WO2022048368A1 - 基于知识图谱的推荐方法、装置、计算机设备及存储介质 - Google Patents

基于知识图谱的推荐方法、装置、计算机设备及存储介质 Download PDF

Info

Publication number
WO2022048368A1
WO2022048368A1 PCT/CN2021/109760 CN2021109760W WO2022048368A1 WO 2022048368 A1 WO2022048368 A1 WO 2022048368A1 CN 2021109760 W CN2021109760 W CN 2021109760W WO 2022048368 A1 WO2022048368 A1 WO 2022048368A1
Authority
WO
WIPO (PCT)
Prior art keywords
node
nodes
analyzed
neighbor
layer
Prior art date
Application number
PCT/CN2021/109760
Other languages
English (en)
French (fr)
Inventor
曹合心
Original Assignee
深圳壹账通智能科技有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 深圳壹账通智能科技有限公司 filed Critical 深圳壹账通智能科技有限公司
Publication of WO2022048368A1 publication Critical patent/WO2022048368A1/zh

Links

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90Details of database functions independent of the retrieved data types
    • G06F16/95Retrieval from the web
    • G06F16/953Querying, e.g. by the use of web search engines
    • G06F16/9535Search customisation based on user profiles and personalisation
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/36Creation of semantic tools, e.g. ontology or thesauri
    • G06F16/367Ontology

Definitions

  • the embodiments of the present application relate to the field of data knowledge graphs, and in particular, to a knowledge graph-based recommendation method, apparatus, computer equipment, and storage medium.
  • the relationship strength between network nodes needs to be calculated, so as to obtain the recommendation result based on the correlation between network nodes.
  • path search path search, similarity between nodes, etc. are usually used.
  • path search can better reflect the relationship between two points, and plays an important role in the calculation of relationship strength and relationship prediction.
  • the path information that first reaches the target point is used to calculate the strength of the relationship.
  • the disadvantage of this method is that it only considers information from a local perspective and does not consider the relationship from an overall perspective. Even if the shortest path between two points is long, if there are many paths between two points, the strength of the relationship is still strong.
  • the purpose of the embodiments of the present application is to provide a recommendation method, device, computer equipment and storage medium based on knowledge graph, which is used to solve the problem that when the recommendation result is obtained by the strength of the relationship between the nodes of the knowledge graph, the high efficiency and low efficiency cannot be obtained. cost to get accurate recommendation results.
  • the embodiment of the present application provides a recommendation method based on a knowledge graph, where the knowledge graph includes a plurality of nodes, and the method includes:
  • Recommendation data between the node to be analyzed and other nodes is generated according to the recommended path.
  • the knowledge graph includes user attribute nodes and product attribute nodes, and the method further includes:
  • the user attribute node of the knowledge graph is generated according to the user attribute table, and the product attribute node of the knowledge graph is generated according to the product attribute table, the user attribute table includes the user's product purchase history, and the product attribute table includes the corresponding product. attribute information; correspondingly, the generating recommendation data between the node to be analyzed and other nodes according to the recommendation path includes:
  • the product attribute node information in the recommendation path is recommended to the user attribute node in the recommendation path.
  • determining the node to be analyzed in the plurality of nodes setting the weight value of the relationship between the node to be analyzed and the first-layer neighbor node as a, and randomly calculating any neighbor node of the node to be analyzed.
  • the relationship weight value I of the Lth layer node to the node to be analyzed includes:
  • the relationship weight between the node to be analyzed and the neighbor node of the first layer is assigned as a;
  • Randomly select any node in the neighbor nodes of the first layer define it as the first node, define the adjacent node of the first node as the neighbor node of the second layer, and associate the first node with the second layer neighbor node;
  • the relationship weight between the node to be analyzed and the neighbor node of the second layer is assigned as a-(a/6).
  • the method further includes:
  • any node in the second-layer neighbor nodes has a neighbor node
  • any node in the second-layer neighbor nodes is randomly selected and defined as the second node, and the adjacent node of the second node is defined as
  • the third-layer neighbor node is associated with the second node and the third-layer neighbor node;
  • the relationship weight between the node to be analyzed and the neighbor node of the third layer is assigned as a-2*(a/6);
  • the method further includes:
  • the first node of calculates the relationship weight value L-1 between the new first node and the neighbor nodes at different levels of the new first node.
  • the method further includes:
  • the average value of the corresponding relationship weight values is taken as the final relationship weight value of the node to be analyzed and other nodes.
  • the method further includes:
  • the recommendation data between the multiple nodes generated according to the recommendation path is uploaded to the blockchain.
  • the present application further provides a recommendation device based on a knowledge graph, the knowledge graph includes a plurality of nodes, and the recommendation device based on the knowledge graph includes:
  • the weight calculation module is used to determine the node to be analyzed in the plurality of nodes, set the weight value of the relationship between the node to be analyzed and the first-layer neighbor node as a, and randomly calculate any neighbor of the node to be analyzed.
  • the relationship weight value I of the node to the L-th layer node of the node to be analyzed, where I a-(L-1)*(a/6);
  • a path search module configured to select a node whose relationship weight value between the node to be analyzed and other nodes is greater than a preset value to generate a recommended path
  • a recommendation output module configured to generate recommendation data between the node to be analyzed and other nodes according to the recommendation path.
  • the present application also provides a computer device, comprising a memory, a processor and a computer program stored in the memory and running on the processor, the processor implements the above method when executing the computer program A step of.
  • the present application also provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the steps of the above method are implemented.
  • the knowledge graph nodes are randomly calculated for multiple times, starting from a single node, the relationship strength values between this node and all other nodes are obtained, and then the relationship strength values that satisfy the predetermined requirements are selected.
  • the nodes with the relationship strength value are set to generate recommendation paths, which solves the problem that accurate recommendation results cannot be obtained with high efficiency and low cost.
  • FIG. 1 is a schematic diagram of an application environment of an embodiment of the present application.
  • FIG. 2 is a schematic flowchart of a recommendation method based on a knowledge graph according to Embodiment 1 of the present application;
  • FIG. 3 is a schematic diagram of a node of a knowledge graph according to an example of Embodiment 1 of the present application;
  • Fig. 4 is the schematic flow chart of calculating the relationship weight value between a plurality of nodes in Fig. 2;
  • FIG. 5 is a schematic diagram of program modules of Embodiment 2 of the recommendation device based on knowledge graph of the present application;
  • FIG. 6 is a schematic diagram of the hardware structure of Embodiment 3 of the computer device of the present application.
  • FIG. 1 a schematic diagram of an implementation environment of an embodiment of the present application is shown.
  • the implementation environment includes: a client 10 and a server 12 .
  • the client 10 is an electronic device with a network access function, and the device may be a smart phone, a tablet computer, a personal computer, or the like.
  • the client 10 is installed with a program 11 that can access the server 12 .
  • the program 11 logs in to the server 12 with an account and a password, the user can perform specific operations and input on the server 12 .
  • the server 12 is a server, a server cluster formed by several servers, or a cloud computing center.
  • a program 13 is stored in the server 12, and the program 13 includes a front-end module and a back-end module.
  • the front-end module and the back-end module can call each other through the interface.
  • the user logs in through the account and password in the program 11 or accesses the server 12 through the account and password. After the program 13 is created, specific operations and inputs can be performed on the program 13.
  • the server 12 stores multiple databases and data tables. Different databases can acquire data in the databases through triggering instructions of the user 10 .
  • the database and data table include, but are not limited to, a user attribute table and a product attribute table.
  • the database may also be stored in different servers in the server 12 cluster, or in different network nodes connected through a network, or stored locally in the user segment 10 .
  • the client 10 and the server 12 are connected through a network, and the network may include multiple network nodes, and the network may be the Internet, a local area network or a blockchain network.
  • the knowledge graph-based recommendation method in this embodiment of the present application can be applied to program 11 or program 13 alone, or can be applied to both program 11 and program 13 in a distributed manner, or stored in multiple nodes of the network in the form of blockchain middle.
  • the relationship strength values between this node and all other nodes can be obtained, and then Selecting nodes that satisfy the preset relationship strength value generates a recommended path, which solves the problem that accurate recommendation results cannot be obtained with high efficiency and low cost.
  • the knowledge graph includes multiple nodes and includes the following steps:
  • Step S100 generating a user attribute node of the knowledge graph according to a user attribute table, generating a product attribute node of the knowledge graph according to a product attribute table, the user attribute table including the user's product purchase history, and the product attribute table including The attribute information of the corresponding product;
  • a knowledge graph is a graph-based data structure consisting of nodes and edges, each node represents an "entity”, and each edge is a "relationship" between entities.
  • data is extracted from different data sources and added to the knowledge graph according to certain rules.
  • the data sources include structured data and unstructured data.
  • the user attribute nodes of the knowledge graph are generated according to the user attribute table
  • the product attribute nodes of the knowledge graph are generated according to the product attribute table.
  • the user attribute table includes the user's product purchase history, and the product attribute table includes attribute information of the corresponding product.
  • the knowledge graph is generated by extracting the user and the historical records of the products purchased by the user as the data source, as shown in FIG. 3 .
  • Nodes 1-14 are entities representing users, and the edges between nodes are relationships between users, and correlations are established based on whether there are products with the same or related attributes in the user's purchased product history records.
  • consumption data of other users that are related to the user can then be recommended to a specific user.
  • Step S200 Determine the node to be analyzed in the plurality of nodes, set the weight value of the relationship between the node to be analyzed and the neighbor node of the first layer as a, and randomly calculate any neighbor node of the node to be analyzed to the node to be analyzed.
  • a is a preset value, such as 1 or 100. Since the solution in the embodiment of the present application is to obtain the relationship weight between nodes, which is essentially a comparison value, any preset value a that is convenient for calculation can be realized.
  • L is greater than or equal to 1, when the node to be analyzed is determined, the knowledge graph is a tree structure, the node to be analyzed is the root node of the knowledge graph, and the first-layer neighbor node of the node to be analyzed is the first layer of the knowledge graph hierarchy, and the Lth layer A node is the corresponding layer extending from the root node to the leaf node.
  • the so-called tree structure is a data structure with hierarchical relationships between elements, and an inverted tree can be used to represent the logical relationship.
  • the so-called root node is the top node of the tree, which is further divided into child nodes. When it is continuously subdivided until there are no more child nodes, it is a leaf node.
  • the second-layer neighbor node of the node to be analyzed is the second layer of the knowledge graph hierarchy, and the nodes in the second layer are the child nodes of the first-layer neighbor node of the root node, that is, the adjacent layer of the first-layer neighbor node The node is determined as the second-level neighbor node.
  • node 1 is the root node
  • nodes 2, 3, 4, and 5 are the first-level neighbor nodes of node 1
  • nodes 6, 7, 8, 10, 11, 12, and 13 are nodes of node 1.
  • the second-level neighbor node is the first-level neighbor node.
  • the node to be analyzed in the knowledge graph has only one layer of neighbor nodes, and the weight of the relationship between the node to be analyzed and other nodes is given the same weight value, that is, the weight of the node to be analyzed and other nodes in the knowledge graph are equal. , the recommendation data with the same priority of the node to be analyzed and other nodes is directly generated.
  • step S200 determine the node to be analyzed among the plurality of nodes, set the weight value of the relationship between the node to be analyzed and the neighbor nodes of the first layer as a, and randomly calculate all the nodes to be analyzed.
  • Step S201 initializing the relationship weight value of the node to be analyzed to other nodes
  • Step S202 associating the node to be analyzed and the neighbor node of the first layer
  • nodes 2, 3, 4, and 5 are first-layer neighbor nodes of node 1.
  • nodes 1, 6, and 7 are the first-layer neighbor nodes of node 2; if it is determined that node 4 is the node to be analyzed, then nodes 1, 8, 9, and 14 is the first-layer neighbor node of node 4, if it is determined that 14 is the node to be analyzed, then only node 9 is the first-layer neighbor node of node 14, and so on, when it is determined that any node as shown in Figure 3 is to be analyzed. Nodes, by obtaining the adjacent nodes of the node to be analyzed as the first-layer neighbor nodes, the node positions of the corresponding first-layer neighbor nodes can be obtained and associated.
  • Step S203 assigning the weight of the relationship between the node to be analyzed to the neighbor node of the first layer as a;
  • node 1 is a node to be analyzed, and the neighbor nodes of node 1 are associated with nodes 2, 3, 4, and 5, the position of nodes 2, 3, 4, and 5 is a.
  • a is preset as 1, and the relationship weight value I becomes ⁇ 1, 1, 1, 1, 0, 0, 0, 0, 0, 0, 0 ⁇ at this time.
  • a may also be preset to other values, such as 100.
  • Step S204 randomly select any node in the neighbor nodes of the first layer, define it as the first node, define the adjacent node of the first node as the neighbor node of the second layer, associate the first node with the The second-level neighbor node;
  • nodes 2, 3, 4, and 5 are first-layer neighbor nodes, and one node is randomly selected from nodes 2, 3, 4, and 5 as the first node. For example, after random selection, node 4 is selected as first node.
  • the random selection of the first node is implemented by invoking a random selector.
  • the principle of the random selector is to randomly select a piece of data from a large amount of or known data through a random selection algorithm, for example, by selecting an algorithm variable Add a perturbation to , and make the perturbation obey a specific probability distribution (pseudo-random), by inputting the identifiers of nodes 2, 3, 4, and 5, and then return to get the corresponding random result.
  • node 4 is randomly selected as the first node, and the neighbor node (except the one already associated) associated with node 4 is node 8 .
  • Step S205 assigning a weight of the relationship between the node to be analyzed and the neighbor node of the second layer as a-(a/6).
  • the strength of the relationship decays as the path increases.
  • the relationship strength can be approximately 0, so the decreasing coefficient of each layer is 1/6 ⁇ 0.167, and the relationship weight I is updated as:
  • I ⁇ 1,1,1,1,0,0,1-0.167,0,0,0,0,0,0 ⁇ , i.e. ⁇ 1,1,1,1,0,0,0.833,0, 0,0,0,0,0 ⁇ .
  • any node in the second-layer neighbor nodes still has a neighbor node, then continue to perform the following steps, and if any node in the second-layer neighbor nodes does not have a neighbor node, do not Perform the following steps, and jump to step S209.
  • node 8 also has neighbor node 9, and the following steps need to be performed. If in step 204, node 2 is randomly selected as the first node, and the neighbor nodes associated with node 2, node 6 and node 2 are randomly selected as the first node. 7. If node 6 and node 7 do not have neighbor nodes, then jump to step S209.
  • node 2 is randomly selected as the first node, and the neighbor nodes 6 and 7 associated with node 2 are associated with node 2, the relationship weight value between node 6 and node 7 is obtained by calculation, and the corresponding I is:
  • Step S206 when any node in the neighbor nodes of the second layer has a neighbor node, randomly select any node in the neighbor nodes of the second layer, define it as the second node, and define the phase of the second node.
  • the adjacent node is a third-layer neighbor node, and is associated with the second node and the third-layer neighbor node;
  • node 9 is selected as the second node
  • the neighbor node of node 9 is node 14
  • node 14 is the third-layer neighbor node.
  • the second node is another node and there is no neighbor node afterward, then jump to step S209.
  • Step S207 assigning a weight of the relationship between the node to be analyzed and the neighbor node of the third layer as a-2*(a/6).
  • the neighbors are randomly sampled on the basis of node 8, which is node 9.
  • l becomes ⁇ 1, 1, 1, 1, 0, 0, 0.833,1-0.167-0.167,0,0,0,0 ⁇ , i.e. ⁇ 1, 1, 1, 1, 0, 0, 0.833, 0.666,0,0,0,0,0 ⁇ ;
  • Step S208 when the third-layer neighbor node still has neighbor nodes, then continue to assign values to the unassigned nodes as a-(L-1)*(a/6) according to the hierarchical order from small to large, where i is the current value.
  • i is the current value.
  • node 9 has neighbor node 14, and the neighbor is randomly sampled on the basis of node 9, which is node 14.
  • the sixth layer and subsequent nodes are assigned a value of 0, because according to the 6-degree space theory, in the sixth layer node Yes, the relationship weight of the node has been approximated to 0.
  • Step S209 if the neighbor nodes of different levels of the first node have completed the relationship weight assignment or the first node has no neighbor nodes, then randomly select other nodes in the neighbor nodes of the first layer and define them as new first nodes. , and calculate the relationship weight value between the new first node and the neighbor nodes at different levels of the new first node.
  • the average value of the corresponding relationship weight values is taken as the final relationship weight value of the node to be analyzed and other nodes.
  • the principle is the same as above, randomly select other nodes except the first neighbor node among the neighbor nodes of the node to be analyzed, and repeat steps S201-S208; another weight combination can be obtained.
  • node 2 is sampled at the neighbor node of 1 in this calculation, it will eventually hit nodes 2, 3, 4, 5, 6, and 7.
  • node 2 is randomly selected as the first node, it will be associated with node 2.
  • the neighbor node 6 and node 7 of the relationship weight value of node 6 and node 7 is obtained by calculation, the corresponding I is: ⁇ 1,1,1,1,0,1-0.167,1-0.167,0,0,0 ,0,0,0 ⁇ , that is, ⁇ 1,1,1,1,0,0.833,0.833,0,0,0,0,0 ⁇ node 2,3,4,5,6,7 weights correspond to is ⁇ 1,1,1,0.833,0.833 ⁇ );
  • nodes 2, 3, 4, and 5 are neighbor nodes of the first layer, they are assigned a fixed value a, so no matter how many times nodes 2, 3, 4, and 5 are randomly calculated, the relationship weight value between them and node 1 are all fixed.
  • the relationship weight of node 6 and node 7 is 0 after the first random calculation, the relationship weight value after the second random calculation is 0.833, and the average value is 0.4165. After the first random calculation, the relationship between node 8 and node 14 is The weight values are 0.666 and 0.499, respectively. After the second random calculation, the relationship weight value is 0, and the average value is 0.333 and 0.2495.
  • the number of random calculations can be set to be greater than 2 times, and the set of relationship weights from other nodes to the remaining nodes can be calculated by the same method.
  • node 10, node 11, node 12 and node 13 are not hit.
  • the number of calculations can be increased to increase the probability of hitting node 10, node 11, node 12 and node 13.
  • node 3 is hit as the first node
  • node 5 is hit as the first node
  • all nodes are finally hit
  • Step S300 selecting a node whose relationship weight value between the node to be analyzed and other nodes is greater than a preset value to generate a recommended path;
  • Step S400 generating recommendation data between the node to be analyzed and other nodes according to the recommended path.
  • the final node relationship weight value is ⁇ 1, 1, 1, 1, 0.4165, 0.4165, 0.4165, 0.333, 0, 0, 0, 0, 0.2495 ⁇ ; select when the relationship strength value between nodes is greater than 0.4, recommend the product attribute node information in the recommended path to the recommended path
  • the user attribute node of the corresponding node for example, recommends the user's product purchase record of the corresponding node to other associated users.
  • the relationship weight value of node 2, node 3, node 4, node 5, node 7, and node 7 is greater than 0.4, then node 2, node 3, node 4, node 5, node 7, node 8 and node 1 are generated. recommended data between.
  • the product attribute node information in the recommended path is recommended to the user attribute node in the recommended path, for example, it is determined that the node to be analyzed is a product attribute node, that is, node 1 is a product attribute node, That is, the product purchase record information of a specific user, in order to obtain the recommended user rule corresponding to the product, it can be recommended to nodes whose relationship weight value is greater than the preset value, such as node 2, node 3, node 4, node 5, node 7, node User attribute node in 8.
  • the final node relationship weight value is ⁇ 1, 1, 1, 1, 0.20825, 0.20825, 0.4165, 0.1665, 0.20825, 0.20825, 0.20825, 0.20825, 0.12475 ⁇
  • the default value is set to 0.2
  • step S400 after generating the recommendation data between the multiple nodes according to the recommendation path, further includes:
  • the corresponding summary information is obtained by generating the recommendation data between the multiple nodes based on the recommendation path.
  • the summary information is obtained by hashing the recommendation data between the multiple nodes generated by the recommendation path. For example, it is obtained by using the sha256s algorithm.
  • Uploading summary information to the blockchain ensures its security and fairness and transparency to users.
  • the user equipment can download the summary information from the blockchain, so as to verify whether the recommendation data between the multiple nodes generated by the recommendation path is tampered with.
  • the blockchain referred to in this example is a new application mode of computer technology such as distributed data storage, point-to-point transmission, consensus mechanism, and encryption algorithm.
  • Blockchain essentially a decentralized database, is a series of data blocks associated with cryptographic methods. Each data block contains a batch of network transaction information to verify its Validity of information (anti-counterfeiting) and generation of the next block.
  • the blockchain can include the underlying platform of the blockchain, the platform product service layer, and the application service layer.
  • a recommendation method based on knowledge graph of the present application by performing probability calculation on multiple random jumps of knowledge graph nodes, starting from a single node, through multiple calculations, the relationship strength values between this node and all other nodes can be obtained, and then selected Nodes that satisfy the preset relationship strength value generate recommended paths, which solves the problem that accurate recommendation results cannot be obtained with high efficiency and low cost.
  • FIG. 5 shows a schematic diagram of program modules of the knowledge graph-based recommendation device of the present application.
  • the knowledge graph-based recommendation apparatus 20 may include or be divided into one or more program modules, and one or more program modules are stored in a storage medium and executed by one or more processors,
  • the program modules referred to in the embodiments of the present application refer to a series of computer program instruction segments capable of performing specific functions, and are more suitable for describing the execution process of the knowledge graph-based recommendation device 20 in the storage medium than the programs themselves. The following description will specifically introduce the functions of each program module in this embodiment:
  • the weight calculation module 202 is used to determine the node to be analyzed in the plurality of nodes, set the weight value of the relationship between the node to be analyzed and the first-layer neighbor node as a, and randomly calculate any one of the nodes to be analyzed.
  • the weight value I of the relationship between the neighbor node and the L-th layer node of the node to be analyzed, where I a-(L-1)*(a/6);
  • a path search module 204 configured to select a node whose relationship weight value between the node to be analyzed and other nodes is greater than a preset value to generate a recommended path;
  • a recommendation output module 206 configured to generate recommendation data between the node to be analyzed and other nodes according to the recommendation path.
  • the computer device 2 is a device that can automatically perform numerical calculation and/or information processing according to pre-set or stored instructions.
  • the computer equipment 2 may be a rack-type server, a blade-type server, a tower-type server or a cabinet-type server (including an independent server, or a server cluster composed of multiple servers) and the like.
  • the computer device 2 at least includes, but is not limited to, a memory 21 , a processor 22 , a network interface 23 , and a knowledge graph-based recommendation device 20 that can communicate with each other through a system bus. in:
  • the memory 21 includes at least one type of computer-readable storage medium, and the readable storage medium includes a flash memory, a hard disk, a multimedia card, a card-type memory (for example, SD or DX memory, etc.), a random access memory ( RAM), static random access memory (SRAM), read only memory (ROM), electrically erasable programmable read only memory (EEPROM), programmable read only memory (PROM), magnetic memory, magnetic disk, optical disk, etc.
  • the memory 21 may be an internal storage unit of the computer device 2 , such as a hard disk or a memory of the computer device 2 .
  • the memory 21 may also be an external storage device of the computer device 2, such as a plug-in hard disk equipped on the computer device 2, a smart memory card (Smart Media Card, SMC), Secure Digital (SD) card, Flash Card (Flash Card), etc.
  • the memory 21 may also include both the internal storage unit of the computer device 2 and its external storage device.
  • the memory 21 is generally used to store the operating system and various application software installed in the computer device 2 , such as the program code of the knowledge graph-based recommendation apparatus 20 in the second embodiment.
  • the memory 21 can also be used to temporarily store various types of data that have been output or will be output.
  • the processor 22 may be a central processing unit (Central Processing Unit) in some embodiments. Processing Unit, CPU), controller, microcontroller, microprocessor, or other data processing chip.
  • the processor 22 is typically used to control the overall operation of the computer device 2 .
  • the processor 22 is configured to run the program code or process data stored in the memory 21 , for example, run the knowledge graph-based recommendation apparatus 20 to implement the knowledge graph-based recommendation method of the first embodiment.
  • the network interface 23 may include a wireless network interface or a wired network interface, and the network interface 23 is generally used to establish a communication connection between the server 2 and other electronic devices.
  • the network interface 23 is used to connect the server 2 with an external terminal through a network, and establish a data transmission channel and a communication connection between the server 2 and the external terminal.
  • the network can be an intranet (Intranet), the Internet (Internet), a global system for mobile communications (Global System of Mobile communication, GSM), Wideband Code Division Multiple Access (Wideband Code Division) Multiple Access, WCDMA), 4G network, 5G network, Bluetooth (Bluetooth), Wi-Fi and other wireless or wired networks.
  • FIG. 6 only shows the computer device 2 having components 20-23, but it should be understood that it is not a requirement to implement all of the components shown, and that more or less components may be implemented instead.
  • the knowledge graph-based recommendation device 20 stored in the memory 21 may also be divided into one or more program modules, and the one or more program modules are stored in the memory 21 and are configured by One or more processors (processor 22 in this embodiment) execute to complete the present application.
  • This embodiment also provides a computer-readable storage medium, which may be non-volatile or volatile, such as flash memory, hard disk, multimedia card, card-type storage (for example, SD or DX memory, etc.), random access memory (RAM), static random access memory (SRAM), read only memory (ROM), electrically erasable programmable read only memory (EEPROM), programmable read only memory (PROM), magnetic memory , magnetic disks, optical discs, servers, App application malls, etc., on which computer-readable instructions are stored, and when executed by the processor, implement corresponding functions.
  • the computer-readable storage medium of this embodiment is used to store computer-readable instructions for a knowledge graph-based recommendation apparatus, and when executed by a processor, implements the knowledge graph-based recommendation method of Embodiment 1.
  • the blockchain referred to in this application is a new application mode of computer technologies such as distributed data storage, point-to-point transmission, consensus mechanism, and encryption algorithm.
  • Blockchain essentially a decentralized database, is a series of data blocks associated with cryptographic methods. Each data block contains a batch of network transaction information to verify its Validity of information (anti-counterfeiting) and generation of the next block.
  • the blockchain can include the underlying platform of the blockchain, the platform product service layer, and the application service layer.

Landscapes

  • Engineering & Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Animal Behavior & Ethology (AREA)
  • Computational Linguistics (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

一种基于知识图谱的推荐方法、装置、计算机设备及存储介质,包括:确定所述多个节点中的待分析节点,将所述待分析节点与第一层邻居节点的关系权重值置为a,并随机计算所述待分析节点的任一邻居节点到所述待分析节点的第L层节点的关系权重值I,其中I=a-(L-1)*(a/6)(S200);选择所述待分析节点到其他节点的关系权重值大于预设值的节点生成推荐路径(S300);根据所述推荐路径生成所述待分析节点与其他节点之间的推荐数据(S400)。本方法通过对知识图谱节点多次随机计算,从单节点出发,获取到其它节点的关系强度值,然后选择满足预设关系强度值的节点生成推荐路径,解决了不能高效率低成本得到精确推荐结果的问题。

Description

基于知识图谱的推荐方法、装置、计算机设备及存储介质
本申请要求于2020年09月02日提交中国专利局、申请号为202010907984.7,发明名称为“基于知识图谱的推荐方法、装置、计算机设备及存储介质”的中国专利申请的优先权,其全部内容通过引用结合在本申请中。
技术领域
本申请实施例涉及数据知识图谱领域,尤其涉及一种基于知识图谱的推荐方法、装置、计算机设备及存储介质。
背景技术
在基于知识图谱的推荐装置中,需要计算网络节点间的关系强度,从而得到基于网络节点之间相关性的推荐结果,在计算网络节点间关系强度时,通常采用路径搜索、节点间相似度等多种方法,其中采用路径搜索的方法,更能够反映两点之间的关联情况,对于关系强度计算和关系预测具有重要作用。
现有的路径搜索方法主要包括:
采用最短路径衡量两点间关系强弱时,采用最先到达目标点的路径信息,计算关系强度。然而此类方法的弊端在于仅从局部的角度考虑信息,未从整体角度考虑关联情况,即使两点最短路径较长,若两点之间存在的路径很多,关系强度仍然偏强。
采用全路径方法计算两点之间的所有路径时,发明人发现,此类方法虽然从全局角度考虑,但是对于大规模图谱,两点间可存在百万条以上路径,执行一次具有较高的时间和空间成本,通常只能作为查询使用,无法一次性计算图中任意两点的关系权重。
因此,现有技术在通过知识图谱节点之间关系强度得到推荐结果时,还存在不能高效率低成本得到精确推荐结果的问题。
发明内容
有鉴于此,本申请实施例的目的是提供一种基于知识图谱的推荐方法、装置、计算机设备及存储介质,用于解决在通过知识图谱节点之间关系强度得到推荐结果时,不能高效率低成本得到精确推荐结果的问题。
为实现上述目的,本申请实施例提供了一种基于知识图谱的推荐方法,所述知识图谱包括多个节点,所述方法包括:
确定所述多个节点中的待分析节点,将所述待分析节点与第一层邻居节点的关系权重值置为a,并随机计算所述待分析节点的任一邻居节点到所述待分析节点的第L层节点的关系权重值I,其中I=a-(L-1)*(a/6);
选择所述待分析节点到其他节点的关系权重值大于预设值的节点生成推荐路径;
根据所述推荐路径生成所述待分析节点与其他节点之间的推荐数据。
进一步的,所述知识图谱包括用户属性节点和产品属性节点,所述方法还包括:
根据用户属性表生成所述知识图谱的用户属性节点,根据产品属性表生成所述知识图谱的产品属性节点,所述用户属性表包括用户的产品购买历史记录,所述产品属性表包括对应产品的属性信息;对应的,所述根据所述推荐路径生成所述待分析节点与其他节点之间的推荐数据包括:
将所述推荐路径中的产品属性节点信息推荐给所述推荐路径中的用户属性节点。
进一步的,所述确定所述多个节点中的待分析节点,将所述待分析节点与第一层邻居节点的关系权重值置为a,并随机计算所述待分析节点的任一邻居节点到所述待分析节点的第L层节点的关系权重值I包括:
初始化所述待分析节点到其他节点的关系权重值;
关联所述待分析节点和所述第一层邻居节点;
对所述待分析节点到所述第一层邻居节点的关系权重赋值为a;
随机选择所述第一层邻居节点中的任一节点,定义为第一节点,定义所述第一节点的相邻节点为第二层邻居节点,关联所述第一节点和所述第二层邻居节点;
对所述待分析节点到所述第二层邻居节点的关系权重赋值为a-(a/6)。
进一步的,所述对所述待分析节点到所述第二层邻居节点的关系权重赋值为a-(a/6)之后还包括:
当所述第二层邻居节点中的任一节点存在邻居节点,则随机选择所述第二层邻居节点中的任一节点,定义为第二节点,定义所述第二节点的相邻节点为第三层邻居节点,关联所述第二节点和所述第三层邻居节点;
对所述待分析节点到所述第三层邻居节点的关系权重赋值为a-2*(a/6);
当所述第L层邻居节点还存在邻居节点,且L大于等于6,则对第6层及之后的节点赋值为0。
进一步的,所述对选择所述待分析节点到其他节点的关系权重值大于预设值的节点生成推荐路径之前还包括:
若所述第一节点的不同层级邻居节点已完成关系权重赋值或所述第一节点没有邻居节点,则随机选择所述第一层邻居节点中除了所述第一节点的其他节点,定义为新的第一节点,计算所述新的第一节点与所述新的第一节点不同层级邻居节点的关系权重值L-1。
进一步的,所述选择关系权重值大于预设值的节点生成推荐路径之前还包括:
若所述待分析节点和其他节点的存在多个关系权重值,则将对应的关系权重值取平均值作为所述待分析节点和其他节点的最终关系权重值。
进一步的,所述根据所述推荐路径生成所述多个节点之间的推荐数据之后还包括:
将根据所述推荐路径生成所述多个节点之间的推荐数据上传至区块链中。
为实现上述目的,本申请还提供一种基于知识图谱的推荐装置,所述知识图谱包括多个节点,所述基于知识图谱的推荐装置包括:
权重计算模块,用于确定所述多个节点中的待分析节点,将所述待分析节点与第一层邻居节点的关系权重值置为a,并随机计算所述待分析节点的任一邻居节点到所述待分析节点的第L层节点的关系权重值I,其中I=a-(L-1)*(a/6);
路径搜索模块,用于选择所述待分析节点到其他节点的关系权重值大于预设值的节点生成推荐路径;
推荐输出模块,用于根据所述推荐路径生成所述待分析节点与其他节点之间的推荐数据。
为实现上述目的,本申请还提供一种计算机设备,包括存储器、处理器以及存储在存储器上并可在所述处理器上运行的计算机程序,所述处理器执行所述计算机程序时实现上述方法的步骤。
为实现上述目的,本申请还提供计算机可读存储介质,其上存储有计算机程序,所述计算机程序被处理器执行时实现上述方法的步骤。
本申请实施例的基于知识图谱的推荐方法、装置、计算机设备及存储介质,通过对知识图谱节点多次随机计算,从单节点出发,获取该节点到其它所有的关系强度值,然后选择满足预设关系强度值的节点生成推荐路径,解决了不能高效率低成本得到精确推荐结果的问题。
附图说明
图1为本申请实施例的应用环境示意图;
图2为本申请实施例一的基于知识图谱的推荐方法的流程示意图;
图3为本申请实施例一示例的知识图谱的节点示意图;
图4为图2中计算多个节点之间的关系权重值的流程示意图;
图5为本申请基于知识图谱的推荐装置之实施例二的程序模块示意图;
图6为本申请计算机设备之实施例三的硬件结构示意图。
本申请目的的实现、功能特点及优点将结合实施例,参照附图做进一步说明。
具体实施方式
为了使本申请的目的、技术方案及优点更加清楚明白,以下结合附图及实施例,对本申请进行进一步详细说明。应当理解,此处所描述的具体实施例仅用以解释本申请,并不用于限定本申请。基于本申请中的实施例,本领域普通技术人员在没有做出创造性劳动前提下所获得的所有其他实施例,都属于本申请保护的范围。
需要说明的是,在本申请中涉及“第一”、“第二”等的描述仅用于描述目的,而不能理解为指示或暗示其相对重要性或者隐含指明所指示的技术特征的数量。由此,限定有“第一”、“第二”的特征可以明示或者隐含地包括至少一个该特征。另外,各个实施例之间的技术方案可以相互结合,但是必须是以本领域普通技术人员能够实现为基础,当技术方案的结合出现相互矛盾或无法实现时应当认为这种技术方案的结合不存在,也不在本申请要求的保护范围之内。
请参阅图1,示出了本申请实施例的实施环境示意图。该实施环境包括:用户端10和服务端12。
用户端10是具有网络访问功能的电子设备,该设备可以是智能手机、平板电脑或者个人计算机等。
其中,用户端10安装有可以访问服务端12的程序11,用户在程序11通过账号和密码登录访问服务端12后,即可对服务端12执行特定操作和输入。
服务端12是一台服务器、若干服务器构称的服务器集群或者云计算中心。服务端12中存储了程序13,程序13包括前端模块和后端模块,前端模块和后端模块可通过接口互相调用,用户在程序11通过账号和密码登录或者通过账号和密码登录访问服务端12的程序13后,即可对程序13执行特定操作和输入。
在本实施例中,服务端12中存储有多个数据库和数据表。不同的数据库可以通过用户10的触发指令获取数据库中的数据。
在本实施例中,所述数据库和数据表包括不限于用户属性表和产品属性表。
在其他实施例中,数据库也可以存储在服务端12集群中的不同服务器,或者通过网络连接的不同网络节点中,或者存储在用户段10本地。
用户端10和服务端12通过网络连接,网络中可包括多个网络节点,网络可以是互联网,局域网或者区块链网络。
本申请实施例的基于知识图谱的推荐方法可单独应用于程序11或程序13中,也可以分布式的同时应用于程序11和程序13中,或者以区块链形式存储于网络的多个节点中。
实施例一
本实施例的一种基于知识图谱的推荐方法,通过对知识图谱节点多次随机跳转进行概率计算,能够从单节点出发,通过多次计算,获取该节点到其它所有的关系强度值,然后选择满足预设关系强度值的节点生成推荐路径,解决了不能高效率低成本得到精确推荐结果的问题。
请参考图2,本实施例的一种基于知识图谱的推荐方法,所述知识图谱包括多个节点,包括以下步骤:
步骤S100,根据用户属性表生成所述知识图谱的用户属性节点,根据产品属性表生成所述知识图谱的产品属性节点,所述用户属性表包括用户的产品购买历史记录,所述产品属性表包括对应产品的属性信息;
知识图谱是一种基于图的数据结构,由节点和边组成,每个节点表示一个“实体”,每条边为实体与实体之间的“关系”。
知识图谱构建的过程中,通过把数据从不同的数据源中抽取出来,然后按一定的规则加入到知识图谱中,数据源包括结构化的数据和非结构化的数据。其中,根据用户属性表生成所述知识图谱的用户属性节点,根据产品属性表生成所述知识图谱的产品属性节点。
所述用户属性表包括用户的产品购买历史记录,所述产品属性表包括对应产品的属性信息。
具体的,在本实施例中,通过将用户和用户购买的产品历史记录作为数据源抽取出来作为数据源,生成知识图谱,如图3所示。
节点1-14为代表用户的实体,节点之间的边为用户之间的关系,通过用户的购买产品历史记录中是否有相同或者具有相关属性的产品进行关联,并建立相关性。
根据用户购买产品的消费习惯建立用户之间的关联并生成知识图谱后,然后就可以对特定用户推荐和该用户具有相关性的其他用户的消费数据。
但是现有技术中,如果只采取节点之间最短路径衡量两点间关系强弱时,仅从局部的角度考虑信息,即使两点最短路径较长,若两点之间存在的路径很多,仍有关系强度偏强的节点符合推荐条件而没有被推荐,从而使得到推荐的结果不准确。
如果采用全路径方法计算两点之间的所有路径时,此类方法虽然从全局角度考虑,但是对于大规模图谱,两点间可存在百万条以上路径,执行一次具有较高的时间和空间成本,通常只能作为查询使用,无法一次性计算图中任意两点的关系权重。
步骤S200,确定所述多个节点中的待分析节点,将所述待分析节点与第一层邻居节点的关系权重值置为a,并随机计算所述待分析节点的任一邻居节点到所述待分析节点的第L层节点的关系权重值I,其中I=a-(L-1)*(a/6)。
其中, a为预设值,例如1或100,由于本申请实施例得方案是为了得到节点之间的关系权重,本质上是一个对比值,所以便于计算的任何预设值a都可以实现。
L大于等于1,当确定待分析节点之后,知识图谱为树形结构,待分析节点为知识图谱的根节点,待分析节点的第一层邻居节点即知识图谱层级的第1层,第L层节点即为自根节点向叶子节点方向扩展的对应层。
所谓的树形结构就是各个元素之间具有分层关系的数据结构,可以用一棵倒置的树来表示逻辑关系。所谓的根节点就是树的最顶端的节点,继续往下分为子节点,当不断细分直到不再有子节点时为叶子节点。
具体的,待分析节点的第二层邻居节点即知识图谱层级的第2层,第2层中的节点为根节点的第一层邻居节点的子节点,即第一层邻居节点的相邻层节点,确定为第二层邻居节点。
例如,如图3所示,节点1为根节点,节点2,3,4,5为节点1的第一层邻居节点,节点6,7,8,10,11,12,13为节点1的第二层邻居节点。
当L=1时,知识图谱的待分析节点只有一层邻居节点,待分析节点到其他节点的关系权重被赋予相同的权重值,即知识图谱的待分析节点和其他节点的权重都是相等的,则直接生成待分析节点和其他节点相同优先级的推荐数据。
当L>1时,则在待分析节点的邻居节点中随机跳转进行概率计算,具体描述如下:
在本实施例中,通过对知识图谱节点多次随机跳转进行概率计算,获取该节点到其它所有的关系强度值,然后选择满足预设关系强度值的节点生成推荐路径,解决了不能高效率低成本得到精确推荐结果的问题。
具体的,请参考图3和图4,步骤S200,确定所述多个节点中的待分析节点,将所述待分析节点与第一层邻居节点的关系权重值置为a,并随机计算所述待分析节点的任一邻居节点到所述待分析节点的第L层节点的关系权重值I,其中I=a-(L-1)*(a/6)包括:
步骤S201,初始化所述待分析节点到其他节点的关系权重值;
例如,在本实施例中,节点1为确定的待分析节点,初始化节点1到其他节点的的关系权重l = {0,0,0,0,0,0,0,0,0,0,0,0,0},第i个位置表示1到第i个节点的关系权重。
步骤S202,关联所述待分析节点和所述第一层邻居节点;
所述待分析节点的相邻节点为第一层邻居节点,在本实施例中,节点2,3,4,5为节点1的第一层邻居节点。
在其他实施例中,若确定节点2为待分析节点,则节点1,6,7为节点2的第一层邻居节点,若确定节点4为待分析节点,则节点1,8,9,14为节点4的第一层邻居节点,若确定14为待分析节点,则只有节点9为节点14的第一层邻居节点,依次类推,当确定了如图3所示的任一节点为待分析节点,通过获取待分析节点的相邻节点为第一层邻居节点,变可得到对应的第一层邻居节点的节点位置,并进行关联。
步骤S203,对所述待分析节点到所述第一层邻居节点的关系权重赋值为a;
在本实施例中,因为节点1为待分析节点,节点1的邻居节点关联到了节点2,3,4,5,则在节点2,3,4,5的位置为a。
在本实施例中,a预设为1,此时关系权重值I变为{1,1,1,1,0,0,0,0,0,0,0,0,0}。在本申请的其他实施例中,a也可预设为其他值,例如100。
步骤S204,随机选择所述第一层邻居节点中的任一节点,定义为第一节点,定义所述第一节点的相邻节点为第二层邻居节点,关联所述第一节点和所述第二层邻居节点;
在本实施例中,节点2,3,4,5为第一层邻居节点,随机在节点2,3,4,5中选择一个节点作为第一节点,例如随机选择后,节点4被选中为第一节点。
在本实施例中,通过调用一个随机选择器来实现第一节点的随机选择,随机选择器的原理是通过随机选择算法,在大量或者已知数据中随机选择一个数据,例如,通过在算法变量中添加一个扰动,并使该扰动服从特定的概率分布(伪随机),通过输入节点2,3,4,5的标识,然后返回得到对应的随机结果。
在本实施例中,节点4被随机选中为第一节点,则关联节点4的邻居节点(已关联除外),为节点8。
步骤S205,对所述待分析节点到所述第二层邻居节点的关系权重赋值为a-(a/6)。
由于此时已关联至第2层,关系强度随路径增加呈衰减关系。由六度关系理论可知,当关系路径到第六层时,关系强度可近似为0,因此每层递减系数为1/6≈0.167,关系权重I更新为:
I={1,1,1,1,0,0,1-0.167,0,0,0,0,0,0},即{1,1,1,1,0,0,0.833,0,0,0,0,0,0}。
在本实施例中,如果所述第二层邻居节点中的任一节点还存在邻居节点,则继续执行以下步骤,如果所述第二层邻居节点中的任一节点不存在邻居节点,则不执行以下步骤,跳转到步骤S209。
例如,在本实施例中,节点8还存在邻居节点9,则需要执行以下步骤,如果在步骤204中,随机选择了节点2为第一节点,关联到节点2的邻居节点,节点6和节点7,节点6和节点7没有邻居节点,则跳转到步骤S209。
对应的,如果随机选择了节点2为第一节点,关联到节点2的邻居节点6和节点7,通过计算得到了节点6和节点7的关系权重值,对应I为:
{1,1,1,1,0,1-0.167,1-0.167,0,0,0,0,0,0},即{1,1,1,1,0,0.833,0.833,0,0,0,0,0,0}
步骤S206,当所述第二层邻居节点中的任一节点存在邻居节点,则随机选择所述第二层邻居节点中的任一节点,定义为第二节点,定义所述第二节点的相邻节点为第三层邻居节点,关联所述第二节点和所述第三层邻居节点;
例如,在本实施例中,由于节点8只存在一个邻居节点9,所以节点9被选中为作为第二节点,节点9的邻居节点为节点14,节点14为第三层邻居节点。在其他实施例中,如果第二节点为其他节点,而且之后不存在邻居节点,则跳转到步骤S209。
步骤S207,对所述待分析节点到所述第三层邻居节点的关系权重赋值为a-2*(a/6)。
在本实施例中,在节点8的基础上随机采样邻居,为节点9,此时l变为{1, 1, 1, 1, 0, 0, 0.833,1-0.167-0.167,0,0,0,0,0},即{1, 1, 1, 1, 0, 0, 0.833, 0.666,0,0,0,0,0};
步骤S208,当所述第三层邻居节点还存在邻居节点,则根据层级顺序继续由小到大依次对未赋值节点赋值为 a-(L-1)*(a/6),其中i为当前赋值节点与待分析节点的层级数;
由于节点9还存在邻居节点,则继续通过随机选择邻居节点,然后根据I=a-(L-1)*(a/6)对邻居节点的关系权重赋值。
例如,在本实施例中,节点9存在邻居节点14,在节点9的基础上随机采样邻居,为节点14,此时l变为{1, 1, 1, 1,0,0,0.833,0.666,0,0,0,0,1-0.167-0.167-0.167},即4度关联后的关系权重值更新为I={1,1,1,1,0,0,0.833,0.666,0,0,0,0,0.499};
由于节点14此时无法继续关联,实验结束,本次计算匹配到{2,3,4,5,8,9,14}7个点,权重分别对应为{1,1,1,1,0.833,0.666,0.499},其余点关系权重为0,本次赋值结束。
在其他实施例中,当所述第L层邻居节点还存在邻居节点,且L大于等于6,则对第6层及之后的节点赋值为0,因为根据6度空间理论,在第6层节点是,节点的关系权重已经近似为0。
步骤S209,若所述第一节点的不同层级邻居节点已完成关系权重赋值或所述第一节点没有邻居节点,则随机选择所述第一层邻居节点中其他节点,定义为新的第一节点,计算所述新的第一节点与所述新的第一节点不同层级邻居节点的关系权重值。
例如,通过随机选择所述第一层邻居节点中其他节点,定义为新的第一节点,
若所述待分析节点和其他节点的存在多个关系权重值,则将对应的关系权重值取平均值作为所述待分析节点和其他节点的最终关系权重值。
具体的,按照上述步骤,原理同上,随机选择待分析节点的邻居节点中除了第一邻居节点的其他节点,重复步骤S201-S208;可得到另外一种权重组合。
例如本次计算在1的邻居节点处采样到节点2,则会最终命中节点2,3,4,5,6,7,对应的,如果随机选择了节点2为第一节点,关联到节点2的邻居节点6和节点7,通过计算得到了节点6和节点7的关系权重值,对应I为:{1,1,1,1,0,1-0.167,1-0.167,0,0,0,0,0,0},即{1,1,1,1,0,0.833,0.833,0,0,0,0,0,0}节点2,3,4,5,6,7权重对应为{1,1,1,1,0.833,0.833});
将两次实验结果叠加并取平均值,则得到节点1到其余节点的关系权重为{1, 1, 1, 1, 0.4165, 0.4165, 0.4165, 0.333, 0, 0, 0, 0, 0.2495}。
第一次随机计算后得到的关系权重值I1为:
{1,1,1,1,0,0,0.833,0.666,0,0,0,0,0.499};
第二次随机计算后得到的关系权重值I2为:
{1,1,1,1,0.833,0.833,0,0,0,0,0,0,0}
经过叠加计算取平均值后得到的最终关系权重I为:
{1, 1, 1, 1, 0.4165, 0.4165, 0.4165, 0.333, 0, 0, 0, 0, 0.2495};
其中,具体的,节点2,3,4,5由于为第一层邻居节点,赋值为固定值a,所以节点2,3,4,5无论随机计算几次,它们和节点1的关系权重值都是固定的。
节点6和节点7在第一次随机计算后关系权重值为0,第二次随机计算后关系权重值为0.833,取平均值后为0.4165,节点8和节点14在第一次随机计算后关系权重值分别为0.666和0.499,第二次随机计算后关系权重值为0,取平均值后对应为0.333和 0.2495。
在本实施例中,设置为只随机计算2次,同理,在其他实施例中,随机计算的次数可以设置为大于2次,通过同样的方法可计算其他节点到剩余节点的关系权重集合。
另外,如果发现两次采样后的结果仍然有部分节点没有命中。通过增加计算次数,可以增大命中其余节点的概率,从而使得结果更加逼近真实情况。
例如在本实施例中,节点10,节点11,节点12和节点13没有命中,为了使得结果更加精确,可以增加计算次数来增大命中节点10,节点11,节点12和节点13的概率。
通过第三次随机计算命中了节点3为第一节点,通过第四次随机计算命中了节点5为第一节点,通过4次随机计算后最终命中所有节点,
第三随机计算后得到的关系权重值I3为:
{1,1,1,1,0,0,0,0,0.833,0.833,0,0,0};
第四随机计算后得到的关系权重值I4为:
{1,1,1,1,0,0,0,0,0,0,0.833,0.833,0}
对应叠加4次随机计算结果取平均值后的关系权重值为:
{1, 1, 1, 1, 0.20825, 0.20825, 0.4165, 0.1665, 0.20825, 0.20825, 0.20825, 0.20825, 0.12475}。
步骤S300,选择所述待分析节点到其他节点的关系权重值大于预设值的节点生成推荐路径;
步骤S400,根据所述推荐路径生成所述待分析节点与其他节点之间的推荐数据。
具体的,在本实施例中,只随机计算了2次,最终的节点关系权重值为{1, 1, 1, 1, 0.4165, 0.4165, 0.4165, 0.333, 0, 0, 0, 0, 0.2495};选择当节点之间的关系强度值大于0.4时,将所述推荐路径中的产品属性节点信息推荐给所述推荐路径中的用户属性节点,例如,将对应节点的用户产品购买记录推荐给关联的其他用户。对应的,节点2、节点3,节点4,节点5,节点7、节点7,的关系权重值大于0.4,则生成节点2、节点3,节点4,节点5,节点7、节点8和节点1之间的推荐数据。
具体的,在本实施例中,将所述推荐路径中的产品属性节点信息推荐给所述推荐路径中的用户属性节点,例如确定待分析节点为产品属性节点,即节点1为产品属性节点,即特定用户的产品购买记录信息,为了得到该产品对应的推荐用户规则,则可以推荐给关系权重值大于预设值的节点,例如节点2、节点3,节点4,节点5,节点7、节点8中的用户属性节点。
例如如果设置随机计算4次,最终的节点关系权重值为{1, 1, 1, 1, 0.20825, 0.20825, 0.4165, 0.1665, 0.20825, 0.20825, 0.20825, 0.20825, 0.12475},预设值设置为0.2,则节点1,2,3,4,5,6,7,9,10,11,12,13的关系权重值大于0.2,则生成节点1,2,3,4,5,6,7,9,10,11,12,13和节点1之间的推荐数据。
在本实施例中,步骤S400,根据所述推荐路径生成所述多个节点之间的推荐数据之后还包括:
将所述推荐路径生成所述多个节点之间的推荐数据上传至区块链中。
基于所述推荐路径生成所述多个节点之间的推荐数据得到对应的摘要信息,具体来说,摘要信息由所述推荐路径生成所述多个节点之间的推荐数据进行散列处理得到,比如利用sha256s算法处理得到。将摘要信息上传至区块链可保证其安全性和对用户的公正透明性。用户设备可以从区块链中下载得该摘要信息,以便查证所述推荐路径生成所述多个节点之间的推荐数据是否被篡改。本示例所指区块链是分布式数据存储、点对点传输、共识机制、加密算法等计算机技术的新型应用模式。区块链(Blockchain),本质上是一个去中心化的数据库,是一串使用密码学方法相关联产生的数据块,每一个数据块中包含了一批次网络交易的信息,用于验证其信息的有效性(防伪)和生成下一个区块。区块链可以包括区块链底层平台、平台产品服务层以及应用服务层等。
本申请的一种基于知识图谱的推荐方法,通过对知识图谱节点多次随机跳转进行概率计算,能够从单节点出发,通过多次计算,获取该节点到其它所有的关系强度值,然后选择满足预设关系强度值的节点生成推荐路径,解决了不能高效率低成本得到精确推荐结果的问题。
实施例二
请继续参阅图5,示出了本申请基于知识图谱的推荐装置的程序模块示意图。在本实施例中,基于知识图谱的推荐装置20可以包括或被分割成一个或多个程序模块,一个或者多个程序模块被存储于存储介质中,并由一个或多个处理器所执行,以完成本申请,并可实现上述基于知识图谱的推荐方法。本申请实施例所称的程序模块是指能够完成特定功能的一系列计算机程序指令段,比程序本身更适合于描述基于知识图谱的推荐装置20在存储介质中的执行过程。以下描述将具体介绍本实施例各程序模块的功能:
权重计算模块202,用于确定所述多个节点中的待分析节点,将所述待分析节点与第一层邻居节点的关系权重值置为a,并随机计算所述待分析节点的任一邻居节点到所述待分析节点的第L层节点的关系权重值I,其中I=a-(L-1)*(a/6);
路径搜索模块204,用于选择所述待分析节点到其他节点的关系权重值大于预设值的节点生成推荐路径;
推荐输出模块206,用于根据所述推荐路径生成所述待分析节点与其他节点之间的推荐数据。
实施例三
参阅图6,是本申请实施例三之计算机设备的硬件架构示意图。本实施例中,所述计算机设备2是一种能够按照事先设定或者存储的指令,自动进行数值计算和/或信息处理的设备。该计算机设备2可以是机架式服务器、刀片式服务器、塔式服务器或机柜式服务器(包括独立的服务器,或者多个服务器所组成的服务器集群)等。如图7所示,所述计算机设备2至少包括,但不限于,可通过系统总线相互通信连接存储器21、处理器22、网络接口23、以及基于知识图谱的推荐装置20。其中:
本实施例中,存储器21至少包括一种类型的计算机可读存储介质,所述可读存储介质包括闪存、硬盘、多媒体卡、卡型存储器(例如,SD或DX存储器等)、随机访问存储器(RAM)、静态随机访问存储器(SRAM)、只读存储器(ROM)、电可擦除可编程只读存储器(EEPROM)、可编程只读存储器(PROM)、磁性存储器、磁盘、光盘等。在一些实施例中,存储器21可以是计算机设备2的内部存储单元,例如该计算机设备2的硬盘或内存。在另一些实施例中,存储器21也可以是计算机设备2的外部存储设备,例如该计算机设备2上配备的插接式硬盘,智能存储卡(Smart Media Card, SMC),安全数字(Secure Digital, SD)卡,闪存卡(Flash Card)等。当然,存储器21还可以既包括计算机设备2的内部存储单元也包括其外部存储设备。本实施例中,存储器21通常用于存储安装于计算机设备2的操作系统和各类应用软件,例如实施例二的基于知识图谱的推荐装置20的程序代码等。此外,存储器21还可以用于暂时地存储已经输出或者将要输出的各类数据。
处理器22在一些实施例中可以是中央处理器(Central Processing Unit,CPU)、控制器、微控制器、微处理器、或其他数据处理芯片。该处理器22通常用于控制计算机设备2的总体操作。本实施例中,处理器22用于运行存储器21中存储的程序代码或者处理数据,例如运行基于知识图谱的推荐装置20,以实现实施例一的基于知识图谱的推荐方法。
所述网络接口23可包括无线网络接口或有线网络接口,该网络接口23通常用于在所述服务器2与其他电子装置之间建立通信连接。例如,所述网络接口23用于通过网络将所述服务器2与外部终端相连,在所述服务器2与外部终端之间的建立数据传输通道和通信连接等。所述网络可以是企业内部网(Intranet)、互联网(Internet)、全球移动通讯系统(Global System of Mobile communication,GSM)、宽带码分多址(Wideband Code Division Multiple Access,WCDMA)、4G网络、5G网络、蓝牙(Bluetooth)、Wi-Fi等无线或有线网络。
需要指出的是,图6仅示出了具有部件20-23的计算机设备2,但是应理解的是,并不要求实施所有示出的部件,可以替代的实施更多或者更少的部件。
在本实施例中,存储于存储器21中的所述基于知识图谱的推荐装置20还可以被分割为一个或者多个程序模块,所述一个或者多个程序模块被存储于存储器21中,并由一个或多个处理器(本实施例为处理器22)所执行,以完成本申请。
实施例四
本实施例还提供一种计算机可读存储介质,所述计算机可读存储介质可以是非易失性,也可以是易失性,如闪存、硬盘、多媒体卡、卡型存储器(例如,SD或DX存储器等)、随机访问存储器(RAM)、静态随机访问存储器(SRAM)、只读存储器(ROM)、电可擦除可编程只读存储器(EEPROM)、可编程只读存储器(PROM)、磁性存储器、磁盘、光盘、服务器、App应用商城等等,其上存储有计算机可读指令,被处理器执行时实现相应功能。本实施例的计算机可读存储介质用于存储计算机可读指令基于知识图谱的推荐装置,被处理器执行时实现实施例一的基于知识图谱的推荐方法。
本申请所指区块链是分布式数据存储、点对点传输、共识机制、加密算法等计算机技术的新型应用模式。区块链(Blockchain),本质上是一个去中心化的数据库,是一串使用密码学方法相关联产生的数据块,每一个数据块中包含了一批次网络交易的信息,用于验证其信息的有效性(防伪)和生成下一个区块。区块链可以包括区块链底层平台、平台产品服务层以及应用服务层等。
上述本申请实施例序号仅仅为了描述,不代表实施例的优劣。
通过以上的实施方式的描述,本领域的技术人员可以清楚地了解到上述实施例方法可借助软件加必需的通用硬件平台的方式来实现,当然也可以通过硬件,但很多情况下前者是更佳的实施方式。
以上仅为本申请的优选实施例,并非因此限制本申请的专利范围,凡是利用本申请说明书及附图内容所作的等效结构或等效流程变换,或直接或间接运用在其他相关的技术领域,均同理包括在本申请的专利保护范围内。

Claims (20)

  1. 一种基于知识图谱的推荐方法,其中,所述知识图谱包括多个节点,所述方法包括:
    确定所述多个节点中的待分析节点,将所述待分析节点与第一层邻居节点的关系权重值置为a,并随机计算所述待分析节点的任一邻居节点到所述待分析节点的第L层节点的关系权重值I,其中I=a-(L-1)*(a/6);
    选择所述待分析节点到其他节点的关系权重值大于预设值的节点生成推荐路径;
    根据所述推荐路径生成所述待分析节点与其他节点之间的推荐数据。
  2. 根据权利要求1所述的基于知识图谱的推荐方法,其中,所述知识图谱包括用户属性节点和产品属性节点,所述方法还包括:
    根据用户属性表生成所述知识图谱的用户属性节点,根据产品属性表生成所述知识图谱的产品属性节点,所述用户属性表包括用户的产品购买历史记录,所述产品属性表包括对应产品的属性信息;对应的,所述根据所述推荐路径生成所述待分析节点与其他节点之间的推荐数据包括:
    将所述推荐路径中的产品属性节点信息推荐给所述推荐路径中的用户属性节点。
  3. 根据权利要求2所述的基于知识图谱的推荐方法,其中,所述确定所述多个节点中的待分析节点,将所述待分析节点与第一层邻居节点的关系权重值置为a,并随机计算所述待分析节点的任一邻居节点到所述待分析节点的第L层节点的关系权重值I包括:
    初始化所述待分析节点到其他节点的关系权重值;
    关联所述待分析节点和所述第一层邻居节点;
    对所述待分析节点到所述第一层邻居节点的关系权重赋值为a;
    随机选择所述第一层邻居节点中的任一节点,定义为第一节点,定义所述第一节点的相邻节点为第二层邻居节点,关联所述第一节点和所述第二层邻居节点;
    对所述待分析节点到所述第二层邻居节点的关系权重赋值为a-(a/6)。
  4. 根据权利要求3所述的基于知识图谱的推荐方法,其中,所述对所述待分析节点到所述第二层邻居节点的关系权重赋值为a-(a/6)之后还包括:
    当所述第二层邻居节点中的任一节点存在邻居节点,则随机选择所述第二层邻居节点中的任一节点,定义为第二节点,定义所述第二节点的相邻节点为第三层邻居节点,关联所述第二节点和所述第三层邻居节点;
    对所述待分析节点到所述第三层邻居节点的关系权重赋值为a-2*(a/6);
    当所述第L层邻居节点还存在邻居节点,且L大于等于6,则对第6层及之后的节点赋值为0。
  5. 根据权利要求4所述的基于知识图谱的推荐方法,其中,所述选择所述待分析节点到其他节点的关系权重值大于预设值的节点生成推荐路径之前还包括:
    若所述第一节点的不同层级邻居节点已完成关系权重赋值或所述第一节点没有邻居节点,则随机选择所述第一层邻居节点中除了所述第一节点的其他节点,定义为新的第一节点,计算所述新的第一节点与所述新的第一节点不同层级邻居节点的关系权重值。
  6. 根据权利要求5所述的基于知识图谱的推荐方法,其中,所述选择关系权重值大于预设值的节点生成推荐路径之前还包括:
    若所述待分析节点和其他节点的存在多个关系权重值,则将对应的关系权重值取平均值作为所述待分析节点和其他节点的最终关系权重值。
  7. 根据权利要求1所述的基于知识图谱的推荐方法,其中,所述根据所述推荐路径生成所述多个节点之间的推荐数据之后,还包括:
    将根据所述推荐路径生成所述多个节点之间的推荐数据上传至区块链中。
  8. 一种基于知识图谱的推荐装置,其中,所述知识图谱包括多个节点,所述基于知识图谱的推荐装置包括:
    权重计算模块,用于确定所述多个节点中的待分析节点,将所述待分析节点与第一层邻居节点的关系权重值置为a,并随机计算所述待分析节点的任一邻居节点到所述待分析节点的第L层节点的关系权重值I,其中I=a-(L-1)*(a/6);
    路径搜索模块,用于选择所述待分析节点到其他节点的关系权重值大于预设值的节点生成推荐路径;
    推荐输出模块,用于根据所述推荐路径生成所述待分析节点与其他节点之间的推荐数据。
  9. 一种计算机设备,其中,所述计算机设备包括存储器、处理器,所述存储器上存储有可在所述处理器上运行的基于知识图谱的推荐计算机可读指令,所述处理器执行所述计算机可读指令时还执行以下步骤:
    从知识图谱的多个节点中确定待分析节点,将所述待分析节点与第一层邻居节点的关系权重值置为a,并随机计算所述待分析节点的任一邻居节点到所述待分析节点的第L层节点的关系权重值I,其中I=a-(L-1)*(a/6);
    选择所述待分析节点到其他节点的关系权重值大于预设值的节点生成推荐路径;
    根据所述推荐路径生成所述待分析节点与其他节点之间的推荐数据。
  10. 根据权利要求9所述的计算机设备,其中,所述知识图谱包括用户属性节点和产品属性节点,所述处理器执行所述计算机可读指令时还执行以下步骤:
    根据用户属性表生成所述知识图谱的用户属性节点,根据产品属性表生成所述知识图谱的产品属性节点,所述用户属性表包括用户的产品购买历史记录,所述产品属性表包括对应产品的属性信息;对应的,所述根据所述推荐路径生成所述待分析节点与其他节点之间的推荐数据包括:
    将所述推荐路径中的产品属性节点信息推荐给所述推荐路径中的用户属性节点。
  11. 根据权利要求10所述的计算机设备,其中,所述处理器执行所述计算机可读指令时还执行以下步骤:
    初始化所述待分析节点到其他节点的关系权重值;
    关联所述待分析节点和所述第一层邻居节点;
    对所述待分析节点到所述第一层邻居节点的关系权重赋值为a;
    随机选择所述第一层邻居节点中的任一节点,定义为第一节点,定义所述第一节点的相邻节点为第二层邻居节点,关联所述第一节点和所述第二层邻居节点;
    对所述待分析节点到所述第二层邻居节点的关系权重赋值为a-(a/6)。
  12. 根据权利要求11所述的计算机设备,其中,所述处理器执行所述计算机可读指令时还执行以下步骤:
    当所述第二层邻居节点中的任一节点存在邻居节点,则随机选择所述第二层邻居节点中的任一节点,定义为第二节点,定义所述第二节点的相邻节点为第三层邻居节点,关联所述第二节点和所述第三层邻居节点;
    对所述待分析节点到所述第三层邻居节点的关系权重赋值为a-2*(a/6);
    当所述第L层邻居节点还存在邻居节点,且L大于等于6,则对第6层及之后的节点赋值为0。
  13. 根据权利要求12所述的计算机设备,其中,所述处理器执行所述计算机可读指令时还执行以下步骤:
    若所述第一节点的不同层级邻居节点已完成关系权重赋值或所述第一节点没有邻居节点,则随机选择所述第一层邻居节点中除了所述第一节点的其他节点,定义为新的第一节点,计算所述新的第一节点与所述新的第一节点不同层级邻居节点的关系权重值。
  14. 根据权利要求9所述的计算机设备,其中,所述处理器执行所述计算机可读指令时还执行以下步骤:
    若所述待分析节点和其他节点的存在多个关系权重值,则将对应的关系权重值取平均值作为所述待分析节点和其他节点的最终关系权重值。
  15. 一种计算机可读存储介质,其中,所述计算机可读存储介质内存储有计算机可读指令,所述计算机可读指令可被至少一个处理器所执行,以使所述至少一个处理器执行以下步骤:
    从知识图谱的多个节点中确定待分析节点,将所述待分析节点与第一层邻居节点的关系权重值置为a,并随机计算所述待分析节点的任一邻居节点到所述待分析节点的第L层节点的关系权重值I,其中I=a-(L-1)*(a/6);
    选择所述待分析节点到其他节点的关系权重值大于预设值的节点生成推荐路径;
    根据所述推荐路径生成所述待分析节点与其他节点之间的推荐数据。
  16. 根据权利要求15所述的计算机可读存储介质,其中,所述知识图谱包括用户属性节点和产品属性节点,所述计算机可读指令可被至少一个处理器所执行,以使所述至少一个处理器还执行以下步骤:
    根据用户属性表生成所述知识图谱的用户属性节点,根据产品属性表生成所述知识图谱的产品属性节点,所述用户属性表包括用户的产品购买历史记录,所述产品属性表包括对应产品的属性信息;对应的,所述根据所述推荐路径生成所述待分析节点与其他节点之间的推荐数据包括:
    将所述推荐路径中的产品属性节点信息推荐给所述推荐路径中的用户属性节点。
  17. 根据权利要求16所述的计算机可读存储介质,其中,所述计算机可读指令可被至少一个处理器所执行,以使所述至少一个处理器还执行以下步骤:
    初始化所述待分析节点到其他节点的关系权重值;
    关联所述待分析节点和所述第一层邻居节点;
    对所述待分析节点到所述第一层邻居节点的关系权重赋值为a;
    随机选择所述第一层邻居节点中的任一节点,定义为第一节点,定义所述第一节点的相邻节点为第二层邻居节点,关联所述第一节点和所述第二层邻居节点;
    对所述待分析节点到所述第二层邻居节点的关系权重赋值为a-(a/6)。
  18. 根据权利要求17所述的计算机可读存储介质,其中,所述计算机可读指令可被至少一个处理器所执行,以使所述至少一个处理器还执行以下步骤:
    当所述第二层邻居节点中的任一节点存在邻居节点,则随机选择所述第二层邻居节点中的任一节点,定义为第二节点,定义所述第二节点的相邻节点为第三层邻居节点,关联所述第二节点和所述第三层邻居节点;
    对所述待分析节点到所述第三层邻居节点的关系权重赋值为a-2*(a/6);
    当所述第L层邻居节点还存在邻居节点,且L大于等于6,则对第6层及之后的节点赋值为0。
  19. 根据权利要求18所述的计算机可读存储介质,其中,所述计算机可读指令可被至少一个处理器所执行,以使所述至少一个处理器还执行以下步骤:
    若所述第一节点的不同层级邻居节点已完成关系权重赋值或所述第一节点没有邻居节点,则随机选择所述第一层邻居节点中除了所述第一节点的其他节点,定义为新的第一节点,计算所述新的第一节点与所述新的第一节点不同层级邻居节点的关系权重值。
  20. 根据权利要求15所述的计算机可读存储介质,其中,所述计算机可读指令可被至少一个处理器所执行,以使所述至少一个处理器还执行以下步骤:
    若所述待分析节点和其他节点的存在多个关系权重值,则将对应的关系权重值取平均值作为所述待分析节点和其他节点的最终关系权重值。
PCT/CN2021/109760 2020-09-02 2021-07-30 基于知识图谱的推荐方法、装置、计算机设备及存储介质 WO2022048368A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202010907984.7A CN111966912B (zh) 2020-09-02 2020-09-02 基于知识图谱的推荐方法、装置、计算机设备及存储介质
CN202010907984.7 2020-09-02

Publications (1)

Publication Number Publication Date
WO2022048368A1 true WO2022048368A1 (zh) 2022-03-10

Family

ID=73392258

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2021/109760 WO2022048368A1 (zh) 2020-09-02 2021-07-30 基于知识图谱的推荐方法、装置、计算机设备及存储介质

Country Status (2)

Country Link
CN (1) CN111966912B (zh)
WO (1) WO2022048368A1 (zh)

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114697188A (zh) * 2022-03-28 2022-07-01 天津大学 一种基于网络态势感知的物联网设备风险预测分析方法
CN116089624A (zh) * 2022-11-17 2023-05-09 昆仑数智科技有限责任公司 基于知识图谱的数据推荐方法、装置和系统
CN116319518A (zh) * 2022-08-31 2023-06-23 王举范 一种基于知识图谱最短路径的信息获取方法及装置
CN117033775A (zh) * 2023-07-28 2023-11-10 广东工业大学 基于知识图谱的工业软件的组件推荐方法及系统

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN111966912B (zh) * 2020-09-02 2023-03-10 深圳壹账通智能科技有限公司 基于知识图谱的推荐方法、装置、计算机设备及存储介质
CN115685765A (zh) * 2021-07-21 2023-02-03 青岛海信智慧生活科技股份有限公司 一种设备联动方案推荐方法、设备及介质

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20160189028A1 (en) * 2014-12-31 2016-06-30 Verizon Patent And Licensing Inc. Systems and Methods of Using a Knowledge Graph to Provide a Media Content Recommendation
CN108804633A (zh) * 2018-06-01 2018-11-13 腾讯科技(深圳)有限公司 基于行为语义知识网络的内容推荐方法
CN109947817A (zh) * 2019-03-01 2019-06-28 武汉达梦数据库有限公司 一种基于关联图谱的六度空间关系分析方法及系统
CN111274495A (zh) * 2020-01-20 2020-06-12 平安科技(深圳)有限公司 用户关系强度的数据处理方法、装置、计算机设备及存储介质
CN111966912A (zh) * 2020-09-02 2020-11-20 深圳壹账通智能科技有限公司 基于知识图谱的推荐方法、装置、计算机设备及存储介质

Family Cites Families (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20140136508A1 (en) * 2012-11-09 2014-05-15 Palo Alto Research Center Incorporated Computer-Implemented System And Method For Providing Website Navigation Recommendations
CN109871445A (zh) * 2019-01-23 2019-06-11 平安科技(深圳)有限公司 欺诈用户识别方法、装置、计算机设备和存储介质
CN110210892B (zh) * 2019-05-05 2023-05-30 平安科技(深圳)有限公司 产品推荐方法、装置和可读存储介质
CN111429255B (zh) * 2020-03-19 2024-02-20 中国建设银行股份有限公司 一种风险评估方法、装置、设备及存储介质

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US20160189028A1 (en) * 2014-12-31 2016-06-30 Verizon Patent And Licensing Inc. Systems and Methods of Using a Knowledge Graph to Provide a Media Content Recommendation
CN108804633A (zh) * 2018-06-01 2018-11-13 腾讯科技(深圳)有限公司 基于行为语义知识网络的内容推荐方法
CN109947817A (zh) * 2019-03-01 2019-06-28 武汉达梦数据库有限公司 一种基于关联图谱的六度空间关系分析方法及系统
CN111274495A (zh) * 2020-01-20 2020-06-12 平安科技(深圳)有限公司 用户关系强度的数据处理方法、装置、计算机设备及存储介质
CN111966912A (zh) * 2020-09-02 2020-11-20 深圳壹账通智能科技有限公司 基于知识图谱的推荐方法、装置、计算机设备及存储介质

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114697188A (zh) * 2022-03-28 2022-07-01 天津大学 一种基于网络态势感知的物联网设备风险预测分析方法
CN116319518A (zh) * 2022-08-31 2023-06-23 王举范 一种基于知识图谱最短路径的信息获取方法及装置
CN116319518B (zh) * 2022-08-31 2024-02-20 王举范 一种基于知识图谱最短路径的信息获取方法及装置
CN116089624A (zh) * 2022-11-17 2023-05-09 昆仑数智科技有限责任公司 基于知识图谱的数据推荐方法、装置和系统
CN116089624B (zh) * 2022-11-17 2024-02-27 昆仑数智科技有限责任公司 基于知识图谱的数据推荐方法、装置和系统
CN117033775A (zh) * 2023-07-28 2023-11-10 广东工业大学 基于知识图谱的工业软件的组件推荐方法及系统
CN117033775B (zh) * 2023-07-28 2024-03-19 广东工业大学 基于知识图谱的工业软件的组件推荐方法及系统

Also Published As

Publication number Publication date
CN111966912A (zh) 2020-11-20
CN111966912B (zh) 2023-03-10

Similar Documents

Publication Publication Date Title
WO2022048368A1 (zh) 基于知识图谱的推荐方法、装置、计算机设备及存储介质
Wu et al. An L 1-and-L 2-norm-oriented latent factor model for recommender systems
US11030247B2 (en) Layered graph data structure
US9866565B2 (en) Method and apparatus for controlling data permissions
US10366335B2 (en) Systems and methods for symbolic analysis
Zhang et al. A weighted kernel possibilistic c‐means algorithm based on cloud computing for clustering big data
US11042662B2 (en) Data aggregation system for enabling query operations on restricted data that originates from multiple independent multiple sources
US8904377B2 (en) Reconfiguration of computer system to allow application installation
US8706716B2 (en) Iterative and dynamic search of publicly available data based on augmentation of search terms and validation of data relevance
Ouadah et al. SEFAP: an efficient approach for ranking skyline web services
WO2018231531A1 (en) Execution optimization of database statements involving encrypted data
KR102410777B1 (ko) 상품 추천 방식 개선 방법 및 장치
CN108230005B (zh) 在线云端服务处理系统与评测方法及其计算机程序产品
CN113452752B (zh) 一种多域物联网场景下的信任管理、博弈方法、系统
Fattah et al. Long-term IaaS selection using performance discovery
CN109657153A (zh) 一种用于确定用户的关联财经信息的方法与设备
US10693855B1 (en) Fraud detection
Zhang et al. Small files storing and computing optimization in Hadoop parallel rendering
Singh et al. Cloud based evaluation of databases for stock market data
CN110321476B (zh) 基于局部敏感哈希的并行矩阵分解计算实现方法
CN108984615B (zh) 一种数据查询方法和系统、存储介质
US11914593B2 (en) Generate digital signature of a query execution plan using similarity hashing
Mistry et al. Probabilistic qualitative preference matching in long-term IaaS composition
CN114253951B (zh) 数据处理方法、系统及第二服务器
US11663506B2 (en) Estimation of power consumption of proof of work consensus based distributed systems

Legal Events

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

Ref document number: 21863436

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

32PN Ep: public notification in the ep bulletin as address of the adressee cannot be established

Free format text: NOTING OF LOSS OF RIGHTS PURSUANT TO RULE 112(1) EPC (EPO FORM 1205 DATED 30/06/2023)

122 Ep: pct application non-entry in european phase

Ref document number: 21863436

Country of ref document: EP

Kind code of ref document: A1