge graph ontology model that needs to be used when the knowledge graph is constructed, the knowledge graph ontology model defining a standardized description of the tuple data in the knowledge graph; receiving a mapping strategy instruction, the mapping strategy instruction being used to indicate a mapping strategy of associatively mapping the plurality of tuple data according to the standardized description of the tuple data; and associatively mapping the plurality of tuple data according to the standardized description of the tuple data and the mapping strategy indicated by the mapping strategy instruction, to obtain a plurality of tuple data that are standardized described by using the standardized description of the tuple data. Correspondingly, the implementation process of constructing the knowledge graph according to the plurality of tuple data includes: constructing the knowledge graph according to the plurality of tuple data that are standardized described.

[0010] The associative mapping is also referred to as knowledge mapping. The knowledge mapping refers to establishing a mapping relationship between an extracted element and an ontology element, and standardizing description of the corresponding extracted element by using the ontology element according to the mapping relationship. The knowledge mapping can realize unified representation of the tuple data, and improves the readability of the knowledge graph.

[0011] In one implementation manner of the mapping strategy, a matching degree of each extracted element and an ontology element can be obtained. When a matching degree of a certain extracted element and an ontology element is greater than a matching degree threshold, a mapping relationship between the extracted element and the ontology element can be established, and the extracted element is indicated to be standardized described by using the ontology element.

[0012] In another implementation manner of the mapping strategy, a user can configure the mapping strategy through a terminal. The implementation process includes: the user can indicate, through the terminal, a mapping relationship between an extracted element in the tuple data and an ontology element of a standardized description defined by the knowledge graph ontology model, and indicate that the extracted element having the mapping relationship is standardized described by using the ontology element.

[0013] The user configures the mapping strategy, and the multi-tuple data is associated and mapped using the configured mapping strategy, so that the knowledge graph construction device can use different mapping strategies for different types of data, can improve the accuracy of the associated mapping of the multi-tuple data, and improves the accuracy of the knowledge graph construction.

[0014] Optionally, before constructing the knowledge graph according to the plurality of multi-tuple data, the method can further include: determining, according to the specified multi-tuple data matching strategy, different multi-tuple data including information indicating the same entity from the plurality of multi-tuple data; and performing merging processing on the different multi-tuple data including information indicating the same entity. Correspondingly, the implementation process of constructing the knowledge graph according to the plurality of multi-tuple data includes: constructing the knowledge graph according to the plurality of multi-tuple data after the merging processing.

[0015] When constructing the knowledge graph according to the plurality of source data, the representation of the information indicating the same entity can be different. If the knowledge graph is directly constructed according to the extracted multi-tuple data, the same entity using different representations can be regarded as different entities, which can cause the constructed knowledge graph to be unable to accurately reflect the content embodied by the source data. By performing merging processing on the different multi-tuple data including elements indicating the same entity, and constructing the knowledge graph according to the multi-tuple data after the merging processing, the accuracy of the constructed knowledge graph can be improved.

[0016] In an implementation manner, before determining, according to the specified multi-tuple data matching strategy, different multi-tuple data including information indicating the same entity from the plurality of multi-tuple data, the method further includes: receiving a matching strategy instruction, the matching strategy instruction being used to indicate a matching algorithm and a matching degree threshold value for judging whether the different multi-tuple data includes information indicating the same entity. Correspondingly, the implementation process of determining, according to the specified multi-tuple data matching strategy, different multi-tuple data including information indicating the same entity from the plurality of multi-tuple data includes: when it is determined according to the matching algorithm indicated by the matching strategy instruction that the matching degree of the information indicating the entity in the two multi-tuple data is not less than the matching degree threshold value, it is determined that the two multi-tuple data includes information indicating the same entity.

[0017] By selecting the matching algorithm through the matching strategy instruction, and using the selected matching algorithm to judge whether the different multi-tuple data includes elements indicating the same entity, different matching algorithms can be used for elements obtained based on data in different fields, the flexibility of knowledge mapping and the accuracy of the obtained matching degree can be improved, and the accuracy and comprehensiveness of the knowledge graph construction are improved.

[0018] Optionally, the source data comprises multi-path data of different sources. That is, the method for constructing a knowledge graph provided by the embodiments of the present application can construct a knowledge graph for multi-path data. Correspondingly, the implementation process of performing information extraction on the source data according to the information extraction strategy indicated by the information extraction instruction to obtain a plurality of multi-tuple data can include: performing information extraction on each path of data according to the information extraction strategy indicated by the information extraction instruction for information extraction on each path of data to obtain a plurality of multi-tuple data corresponding to the multi-path data, respectively. At this time, the implementation process of constructing a knowledge graph according to the plurality of multi-tuple data includes: constructing a knowledge graph according to the plurality of multi-tuple data corresponding to the multi-path data. In this way, the construction efficiency of constructing a knowledge graph according to multi-path data can be improved.

[0019] Among them, after constructing a knowledge graph according to a plurality of multi-tuple data, the method can further include: after determining that the source data is updated, performing information extraction on the incremental data in the updated source data according to the strategy indicated by the information extraction instruction to obtain a plurality of multi-tuple data corresponding to the incremental data; and updating the knowledge graph according to the plurality of multi-tuple data corresponding to the incremental data.

[0020] By incrementally updating the knowledge graph, the amount of calculation in the process of constructing a knowledge graph according to the updated source data can be reduced, and the construction efficiency of constructing a knowledge graph can be improved.

[0021] In one implementation manner, the implementation process of performing information extraction on the source data according to the information extraction strategy indicated by the information extraction instruction can include: performing information extraction on the source data according to the AI model indicated by the information extraction instruction. Among them, the AI model is a trained model, and the training samples of the AI model are labeled using the standardized description of the multi-tuple data in the knowledge graph ontology model, and the knowledge graph ontology model defines the standardized description of the multi-tuple data in the knowledge graph.

[0022] Since the training samples of the AI model are labeled using the standardized description of the multi-tuple data in the knowledge graph ontology model, when the AI model trained using the labeled samples extracts information, the multi-tuple data extracted using the AI model is information represented by the ontology elements defined in the knowledge graph ontology model, which can reduce the process of standardizing the extracted multi-tuple data according to the ontology elements, simplify the process of constructing a knowledge graph, and improve the construction efficiency of the knowledge graph.

[0023] In a second aspect, the present application provides a knowledge graph construction device, which comprises: a receiving module configured to receive an information extraction instruction, the information extraction instruction being used to indicate an information extraction strategy for information extraction on source data for constructing a knowledge graph; an extraction module configured to perform information extraction on the source data by using the information extraction strategy indicated by the information extraction instruction, to obtain a plurality of multi-tuple data, each multi-tuple data comprising information indicating an entity type of an entity, information of an entity attribute, and information of an association relationship; and a construction module configured to construct the knowledge graph according to the plurality of multi-tuple data, the knowledge graph recording entities included in the source data and relationships between different entities.

[0024] Optionally, the device further comprises: an acquisition module configured to acquire a knowledge graph ontology model required for constructing the knowledge graph, the knowledge graph ontology model defining a standardized description of the multi-tuple data in the knowledge graph; the receiving module is further configured to receive a mapping strategy instruction, the mapping strategy instruction being used to indicate a mapping strategy for associatively mapping the plurality of multi-tuple data according to the standardized description of the multi-tuple data; and a mapping module configured to associatively map the plurality of multi-tuple data according to the standardized description of the multi-tuple data and the mapping strategy indicated by the mapping strategy instruction, to obtain a plurality of multi-tuple data standardized by the standardized description of the multi-tuple data.

[0025] Correspondingly, the construction module is specifically configured to construct the knowledge graph according to the plurality of multi-tuple data after the standardized description.

[0026] Optionally, the device further comprises: a determination module configured to determine, according to a specified multi-tuple data matching strategy, different multi-tuple data comprising information indicating a same entity from the plurality of multi-tuple data; and a merging module configured to perform merging processing on the different multi-tuple data comprising the information indicating the same entity.

[0027] Correspondingly, the construction module is specifically configured to construct the knowledge graph according to the plurality of multi-tuple data after the merging processing.

[0028] Optionally, the receiving module is further configured to receive a matching strategy instruction, the matching strategy instruction being used to indicate a matching algorithm and a matching degree threshold value for judging whether the different multi-tuple data comprises information indicating a same entity.

[0029] Correspondingly, the determination module is specifically configured to determine that the two multi-tuple data comprises the information indicating the same entity when the matching degree of the information indicating the entity in the two multi-tuple data is not less than the matching degree threshold value according to the matching algorithm indicated by the matching strategy instruction.

[0030] The source data includes multiple paths of data from different sources, and the extraction module is specifically configured to perform information extraction on each path of data by using an information extraction strategy indicated by the information extraction instruction, to obtain multiple tuple data corresponding to the multiple paths of data.

[0031] The construction module is configured to construct the knowledge graph according to the multiple tuple data corresponding to the multiple paths of data.

[0032] The extraction module is further configured to perform information extraction on incremental data in the updated source data according to the strategy indicated by the information extraction instruction, to obtain multiple tuple data corresponding to the incremental data, after determining that the source data is updated.

[0033] The construction module is further configured to update the knowledge graph according to the multiple tuple data corresponding to the incremental data.

[0034] The extraction module is specifically configured to perform information extraction on the source data by using an AI model indicated by the information extraction instruction, wherein the AI model is a trained model, and training samples of the AI model are labeled by using standardized descriptions of the tuple data in a knowledge graph ontology model, and the knowledge graph ontology model defines the standardized descriptions of the tuple data in the knowledge graph.

[0035] In a third aspect, the present application provides a computing device, comprising a processor and a memory; the memory stores a computer program; when the processor executes the computer program, the computing device implements the knowledge graph construction method provided in the first aspect.

[0036] In a fourth aspect, the present application provides a non-volatile storage medium, when instructions in the storage medium are executed by a processor, the knowledge graph construction method provided in the first aspect is implemented. BRIEF DESCRIPTION OF DRAWINGS

[0037] Figure 1 is a deployment schematic diagram of a knowledge graph construction device provided by an embodiment of the present application;

[0038] Figure 2 is another deployment schematic diagram of a knowledge graph construction device provided by an embodiment of the present application;

[0039] Figure 3 is a structural schematic diagram of a computing device provided by an embodiment of the present application;

[0040] Figure 4 is a flowchart of a knowledge graph construction method provided by an embodiment of the present application;

[0041] Figure 5is a logical block diagram of constructing a knowledge graph according to two-way data provided by an embodiment of the present application;

[0042] Figure 6 is an interface schematic diagram of selecting a knowledge graph ontology model provided by an embodiment of the present application;

[0043] Figure 7 is a schematic diagram of a knowledge graph ontology model provided by an embodiment of the present application;

[0044] Figure 8 is an interface schematic diagram of selecting source data provided by an embodiment of the present application;

[0045] Figure 9 is an interface schematic diagram of selecting an information extraction strategy provided by an embodiment of the present application;

[0046] Figure 10 is an interface schematic diagram of selecting a mapping strategy provided by an embodiment of the present application;

[0047] Figure 11 is an interface schematic diagram of selecting a matching strategy provided by an embodiment of the present application;

[0048] Figure 12 is a schematic diagram of a knowledge graph provided by an embodiment of the present application;

[0049] Figure 13 is a structural schematic diagram of a knowledge graph construction device provided by an embodiment of the present application;

[0050] Figure 14 is a structural schematic diagram of a knowledge graph construction device provided by an embodiment of the present application. DETAILED DESCRIPTION

[0051] In order to make the purpose, technical scheme and advantages of the present application clearer, the embodiments of the present application will be further described in detail below with reference to the drawings.

[0052] In order to facilitate understanding of the knowledge graph construction method provided by the embodiments of the present application, the related knowledge of the knowledge graph will be introduced first.

[0053] A knowledge graph is a semantic network that describes objective things in the form of a graph. The knowledge graph is composed of a plurality of nodes and connections between different nodes. The nodes are used to represent entity types or entity attributes of entities such as persons or organizations. The connections (also referred to as edges) between the nodes represent that the entities represented by the nodes have certain association relationships. Among them, an entity can be collectively represented by an entity type, an entity attribute, and an association relationship. The association relationship between the node representing the entity type of a certain entity and the node representing the entity attribute of the entity can include an ownership relationship between the entity type and the entity attribute. The association relationship between the node representing the entity type of a certain entity and the node representing the entity type of another entity can include an external contact between the entity and the other entity.

[0054] In the embodiments of the present application, the knowledge graph can be applied to various application scenarios. For example, in an information recommendation system, information can be recommended based on the knowledge graph. Or, in a text classification process, classification can be performed based on the knowledge graph. Or, in a semantic search process, search can be performed based on the knowledge graph. Or, in a fault analysis system, for a fault that occurs, the cause of the fault can be determined according to the attributes of each entity and the association relationships between entities presented by the knowledge graph, and the analysis of the fault is realized.

[0055] An entity is a certain thing that has distinguishability and exists independently. For example, a person, a city, a plant, or a commodity, etc. An entity is the most basic element in a knowledge graph. The relationships between different entities can be different, and the entity attributes of different entities can be different.

[0056] For example, in a knowledge graph representing the basic information of an actor, the nodes can represent entity types such as family members, friends, partners, representative works, agencies, and alma mater of the actor; or the nodes can represent entity attributes such as the name, height, and nationality of the actor of the entities indicated by the entity types; the edges between the nodes representing the entity types and the nodes representing the entity attributes can represent the ownership relationship between the entity attribute and the entity type; the edges between the node representing the actor and the node representing the family members can represent the husband-wife relationship, parent-child relationship, and father-son relationship, etc. between the actor and the family members; the edges between the node representing the actor and the node representing the friends can represent the friendship between the actor and the friends; the edges between the node representing the actor and the node representing the partners can represent the cooperation relationship between the actor and the partners; the edges between the node representing the actor and the node representing the representative works of the actor can represent the ownership relationship between the actor and the representative works of the actor; the edges between the node representing the actor and the node representing the agency can represent the contract relationship between the actor and the agency; and the edges between the node representing the actor and the node representing the alma mater can represent the relationship between the actor and the alma mater.

[0057] In the knowledge graph, data can be organized by multi-tuple data. The multi-tuple data can include triple data, quadruple or quintuple, etc. The representation form of the triple data includes "node-edge-node" and "node-property name-property value". The first word in the triple can be regarded as the subject, the second word in the triple can be regarded as the predicate, and the third word in the triple can be regarded as the object. The subject-predicate-object relationship is the relationship between the first word and the third word in the triple. For example, in the triple data "Cao Cao-pseudonym-A Mo" represented by the representation form "node-property name-property value", the subject is Cao Cao, the predicate is pseudonym, and the object is A Mo. The subject-predicate-object relationship is that the pseudonym of Cao Cao is A Mo, which is the relationship between the node representing "Cao Cao" and the property value representing "A Mo".

[0058] The embodiment of the present application provides a knowledge graph construction method. By receiving an information extraction instruction, determining an information extraction strategy for information extraction of source data for constructing a knowledge graph, and using the information extraction strategy to perform information extraction on the source data to obtain a plurality of multi-tuple data, and then constructing a knowledge graph according to the plurality of multi-tuple data, compared with related technologies, the information extraction strategy can be configured according to business requirements, and different information extraction strategies can be used for source data in different fields, so that a knowledge graph can be constructed according to source data in different fields, ensuring the application range of the knowledge graph construction method and improving the flexibility of constructing the knowledge graph.

[0059] The knowledge graph construction method provided by the embodiment of the present application can be executed by a knowledge graph construction device. The knowledge graph construction device can establish a communication connection with a terminal through a wired network or a wireless network, so that the terminal can send an instruction to the knowledge graph construction device through the communication connection to control the knowledge graph construction device to execute the knowledge graph construction method provided by the embodiment of the present application according to the content indicated by the instruction. For example, the terminal can send an instruction to the knowledge graph construction device to indicate to obtain source data for constructing a knowledge graph. After receiving the instruction, the knowledge graph construction device can obtain the source data according to the instruction, and execute the knowledge graph construction method provided by the embodiment of the present application according to the source data. Alternatively, the terminal can send an information extraction instruction to the knowledge graph construction device. After receiving the information extraction instruction, the knowledge graph construction device can use the information extraction strategy indicated by the information extraction instruction to perform information extraction on the source data, and construct a knowledge graph according to the plurality of multi-tuple data extracted.

[0060] The terminal may be a smartphone, laptop, tablet, personal desktop computer, or smart camera. A client may be installed in the terminal, and the user may interact with the knowledge graph construction device through the client. Alternatively, the user may interact with the knowledge graph construction device through a web page in the terminal.

[0061] Figure 1 This is a schematic diagram of the deployment of a knowledge graph construction device provided in an embodiment of the present application. Figure 1 As shown, the knowledge graph construction device 01 can be deployed in a cloud environment. A cloud environment is an entity that uses basic resources to provide cloud services to users in a cloud computing model. The cloud environment includes a cloud data center and a cloud service platform, and the cloud data center includes a large number of basic resources owned by the cloud service provider. For example, a cloud data center includes computing resources, storage resources, network resources, etc., and the computing resources can be a large number of computing devices (such as servers). Optionally, the knowledge graph construction device 01 can be independently deployed on a server or virtual machine in a cloud data center, or the knowledge graph construction device 01 can be distributedly deployed on multiple servers in a cloud data center, or the knowledge graph construction device 01 can be distributedly deployed on multiple virtual machines in a cloud data center, or the knowledge graph construction device 01 can be distributedly deployed on servers and virtual machines in a cloud data center.

[0062] like Figure 1 As shown, the knowledge graph construction device 01 can be abstracted into a cloud service for constructing a knowledge graph by a cloud service provider on a cloud service platform. After the user purchases the cloud service on the cloud service platform, the cloud environment can use the knowledge graph construction device 01 to construct a cloud service for the knowledge graph for the user. In addition, the user can upload the source data for constructing the knowledge graph to the cloud environment through the application program interface (API) or the web interface provided by the cloud service platform on the terminal, so that the knowledge graph construction device 01 can construct the knowledge graph based on the source data. After completing the knowledge graph construction, the knowledge graph construction device 01 can send the constructed knowledge graph to the terminal used by the user, or store the knowledge graph in the cloud environment, for example: present it on the web interface of the cloud service platform for the user to view.

[0063] In addition, there are many ways to deploy the knowledge graph construction device 01. In another deployment method, the knowledge graph construction device 01 can be logically divided into multiple parts, each part has different functions, and the multiple parts can be distributed and deployed in different environments. The multiple parts deployed in different environments work together to realize the function of building a knowledge graph for the user. For example: Figure 2As shown, the plurality of parts can be respectively deployed in any two or three of the terminal computing device, the edge environment and the cloud environment. The terminal computing device includes a terminal server, a smart phone, a notebook computer, a tablet computer, a personal desktop computer, a smart camera and the like. The edge environment is an environment including a set of edge computing devices close to the terminal computing device. The edge computing device includes an edge server, an edge small station with computing power and the like.

[0064] It should be understood that the present application does not limit which parts of the knowledge graph construction device 01 are specifically deployed in which environment. In actual application, the deployment can be adapted according to the computing power of the terminal computing device, the resource occupation of the edge environment and the cloud environment or specific application requirements.

[0065] In another deployment mode of the knowledge graph construction device 01, when the knowledge graph construction device 01 is a software device, the knowledge graph construction device 01 can be published by a service provider in the form of an application program. The user can download the application program to the terminal used by the user, and use the function of the knowledge graph construction device 01 in the terminal.

[0066] In another deployment mode of the knowledge graph construction device 01, the knowledge graph construction device 01 can also be deployed on a computing device in any environment alone. As shown in the figure, Figure 3 As shown, the computing device 100 can include a bus 101, a processor 102, a communication interface 103 and a memory 104. The processor 102, the memory 104 and the communication interface 103 communicate through the bus 101.

[0067] The processor 102 can be a hardware chip, which can be an application-specific integrated circuit (ASIC), a programmable logic device (PLD) or a combination thereof. The above-mentioned PLD can be a complex programmable logic device (CPLD), a field-programmable gate array (FPGA), a generic array logic (GAL) or any combination thereof. The processor 810 can also be a general-purpose processor, for example, a central processing unit (CPU), a network processor (NP) or a combination of CPU and NP.

[0068] The memory 104 can include volatile memory, such as a random access memory (RAM). The memory 104 can also include non-volatile memory, such as read-only memory (ROM), Flash memory, HDD, or SSD. The memory 104 stores executable code for constructing a knowledge graph, which is read by the processor 102 to execute the knowledge graph construction method provided by the embodiments of the present application. The memory 104 can also include software modules and data required by other running processes, such as an operating system. The operating system can be LINUX TM , UNIX TM , WINDOWS TM , etc.

[0069] Figure 4 A flowchart of a knowledge graph construction method provided by the embodiments of the present application. The knowledge graph construction method can construct a knowledge graph according to one or more data. In the following, the construction of a knowledge graph according to multiple data, and the process of constructing a knowledge graph is executed by a knowledge graph construction device, is taken as an example to illustrate the process of constructing a knowledge graph. At the same time, in order to facilitate understanding, the embodiments of the present application also provide a logic block diagram of constructing a knowledge graph according to two data (source data 1 and source data 2, respectively) Figure 5 . As shown in Figure 4 and Figure 5 , the knowledge graph construction method includes the following steps:

[0070] Step 401, receiving a knowledge graph construction request.

[0071] When a user needs to construct a knowledge graph using a knowledge graph construction device, the user can send a knowledge graph construction request to the knowledge graph construction device through a terminal to request the construction of a knowledge graph.

[0072] Step 402, receiving a knowledge graph ontology model instruction.

[0073] The knowledge graph ontology model instruction is used to instruct a knowledge graph ontology model used for constructing a knowledge graph. The knowledge graph ontology model (also referred to as ontology) is a skeleton and basis of the knowledge graph. The knowledge graph ontology model is a standardized description of the multi-tuple data in a specific field. That is, the knowledge graph ontology specifies standardized descriptions of entity types, entity attributes, and association relationships in the knowledge graph, and standardized descriptions of elements in the multi-tuple data. Since the knowledge graph ontology specifies the standardized description of the multi-tuple data included in the knowledge graph, the knowledge graph can be constructed according to the knowledge graph ontology model, so as to avoid including useless information in the knowledge graph, and ensure that the elements such as entity types, entity attributes, and association relationships in the knowledge graph can be described in a unified manner. For ease of description, the elements in the multi-tuple data obtained through information extraction are referred to as extracted elements, and the standardized description of the elements in the multi-tuple data is referred to as an ontology element.

[0074] The user can send the knowledge graph ontology model instruction to the knowledge graph construction apparatus through the terminal to instruct the knowledge graph ontology model to be used when constructing the knowledge graph. In addition, the knowledge graph ontology model instruction can carry the knowledge graph ontology model. Alternatively, the knowledge graph ontology model instruction can carry an identification number or a storage address of the knowledge graph ontology model, so that the knowledge graph construction apparatus can obtain the corresponding knowledge graph ontology model according to the knowledge graph ontology model instruction.

[0075] In the deployment environment of the knowledge graph construction apparatus, the knowledge graph ontology model can be stored, and the stored knowledge graph ontology model can be a model constructed in the knowledge graph construction apparatus, or a model constructed in the terminal and stored in the deployment environment. In addition, to improve the flexibility of constructing the knowledge graph, the knowledge graph construction apparatus has the functions of creating the knowledge graph ontology model, modifying and deleting the created knowledge graph ontology model, and adding, deleting, and modifying the ontology elements in the knowledge graph ontology model.

[0076] In an implementation manner, the deployment environment of the knowledge graph construction apparatus can prestore a plurality of alternative knowledge graph ontology models. At this time, the user can select the knowledge graph ontology model in the setting interface of the knowledge graph construction apparatus through the terminal, and after the selection is completed, the user can execute a specified operation in the setting interface to trigger the sending of the knowledge graph ontology model instruction. For example, Figure 6 is a schematic diagram of a setting interface of a knowledge graph construction apparatus provided by an embodiment of the present application, as shown in Figure 6As shown, the user can select the knowledge graph ontology model to be used when building the knowledge graph in the setting interface, and click the "Next" button to trigger the sending of the knowledge graph ontology model instruction.

[0077] Step 403: According to the knowledge graph ontology model instruction, obtain the knowledge graph ontology model required to construct the knowledge graph.

[0078] After receiving the knowledge graph ontology model instruction, the knowledge graph construction device can obtain the knowledge graph ontology model according to the instructions of the knowledge graph ontology model instruction. For example, when the knowledge graph ontology model instruction carries the identification number of the knowledge graph ontology model, the knowledge graph construction device can search for the knowledge graph ontology model indicated by the identification number in its deployment environment based on the identification number to obtain the knowledge graph ontology model indicated by the identification number.

[0079] For example, Figure 7 Schematic diagram of the knowledge graph ontology model obtained according to the knowledge graph ontology model instruction in step 402. Figure 7 As shown in the figure, the knowledge graph ontology model defines the standardized description of entity types, standardized description of entity attributes and standardized description of association relationships that should be included in the knowledge graph. Figure 7 The entity attributes of a person (such as Figure 7 Entity attributes for songs include: name, birthday, nationality, height, and gender. Entity attributes for movies include: release date and title. Entity attributes for films include: release time and country. Relationships between characters include: spouse, clan member, parent, and child. Relationships between characters and songs include: singer. Relationships between characters and movies include: starring or directing. Relationships between movies and songs include: usage.

[0080] It should be noted that, in the process of constructing the knowledge graph, it can be determined whether to execute step 402 according to business needs. In addition, the knowledge graph construction device can be configured by default with a knowledge graph ontology model for constructing the knowledge graph. When step 402 is not executed, in step 403, the knowledge graph construction device can obtain the knowledge graph ontology model of the default configuration and use the knowledge graph ontology model of the default configuration to construct the knowledge graph. However, when executing step 402, if the knowledge graph ontology model is selected according to application requirements, different knowledge graph ontology models can be used for different fields, which can improve the adaptability of the constructed knowledge graph to the field, thereby improving the accuracy of the knowledge graph construction.

[0081] Step 404, receiving a source data indication instruction.

[0082] The terminal can send a source data indication instruction to the knowledge graph construction apparatus, where the source data indication instruction is used to indicate source data for constructing the knowledge graph. In one possible implementation, the source data indication instruction can carry the source data for constructing the knowledge graph. In another possible implementation, the source data indication instruction can carry a storage address of the source data for constructing the knowledge graph, so as to inform the knowledge graph construction apparatus to obtain the source data from the storage address indicated by the storage address.

[0083] For example, when the knowledge graph construction apparatus is deployed in a cloud environment, the user can store the source data in a cloud data center in advance through the terminal, and send the source data indication instruction to the knowledge graph construction apparatus through the terminal, where the source data indication instruction carries a storage address of the source data in the cloud data center, so as to inform the knowledge graph construction apparatus to obtain the source data from the cloud data center according to the storage address.

[0084] In addition, the source data indicated by the source data indication instruction can be preprocessed data. The preprocessing can include converting the data type of the data into a data category that can be directly used by the knowledge graph construction apparatus. For example, after the terminal stores the source data in the data center, the cloud data center can convert the data type of the source data into a JSON data format, or convert the source data into data in a comma separated values (CSV) file format, and the like, so that the knowledge graph construction apparatus can directly use the preprocessed data after obtaining the source data, without the need of data conversion, so as to reduce the data processing amount of the knowledge graph construction apparatus when constructing the knowledge graph.

[0085] Optionally, the source data indication instruction can further carry a data category of the source data, an encoding mode, and a separator used by the source data, and the like, so as to inform the knowledge graph construction apparatus of the data category of the source data, the encoding mode, and the separator used by the source data, and the like. It should be noted that the knowledge graph construction apparatus can also automatically identify the data category of the source data, the encoding mode, and the separator used by the source data, and the like, and the embodiments of the present application do not make specific limitation thereto.

[0086] Further, whether the above information needs to be carried in the source data indication instruction can be selected in a setting interface of the knowledge graph construction apparatus. In addition, after the selection is completed, a specified operation can be performed in the setting interface to trigger the sending of the source data indication instruction carrying the corresponding information. For example, Figure 8 is a schematic diagram of a setting interface of a knowledge graph construction apparatus provided by an embodiment of the present application, as shown in Figure 8As shown, the user can select one or more data required for building the knowledge graph in the setting interface, set the name of the source data, add the storage address of each piece of data, fill in the data category, coding mode and separator of the source data, and select whether to set the title line of the source data. After completing the configuration of the setting interface, the user can click the "next" button in the setting interface to trigger the sending of the source data indication instruction.

[0087] It should be noted that the embodiments of the present application do not limit the type and source of the source data for building the knowledge graph. For example, the type of source data can be table structured data or text unstructured data, etc. The source data can be data from Baidu Encyclopedia, data from Douban movie, entertainment news text data, or data from enterprise internal database or document library, etc. Moreover, the embodiments of the present application do not limit the acquisition method of the source data, for example, the data from the web page can be acquired by a distributed crawler.

[0088] Step 405, acquiring multiple pieces of data according to the source data indication instruction.

[0089] After receiving the source data indication instruction, the knowledge graph construction device can acquire the source data according to the indication of the source data indication instruction. For example, when the source data indication instruction carries the storage address of the source data, the knowledge graph construction device can acquire the source data in the storage position indicated by the storage address. Or, when the source data indication instruction carries the source data, the knowledge graph construction device can directly read the source data carried in the source data indication instruction. For example, assuming that two pieces of data are acquired according to the source data indication instruction, and the two pieces of data are both related introduction information of Zhang Ziyi, Table 1 is one piece of data acquired by the knowledge graph construction device from a website according to the source data indication instruction, and Table 2 is another piece of data acquired by the knowledge graph construction device from a database according to the source data indication instruction.

[0090] Table 1

[0091]

[0092] Table 2

[0093] Name: Zhang Ziyi Star relationship: Zhang Ziman (brother) Alias: International Zhang Nationality: China Gender: Female Occupation: Actor, producer, singer Height: 164 cm Representative work: Hero, My Father Mother, Ten Face Buried Date of birth: February 9, 1979 Song: Ten Face Buried

[0094] Step 406, receiving information extraction instruction.

[0095] The information extraction instruction is used to indicate an information extraction strategy for information extraction on the source data. The information extraction refers to extracting multi-tuple data from the source data. The multi-tuple data can include information of an entity type indicating an entity, information of an entity attribute, and information of a correlation relationship, etc. The information extraction instruction indicates an implementation manner of the information extraction strategy, which can include an algorithm identifier of an information extraction algorithm carried in the information extraction instruction. The knowledge graph construction apparatus pre-stores program instructions of a plurality of candidate information extraction algorithms. After receiving the algorithm identifier carried in the information extraction instruction, the knowledge graph construction apparatus can determine the information extraction algorithm indicated by the algorithm identifier from the plurality of candidate information extraction algorithms, and use the information extraction algorithm to perform information extraction on the source data. When the knowledge graph is constructed according to the multi-path data, the information extraction strategies for information extraction on the multi-path data can be the same or different, which is not limited in the embodiments of the present application.

[0096] In an implementable manner, the information extraction instruction can be triggered by performing a specified operation after selecting an information extraction algorithm in a setting interface of the knowledge graph construction apparatus. For example, Figure 9 is a schematic diagram of a setting interface of a knowledge graph construction apparatus provided by an embodiment of the present application. As shown in Figure 9 , the user can select a corresponding information extraction strategy for different source data in the setting interface respectively, and click the "next step" button to trigger sending of the information extraction instruction.

[0097] Step 407, information extraction is performed on each path of data by using the information extraction strategy corresponding to each path of data indicated by the information extraction instruction, to obtain a plurality of multi-tuple data corresponding to each path of data.

[0098] The information extraction strategies used for information extraction on different types of data can be different. For example, for structured data and semi-structured data, fixed rules can be used for information extraction, or an artificial intelligence (AI) model can be used for information extraction. The representation manner of the fixed rules can include representation by a general algorithm model, a pre-installed plug-in script, and a configured function plug-in, etc. Optionally, the fixed rules can be regular expressions, rule functions, or semantic-based analysis methods, etc.

[0099] For unstructured data, rules that adaptively change according to data can be used for information extraction. For example, an AI model can be used for information extraction. Moreover, before using the AI model for information extraction, the AI model can be trained using labeled samples to ensure that the AI model has better information extraction performance. Further, the labeled samples can be labeled using ontology elements in the knowledge graph ontology model. When the AI model trained using the labeled samples extracts information, the multi-tuple data extracted by the AI model is information represented using ontology elements defined in the knowledge graph ontology model, which can reduce the process of standardizing the extracted multi-tuple data according to the ontology elements, simplify the process of constructing the knowledge graph, and improve the construction efficiency of the knowledge graph.

[0100] Moreover, the knowledge graph construction apparatus can also be configured with a function plug-in custom function. The function plug-in custom function refers to reserving an input interface and an output interface for connecting a function plug-in when deploying the knowledge graph construction apparatus, and specifying conditions that the input interface and the output interface need to meet, so as to facilitate a user to customize a function plug-in according to application requirements, and use the customized function plug-in to extract information from source data when the input of the customized function plug-in meets the restriction conditions of the input interface and the output meets the restriction conditions of the output interface. By configuring the function plug-in custom function, the user can configure the function plug-in according to application requirements, which can further improve the flexibility of constructing the knowledge graph, so that the knowledge graph construction method provided by the embodiments of the present application can be applied to more knowledge graph construction scenarios, and the application range of the knowledge graph construction method is ensured.

[0101] The following takes the AI model for information extraction as an example, and describes the implementation process of information extraction for three information extraction scenarios. The three information extraction scenarios are: information extraction under mode constraints, open information extraction, and event extraction.

[0102] In the information extraction scenario under the mode constraint, each time of information extraction process extracts a multi-tuple data of a specified type. In each time of information extraction process, a predicate model, a subject model and an object model are used in turn to perform information extraction on the to-be-extracted data. The to-be-extracted data can be part of the source data, for example, can be a sentence in the source data. The predicate model is used to determine whether the to-be-extracted data contains the multi-tuple data of the specified type. The input of the predicate model is the to-be-extracted data, and the output of the predicate model is the result of whether the to-be-extracted data contains the multi-tuple data of the specified type. The subject model is used to extract the subject of the multi-tuple data of the specified type from the to-be-extracted data when the to-be-extracted data contains the multi-tuple data of the specified type. The input of the subject model is the to-be-extracted data and the type information of the multi-tuple data of the specified type. The output of the subject model is the subject of the multi-tuple data of the specified type. The object model is used to extract the object of the multi-tuple data of the specified type from the to-be-extracted data when the to-be-extracted data contains the multi-tuple data of the specified type. The input of the object model is the to-be-extracted data, the type information of the multi-tuple data of the specified type, and the subject of the multi-tuple data of the specified type. The output of the object model is the object of the multi-tuple data of the specified type.

[0103] The predicate model, the subject model and the object model each have an input layer, a feature extraction layer and an output layer. The input layer is used to divide the to-be-extracted data according to words or sentences, use vectors to represent each part of the divided data, and indicate the position of each part of the divided data in the to-be-extracted data (i.e. position embedding function). The feature extraction layer is used to extract the features of the vectors input from the input layer. The output layer is used to determine the type of each part of the divided data according to the features extracted by the feature extraction layer.

[0104] Optionally, the input layer of the predicate model, the subject model and the object model can be implemented using a Bert model (a language representation model). The feature extraction layer of the predicate model, the subject model and the object model can be implemented using a dilate gated convolutional neural network (DGCNN) model (a language representation model). The output layer of the predicate model, the subject model and the object model can be implemented using a Sigmoid function (a S-shaped function).

[0105] For example, from the sentence "The Forest Report-Autumn is a book published by the 21st Century Press in 2007, and the author is (Soviet Union) V Bianqi", the triple data contained therein is (The Forest Report-Autumn, author, V Bianqi), (The Forest Report-Autumn, publication time, 2007), (The Forest Report-Autumn, publisher, 21st Century Press), (The Forest Report-Autumn, type, book), (V Bianqi, nationality, Soviet Union), (V Bianqi, type, person), and the like. In the information extraction scene under the mode constraint, the specified type of triple data is (book, author, person), (book, publisher, publisher), (person, country, nationality), and then the results that can be extracted from the above sentence are (The Forest Report-Autumn, author, V Bianqi), (The Forest Report-Autumn, publisher, 21st Century Press), (V Bianqi, nationality, Soviet Union).

[0106] In the open information extraction scene, there is no need to limit the extraction of specified type of multi-tuple data, and the subject, predicate and object of the extracted multi-tuple data are the words directly appearing in the data to be extracted. In each information extraction process, the predicate model, the subject model and the object model are used in turn to extract information from the data to be extracted. Among them, the predicate model is used to extract the predicate of the multi-tuple data from the data to be extracted. The input of the predicate model is the data to be extracted, and the output of the predicate model is the predicate of the multi-tuple data. The subject model is used to extract the subject of the multi-tuple data from the data to be extracted. The input of the subject model is the data to be extracted and the predicate of the multi-tuple data. The output of the subject model is the subject of the multi-tuple data. The object model is used to extract the object of the multi-tuple data from the data to be extracted. The input of the object model is the data to be extracted, the subject and the predicate of the multi-tuple data. The output of the object model is the object of the multi-tuple data. Among them, the implementation of the predicate model, the subject model and the object model can be referred to the implementation of the predicate model, the subject model and the object model in the aforementioned information extraction scene under the mode constraint.

[0107] For example, from the sentence "The Forest Report-Autumn is a book published by the 21st Century Press in 2007, and the author is (Soviet Union) V Bianqi", the triple data contained therein is (The Forest Report-Autumn, author, V Bianqi), (The Forest Report-Autumn, publication time, 2007), (The Forest Report-Autumn, publisher, 21st Century Press), (The Forest Report-Autumn, type, book), (V Bianqi, nationality, Soviet Union), (V Bianqi, type, person), and the like. In the open information extraction scene, since the subject, predicate and object of the extracted multi-tuple data need to be the words directly appearing in the data to be extracted, the results that can be extracted from the above sentence are (The Forest Report-Autumn, author, V Bianqi).

[0108] In the event extraction scenario, the data extracted each time is an event composed of multiple tuples of multiple specified types. Before performing the information extraction operation, the event type and event attribute need to be defined in advance. The information extraction logic is: first, identify the trigger word and event type of the event, then extract the event element, and judge the role of each event element. In each information extraction process, the subject model, predicate model and object model are used in turn to extract information from the data to be extracted. Among them, the subject model is used to judge whether the pre-defined event type and trigger word exist in the data to be extracted. The input of the subject model is the data to be extracted. The output of the subject model is the result of whether the pre-defined event type exists in the data to be extracted. The predicate model is used to judge whether the pre-defined event attribute exists in the data to be extracted. The input of the predicate model is the data to be extracted and the type information of the pre-defined event type, and the output of the predicate model is the event attribute existing in the data to be extracted. The object model is used to extract the attribute value of the event attribute from the data to be extracted. The input of the object model is the data to be extracted, the type information of the pre-defined event type, and the attribute information of the event attribute existing in the data to be extracted. The output of the object model is the attribute value of each event attribute. The output of the subject model, predicate model and object model constitutes an event. The implementation of the predicate model, subject model and object model can be referred to the implementation of the predicate model, subject model and object model in the information extraction scenario under the mode constraint.

[0109] For example, the data to be extracted is "Banana Company will hold a new product launch conference at 10:00 am on September 12, 2021 (1:00 am on September 13, 2021 Beijing time) in the west. The launch conference is held at the newly built Steve Jobs Theater. According to the current news, Banana Company will release ichne8, ichne7s, ichne7s Plus, ichne ch 3 and new ichne TV at this launch conference". The event type is defined as "launch conference", and the event attributes include "time", "location", "company" and "product".

[0110] In the extraction process, the subject model is used to judge whether the event type "launch conference" appears in the data to be extracted. The input is the data to be extracted, and the output is the result of whether the event type "launch conference" exists in the data to be extracted. The subject model can also mark the trigger word "new product launch conference" in the data to be extracted, so as to distinguish multiple events of the same type that may appear in the data to be extracted.

[0111] The predicate model is used to judge whether the event attributes "time", "location", "company" and "product" appear in the data to be extracted according to the event type appearing in the data to be extracted. The input is the data to be extracted and the type information of the event type, and the output is the event attribute existing in the data to be extracted.

[0112] The object model is used to extract attribute values of event attributes from the data to be extracted. The input is the data to be extracted, the event type "conference", and the event attributes "time", "location", "company", and "product". The output is the attribute value of each event attribute in the data to be extracted, for example, for the event attribute "time", the output is: 10:00 am on September 12, Pacific Time, for the event attribute "location", the output is: Steve Jobs Theater, for the event attribute "company", the output is: Apple Inc., and for the event attribute "product", the output is: ichne8, ichne7s, ichne7s Plus, ichne ch 3, and new ichne TV.

[0113] According to the outputs of the subject model, the predicate model, and the object model, a plurality of triple data can be obtained: (conference, company, Banana Inc.), (conference, time, 10:00 am on September 12, Pacific Time), (conference, location, Steve Jobs Theater), (conference, product, ichne8), (conference, product, ichne7s), and the like. These triple data constitute the result of event extraction:

[0114] Event type: conference;

[0115] Company: Banana Inc.;

[0116] Time: 10:00 am on September 12, Pacific Time;

[0117] Location: Steve Jobs Theater;

[0118] Product: ichne8, ichne7s, ichne7s Plus, ichne ch 3, and ichne TV.

[0119] It should be noted that in the process of constructing the knowledge graph, it can be determined according to business requirements whether to perform step 406. Moreover, the knowledge graph construction device can be configured with a default information extraction strategy, and when step 406 is not performed, the knowledge graph construction device can use the default information extraction strategy to perform information extraction on the source data in step 407. However, by selecting an information extraction strategy for information extraction on the source data, the knowledge graph construction device can use different information extraction strategies for source data in different fields, improve the accuracy of information extracted from the source data, ensure the accuracy of the knowledge graph constructed according to the source data in different fields, ensure the application scope of the knowledge graph construction method, and improve the flexibility of constructing the knowledge graph.

[0120] Step 408, receiving a mapping strategy instruction.

[0121] The mapping strategy instruction is used to instruct a mapping strategy of associatively mapping (also referred to as knowledge mapping) a plurality of tuples of data according to the ontology elements. The knowledge mapping refers to establishing a mapping relationship between the extracted elements and the ontology elements, and standardizing the description of the corresponding extracted elements by using the ontology elements according to the mapping relationship. For example, when the formalized expression of the subject in the tuple data defined by the knowledge graph ontology model is "name", if the subject in the extracted tuple data is "name", the mapping relationship between "name" and "name" can be established according to the mapping strategy, and "name" is standardized to "name" according to the mapping relationship. When the knowledge graph is constructed according to the multi-path data, the mapping strategies corresponding to the multi-path data can be the same or different, and the embodiments of the present application do not make specific limitations thereto.

[0122] In an implementation manner of the mapping strategy, the knowledge graph construction apparatus can acquire the matching degree of each extracted element and ontology element. When the matching degree of a certain extracted element and an ontology element is greater than a matching degree threshold, the knowledge graph construction apparatus can establish the mapping relationship between the extracted element and the ontology element, and instruct to use the ontology element to standardize the description of the extracted element. For example, when the matching degree of the extracted element "name" and the ontology element "name" is greater than the matching degree threshold, the mapping relationship between "name" and "name" can be established, and "name" is standardized to "name" according to the mapping relationship.

[0123] At this time, the mapping strategy instruction is used to instruct to establish the mapping relationship between the ontology elements and the extracted elements according to the matching degree, and to acquire the matching degree algorithm used by the matching degree. For example, the mapping strategy instruction can instruct to establish the mapping relationship between the ontology elements and the extracted elements according to the matching degree, and the matching degree algorithm used to acquire the matching degree can be an edit distance similarity algorithm.

[0124] In another implementation manner of the mapping strategy, the user can configure the mapping strategy in the setting interface of the knowledge graph construction apparatus through the terminal. The implementation process includes that the user can instruct the mapping relationship between the extracted elements and the ontology elements through the terminal, and instruct to use the ontology elements to standardize the description of the extracted elements having the mapping relationship therewith. After the user completes the configuration, the mapping strategy instruction can be triggered by executing a specified operation in the setting interface. Moreover, since the knowledge graph ontology model is determined in step 403, the ontology elements defined by the knowledge graph ontology model are determined, and therefore, the process of configuring the mapping strategy is essentially a process of respectively instructing the extracted elements having the mapping relationship with different ontology elements according to the determined ontology elements.

[0125] An example of the mapping strategy is shown in Table 1. Figure 10 is a schematic diagram of a setting interface of a knowledge graph construction apparatus provided by an embodiment of the present application, as shown in Figure 10As shown, in the setting interface, the user can add an extracted element having a mapping relationship with an ontology element, respectively. For example, for an entity type (i.e., ontology entity type) "name" in the known ontology element, an entity type (i.e., extracted entity type) "name" having a mapping relationship with the ontology element can be added to map the entity type. For an association relationship (i.e., ontology association relationship) in the ontology element, an association relationship (i.e., extracted association relationship) having a mapping relationship with the ontology element can be added to map the association relationship. For an entity attribute (i.e., ontology entity attribute) in the known ontology element, an entity attribute (i.e., extracted entity attribute) having a mapping relationship with the ontology element can be added to map the entity attribute. In addition, the type of the knowledge graph can be mapped according to the category (i.e., ontology category) of the knowledge graph ontology model. After the configuration is completed, the "next" button can be clicked to trigger the sending of the mapping strategy instruction.

[0126] In step 409, according to the mapping strategy indicated by the mapping strategy instruction and the standardized description of the multi-tuple data, the multi-tuple data obtained by extracting each piece of data is associated and mapped respectively to obtain a plurality of multi-tuple data standardized by the standardized description of the multi-tuple data.

[0127] After the knowledge graph construction device obtains the mapping strategy instruction, the knowledge graph construction device can perform knowledge mapping on the plurality of multi-tuple data according to the ontology element according to the mapping strategy indicated by the mapping strategy instruction, to obtain a plurality of multi-tuple data standardized by the ontology element. Through knowledge mapping, the extracted element can be standardized by the ontology element defined by the knowledge graph ontology model, realizing the unified representation of the extracted element and improving the readability of the knowledge graph.

[0128] It should be noted that in the process of constructing the knowledge graph, it can be determined whether to perform step 408 according to the business requirements. In addition, the mapping strategy can be configured by default in the knowledge graph construction device, and when step 408 is not performed, the knowledge graph construction device can use the default mapping strategy to associate and map the multi-tuple data in step 409. However, by selecting the mapping strategy and using the selected mapping strategy to associate and map the multi-tuple data, the knowledge graph construction device can use different mapping strategies for different types of data, which can improve the accuracy of associating and mapping the multi-tuple data and improve the accuracy of constructing the knowledge graph.

[0129] In step 410, a matching strategy instruction is received.

[0130] When constructing a knowledge graph according to multiple source data, the representation of information indicating the same entity can be different. If a knowledge graph is directly constructed according to the extracted multiple tuple data, the same entity represented in different ways can be regarded as different entities, which can cause the constructed knowledge graph to fail to accurately reflect the content embodied by the source data. Therefore, before constructing a knowledge graph according to multiple tuple data, it can also be determined whether the different multiple tuple data includes elements indicating the same entity, and the different multiple tuple data including elements indicating the same entity is processed (also referred to as knowledge conflation) to facilitate the construction of a knowledge graph according to the multiple tuple data after the processing, thereby improving the accuracy of the constructed knowledge graph. For example, the entity type information extracted from the source data shown in Table 1 is "name: Zhang Ziyi", and the entity type information extracted from the source data shown in Table 2 is "name: International Chapter". Although the representation is different, both of them are used to indicate the same entity. At this time, the two can be knowledge conflated.

[0131] The matching strategy instruction is used to indicate the matching algorithm and the matching degree threshold value for determining whether the different multiple tuple data includes elements indicating the same entity. The knowledge graph construction device can obtain the matching degree of the elements in the different multiple tuple data according to the matching degree algorithm. When the matching degree of the elements in the different multiple tuple data is not less than the matching degree threshold value, it is determined that the elements in the different multiple tuple data are used to indicate the same entity. At this time, the elements in the different multiple tuple data indicating the same entity can be merged.

[0132] In an implementable manner, the programs of multiple matching algorithms can be pre-stored in the deployment environment of the knowledge graph construction device. At this time, the matching algorithm to be used can be selected in the setting interface of the knowledge graph construction device, and after the selection is completed, the matching strategy instruction is triggered by performing a specified operation in the setting interface. For example, Figure 11 is a schematic diagram of a setting interface of a knowledge graph construction device provided by an embodiment of the present application, as Figure 11 shown, the user can select the matching algorithm and the matching degree threshold value to be used when knowledge fusion is performed on different elements in the setting interface. Moreover, the matching algorithm and the matching degree threshold value can also be set for different entity attributes of an entity respectively, and for an entity having multiple entity attributes, the judgment result when determining whether the entity and other entities are the same entity can be the "integration" of the algorithm results of the matching algorithms corresponding to the different entity attributes of the entity. For example, it can be the intersection of the algorithm results of the matching algorithms corresponding to the different entity attributes of the entity. Similarly, each attribute can also be configured with multiple matching algorithms. After the setting is completed, the "next" button can be clicked to trigger the matching strategy instruction.

[0133] In step 411, according to the multi-tuple data matching strategy indicated by the matching strategy instruction, different multi-tuple data including elements indicating the same entity in the standardized description of the plurality of multi-tuple data are determined, and the different multi-tuple data including elements indicating the same entity are merged to obtain the plurality of multi-tuple data after the merging processing.

[0134] The merging processing of the different multi-tuple data including elements indicating the same entity refers to using the same representation to represent the same entity using different representations, so that the representation of the elements indicating the same entity is the same.

[0135] For example, the triple data obtained by information extraction according to the source data shown in Table 1 are (Zhang Ziyi, height, 164 cm), (Zhang Ziyi, gender, female), (Zhang Ziyi, nationality, China), (Zhang Ziyi, birthday, February 9, 1979), (Zhang Ziyi, brother and sister, Zhang Ziman), (Zhang Ziyi, starring, My Mother and Father), and (Zhang Ziyi, starring, Crouching Tiger, Hidden Dragon). The triple data obtained by information extraction according to the source data shown in Table 2 are (International Chapter, height, 164 cm), (International Chapter, gender, female), (International Chapter, brother and sister, Zhang Ziman), (International Chapter, starring, My Mother and Father), (International Chapter, starring, Hero), (Zhang Ziyi, starring, House of Flying Daggers), and (International Chapter, singer, House of Flying Daggers). After knowledge fusion according to the multi-tuple data matching strategy indicated by the matching strategy instruction, the following triple data are obtained: (Zhang Ziyi, height, 164 cm), (Zhang Ziyi, gender, female), (Zhang Ziyi, nationality, China), (Zhang Ziyi, birthday, February 9, 1979), (Zhang Ziyi, brother and sister, Zhang Ziman), (Zhang Ziyi, starring, My Mother and Father), (Zhang Ziyi, starring, House of Flying Daggers), (Zhang Ziyi, starring, Hero), and (Zhang Ziyi, singer, House of Flying Daggers).

[0136] It should be noted that in the process of constructing the knowledge graph, whether step 410 is performed can be determined according to business requirements. In addition, the matching algorithm and the corresponding matching degree threshold can be configured by default in the knowledge graph construction device. When step 410 is not performed, in step 411, the knowledge graph construction device can use the default configured matching algorithm and the corresponding matching degree threshold to determine whether the different multi-tuple data include elements indicating the same entity. However, by selecting a matching algorithm and using the selected matching algorithm to determine whether the different multi-tuple data include elements indicating the same entity, the knowledge graph construction device can use different matching algorithms for elements obtained based on data in different fields, which can improve the flexibility of knowledge mapping and the accuracy of obtaining matching degrees, and improve the accuracy and comprehensiveness of knowledge graph construction.

[0137] Step 412, constructing a knowledge graph according to the plurality of tuples of data after the merging processing.

[0138] The knowledge graph records the entities included in the source data and the relationships between different entities. The steps 401 to 411 are all preparatory work for constructing the knowledge graph. After the preparatory work is completed, the knowledge graph can be constructed according to the plurality of tuples of data after the merging processing. The process of constructing the knowledge graph according to the plurality of tuples of data can be understood as a process of connecting the plurality of tuples of data into a semantic network according to the relationships between the elements in the plurality of tuples of data after the merging processing. Moreover, each node in the semantic network corresponds to an entity type or an entity attribute in a tuple of data, the relationships between the nodes correspond to the information of the association relationships in the tuple of data, and the starting point of the arrow between the nodes corresponds to the element used as the subject in the tuple of data, and the ending point of the arrow corresponds to the element used as the object in the tuple of data.

[0139] For example, Figure 12 FIG. 1 is a schematic diagram of the knowledge graph constructed according to the plurality of tuples of data after the merging processing in step 411. As shown in FIG. 1, the knowledge graph records the entity types, entity attributes and association relationships in the tuples of data for indicating the entities. The knowledge graph represents the source data in Table 1 and Table 2 in the form of a graph, improves the visualization degree of the source data, and improves the convenience of analyzing the source data. Figure 12

[0140] Step 413, after determining that the source data is updated, performing information extraction on the incremental data in the updated source data according to the strategy indicated by the information extraction instruction, obtaining a plurality of tuples of data corresponding to the incremental data, and updating the knowledge graph according to the plurality of tuples of data corresponding to the incremental data.

[0141] When the source data of the constructed knowledge graph is updated, the incremental data of the updated source data relative to the source data can be obtained, and the constructed knowledge graph can be updated according to the incremental data to obtain the knowledge graph corresponding to the updated source data. For example, the information extraction can be performed on the incremental data to obtain a plurality of tuples of data corresponding to the incremental data, then the knowledge mapping can be performed on the plurality of tuples of data corresponding to the incremental data, then the knowledge fusion can be performed on the plurality of tuples of data corresponding to the incremental data after the association mapping, and then the knowledge graph can be updated according to the plurality of tuples of data after the knowledge fusion. Through the incremental updating of the knowledge graph, the calculation amount in the process of constructing the knowledge graph according to the updated source data can be reduced, and the construction efficiency of the knowledge graph can be improved.

[0142] ​To sum up, the knowledge graph construction method provided by the embodiments of the present application receives an information extraction instruction, determines an information extraction strategy used for information extraction on source data for constructing a knowledge graph, extracts information from the source data by using the information extraction strategy to obtain a plurality of tuple data, and then constructs a knowledge graph according to the plurality of tuple data. Compared with the related art, the information extraction strategy can be configured according to business requirements, and different information extraction strategies can be used for source data in different fields, so that a knowledge graph can be constructed according to source data in different fields, the application range of the knowledge graph construction method is ensured, and the flexibility of constructing a knowledge graph is improved.

[0143] The steps of the knowledge graph construction method provided by the embodiments of the present application can be appropriately adjusted, and the steps can be appropriately increased or decreased according to circumstances. For example, whether to perform steps 402, 406, 408 and 410 can be selected according to application requirements. Any person skilled in the art can easily think of changes within the technical range disclosed in the present application, which should be covered within the protection scope of the present application, and therefore will not be described in detail.

[0144] The embodiments of the present application also provide a knowledge graph construction device. As shown in Figure 13 The knowledge graph construction device 80 can include:

[0145] The receiving module 801 is configured to receive an information extraction instruction, and the information extraction instruction is used to indicate an information extraction strategy used for information extraction on source data for constructing a knowledge graph.

[0146] The extraction module 802 is configured to extract information from the source data by using the information extraction strategy indicated by the information extraction instruction to obtain a plurality of tuple data, and each tuple data includes information indicating an entity type of an entity, information of an entity attribute and information of an association relationship.

[0147] The construction module 803 is configured to construct a knowledge graph according to the plurality of tuple data, and the knowledge graph records entities included in the source data and relationships between different entities.

[0148] Optionally, as shown in Figure 14 The knowledge graph construction device 80 further includes:

[0149] The acquisition module 804 is configured to acquire a knowledge graph ontology model used for constructing a knowledge graph, and the knowledge graph ontology model defines a standardized description of the tuple data in the knowledge graph.

[0150] The receiving module 801 is further configured to receive a mapping strategy instruction, and the mapping strategy instruction is used to indicate a mapping strategy for associatively mapping the plurality of tuple data according to the standardized description of the tuple data.

[0151] The mapping module 805 is configured to perform associated mapping on the plurality of multi-tuple data according to a mapping strategy indicated by the standardized description of the multi-tuple data and the mapping strategy instruction, to obtain a plurality of multi-tuple data standardized by the standardized description of the multi-tuple data.

[0152] Correspondingly, the construction module 803 is specifically configured to construct the knowledge graph according to the plurality of multi-tuple data after the standardization.

[0153] Optionally, as shown in Figure 14 The knowledge graph construction apparatus 80 further includes:

[0154] The determination module 806 is configured to determine, according to a specified multi-tuple data matching strategy, different multi-tuple data including information indicating the same entity from the plurality of multi-tuple data.

[0155] The merging module 807 is configured to perform merging processing on the different multi-tuple data including information indicating the same entity.

[0156] Correspondingly, the construction module 803 is specifically configured to construct the knowledge graph according to the plurality of multi-tuple data after the merging processing.

[0157] Optionally, the receiving module 801 is further configured to receive a matching strategy instruction, the matching strategy instruction being used to indicate a matching algorithm and a matching degree threshold value for judging whether different multi-tuple data include information indicating the same entity.

[0158] Correspondingly, the determination module 806 is specifically configured to determine that two multi-tuple data include information indicating the same entity when a matching degree of information indicating an entity in the two multi-tuple data is not less than the matching degree threshold value according to the matching algorithm indicated by the matching strategy instruction.

[0159] Optionally, the source data includes multi-path data with different sources, and the extraction module 802 is specifically configured to perform information extraction on each path of data according to an information extraction strategy indicated by the information extraction instruction for information extraction on each path of data, to obtain a plurality of multi-tuple data corresponding to the multi-path data respectively.

[0160] Correspondingly, the construction module 803 is specifically configured to construct the knowledge graph according to the plurality of multi-tuple data corresponding to the multi-path data.

[0161] Optionally, the extraction module 802 is further configured to perform information extraction on incremental data in the updated source data according to a strategy indicated by the information extraction instruction after determining that the source data is updated, to obtain a plurality of multi-tuple data corresponding to the incremental data.

[0162] Correspondingly, the constructing module 803 is further configured to update the knowledge graph according to the plurality of tuple data corresponding to the incremental data.

[0163] Optionally, the extracting module 802 is specifically configured to perform information extraction on the source data by using an AI model indicated by the information extraction instruction.

[0164] The AI model is a trained model, and the training samples of the AI model are labeled by using the standardized description of the plurality of tuple data in the knowledge graph ontology model, and the knowledge graph ontology model defines the standardized description of the plurality of tuple data in the knowledge graph.

[0165] In summary, the knowledge graph constructing apparatus provided in the embodiments of the present application receives an information extraction instruction by using a receiving module, determines an information extraction strategy for performing information extraction on source data for constructing a knowledge graph, extracts a plurality of tuple data from the source data by using the information extraction strategy, and then constructs a knowledge graph according to the plurality of tuple data. Compared with related technologies, the information extraction strategy can be configured according to business requirements, and different information extraction strategies can be used for source data in different fields, so that a knowledge graph can be constructed according to source data in different fields, the application range of the knowledge graph construction method is ensured, and the flexibility of constructing a knowledge graph is improved.

[0166] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the apparatus and the module described above can refer to the corresponding process in the foregoing method embodiments, which will not be described herein.

[0167] The embodiments of the present application further provide a computing device, which comprises a processor and a memory. The memory stores a computer program. When the processor executes the computer program, the computing device implements the knowledge graph construction method provided in the embodiments of the present application. The computing device can be a server or a terminal. The structure of the computing device can be referred to the structure of the computing device in the foregoing method embodiments, which will not be described herein. Figure 3

[0168] Optionally, the computing device can work on an AI platform and a big data platform, so as to use the AI platform to construct, train and deploy the AI model used in the knowledge graph construction method provided in the embodiments of the present application, obtain source data from the big data platform, and use the big data platform to process data.

[0169] The embodiments of the present application further provide a storage medium, which is a non-volatile computer readable storage medium. When the instructions in the storage medium are executed by a processor, the knowledge graph construction method provided in the embodiments of the present application is implemented.

[0170] ​The embodiment of the present application further provides a computer program product containing instructions, which, when the computer program product is executed on a computer, causes the computer to execute the knowledge graph construction method provided by the embodiment of the present application.

[0171] Those skilled in the art can understand that all or part of the steps of the above-mentioned embodiments can be completed by hardware, or by program instructing relevant hardware to complete, and the program can be stored in a computer readable storage medium. The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk.

[0172] In the embodiment of the present application, the terms "first", "second" and "third" are only used for descriptive purposes, and cannot be understood as indicating or implying relative importance. The term "at least one" means one or more, and the term "multiple" means two or more, unless otherwise explicitly limited.

[0173] In the present application, the term "and / or" only describes the association relationship of the associated objects, which means that there can be three relationships, for example, A and / or B can mean that A exists alone, A and B exist together, and B exists alone. In addition, the character " / " in this paper generally represents that the front and rear associated objects are a "or" relationship.

[0174] The above is only an optional embodiment of the present application, and is not used to limit the present application. Any modification, equivalent replacement, improvement, etc. within the concept and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for constructing a knowledge graph, characterized in that, The method comprises: receiving a user-triggered information extraction instruction, the information extraction instruction being used to indicate an information extraction strategy for information extraction on source data for constructing a knowledge graph, the information extraction strategy being used to indicate an artificial intelligence (AI) model for information extraction on the source data, the information extraction strategy for information extraction on source data in different fields being different, wherein the AI model is a trained model, and the training samples of the AI model are labeled using standardized descriptions of multi-tuple data in a knowledge graph ontology model, the knowledge graph ontology model defining the standardized descriptions of multi-tuple data in the knowledge graph, and the knowledge graph ontology models used in different fields being different; performing information extraction on the source data using the information extraction strategy indicated by the information extraction instruction to obtain a plurality of multi-tuple data, each multi-tuple data comprising information indicating an entity type of an entity, information of an entity attribute, and information of an association relationship, the entity being used to indicate a transaction that is distinguishable and independently existing, the transaction indicated by the entity comprising one or more of a person, a city, a plant, or a commodity, and the association relationship comprising one or more of a possession relationship or an external contact; according to the standardized descriptions of the multi-tuple data and a mapping strategy, associatively mapping the plurality of multi-tuple data to obtain a plurality of multi-tuple data that are standardized described using the standardized descriptions of the multi-tuple data, different types of source data using different mapping strategies; constructing the knowledge graph according to the plurality of multi-tuple data that are standardized described, the knowledge graph recording entities included in the source data and relationships between different entities, the knowledge graph comprising a plurality of nodes and connections between different nodes, the nodes being used to represent entity types or entity attributes of the entities, and the connections between the nodes representing association relationships between the entities represented by the nodes.

2. The method of claim 1, wherein, Before the associatively mapping the plurality of multi-tuple data according to the standardized descriptions of the multi-tuple data and the mapping strategy, the method further comprises: obtaining a knowledge graph ontology model needed for constructing the knowledge graph; receiving a mapping strategy instruction, the mapping strategy instruction being used to indicate a mapping strategy for associatively mapping the plurality of multi-tuple data according to the standardized descriptions of the multi-tuple data.

3. The method according to claim 1 or 2, characterized in that, Before the constructing the knowledge graph according to the plurality of multi-tuple data that are standardized described, the method further comprises: determining, according to a specified multi-tuple data matching strategy, different multi-tuple data comprising information indicating a same entity from the plurality of multi-tuple data that are standardized described; performing merging processing on the different multi-tuple data comprising information indicating the same entity; the constructing the knowledge graph according to the plurality of multi-tuple data that are standardized described comprises: constructing the knowledge graph according to the plurality of multi-tuple data that are standardized described and are subjected to the merging processing.

4. The method of claim 3, wherein, Before the determining, according to the specified multi-tuple data matching strategy, different multi-tuple data including information indicating the same entity from the standardized description of the plurality of multi-tuple data, the method further comprises: receiving a matching strategy instruction, the matching strategy instruction being used for indicating a matching algorithm and a matching degree threshold value for judging whether different multi-tuple data include information indicating the same entity; the determining, according to the specified multi-tuple data matching strategy, different multi-tuple data including information indicating the same entity from the standardized description of the plurality of multi-tuple data, comprises: when it is determined, according to the matching algorithm indicated by the matching strategy instruction, that the matching degree of information indicating an entity in two multi-tuple data is not less than the matching degree threshold value, determining that the two multi-tuple data include information indicating the same entity.

5. The method of any of claims 1, 2, or 4, wherein, The source data comprises multi-path data of different sources, and the information extraction is performed on the source data by using the information extraction strategy indicated by the information extraction instruction to obtain a plurality of multi-tuple data, which comprises: respectively performing information extraction on each path of data by using the information extraction strategy indicated by the information extraction instruction for information extraction on each path of data to obtain a plurality of multi-tuple data corresponding to the multi-path data respectively; the constructing, according to the plurality of multi-tuple data corresponding to the multi-path data, the knowledge graph, comprises: constructing the knowledge graph according to the plurality of multi-tuple data corresponding to the multi-path data.

6. The method of any of claims 1, 2, or 4, wherein, After the constructing, according to the plurality of multi-tuple data after the standardized description, the knowledge graph, the method further comprises: after determining that the source data is updated, performing information extraction on incremental data in the updated source data according to the strategy indicated by the information extraction instruction to obtain a plurality of multi-tuple data corresponding to the incremental data; updating the knowledge graph according to the plurality of multi-tuple data corresponding to the incremental data.

7. A knowledge graph construction apparatus, characterized by comprising: The device comprises: a receiving module configured to receive an information extraction instruction triggered by a user, the information extraction instruction being used for indicating an information extraction strategy for performing information extraction on source data for constructing a knowledge graph, the information extraction strategy being used for indicating an artificial intelligence (AI) model for performing information extraction on the source data, the information extraction strategy for different fields being different, wherein the AI model is a trained model, and training samples of the AI model are labeled by using standardized descriptions of multi-tuple data in a knowledge graph ontology model, the knowledge graph ontology model defining the standardized descriptions of the multi-tuple data in the knowledge graph, and the knowledge graph ontology models used in different fields being different; extracting information from the source data according to an information extraction strategy indicated by the information extraction instruction to obtain a plurality of tuple data, each of the tuple data including information indicating an entity type of an entity, information of an entity attribute, and information of a correlation relationship, the entity being used to indicate a transaction having distinguishability and independent existence, the transaction indicated by the entity including one or more of a person, a city, a plant, or a commodity, and the correlation relationship including one or more of an ownership relationship or an external contact; mapping the plurality of tuple data according to a standardized description of the tuple data and a mapping strategy to obtain a plurality of tuple data standardized according to the standardized description of the tuple data, different types of source data using different mapping strategies; constructing the knowledge graph according to the plurality of tuple data after the standardization, the knowledge graph recording entities included in the source data and relationships between different entities, the knowledge graph including a plurality of nodes and connections between different nodes, the nodes being used to represent the entity type or the entity attribute of the entity, and the connections between the nodes representing the correlation relationship between the entities represented by the nodes.

8. The apparatus of claim 7, wherein, The device further includes: an obtaining module configured to obtain a knowledge graph ontology model required when the knowledge graph is constructed; the receiving module is further configured to receive a mapping strategy instruction, the mapping strategy instruction being used to indicate a mapping strategy of associatively mapping the plurality of tuple data according to the standardized description of the tuple data.

9. The apparatus of claim 7 or 8, wherein, The device further includes: a determining module configured to determine, according to a specified tuple data matching strategy, different tuple data including information indicating a same entity from the plurality of tuple data after the standardization; a merging module configured to perform merging processing on the different tuple data including the information indicating the same entity; the constructing module is specifically configured to: construct the knowledge graph according to the plurality of tuple data after the merging processing.

10. The device of claim 9, wherein the receiving module is further configured to receive a matching strategy instruction, the matching strategy instruction being used to indicate a matching algorithm and a matching degree threshold value of judging whether different tuple data include information indicating a same entity; the determining module is specifically configured to: determine that the two tuple data include the information indicating the same entity when it is determined according to the matching algorithm indicated by the matching strategy instruction that a matching degree of the information indicating the entity in the two tuple data is not less than the matching degree threshold value.

11. The apparatus of any of claims 7, 8, or 10, wherein, The source data includes a plurality of pieces of data having different sources, and the extracting module is specifically configured to: perform information extraction on each piece of data according to an information extraction strategy indicated by the information extraction instruction for performing information extraction on each piece of data to obtain a plurality of tuple data corresponding to the plurality of pieces of data respectively; the constructing module is specifically configured to: construct the knowledge graph according to the plurality of tuple data corresponding to the plurality of pieces of data.

12. The apparatus of any one of claims 7, 8 or 10, characterized in that, the extraction module is further configured to, after determining that the source data is updated, perform information extraction on incremental data in the updated source data according to a strategy indicated by the information extraction instruction, to obtain a plurality of tuples of data corresponding to the incremental data; the construction module is further configured to update the knowledge graph according to the plurality of tuples of data corresponding to the incremental data.

13. A computing device, comprising: the computing device comprises a processor and a memory; the memory stores a computer program; when the processor executes the computer program, the computing device implements the knowledge graph construction method of any one of claims 1 to 6.

14. A non-volatile storage medium, comprising: when the instructions in the storage medium are executed by the processor, the knowledge graph construction method of any one of claims 1 to 6 is implemented.

Images