WO2020232879A1 - 风险传导关联图谱优化方法、装置、计算机设备和存储介质 - Google Patents

风险传导关联图谱优化方法、装置、计算机设备和存储介质 Download PDF

Info

Publication number
WO2020232879A1
WO2020232879A1 PCT/CN2019/102962 CN2019102962W WO2020232879A1 WO 2020232879 A1 WO2020232879 A1 WO 2020232879A1 CN 2019102962 W CN2019102962 W CN 2019102962W WO 2020232879 A1 WO2020232879 A1 WO 2020232879A1
Authority
WO
WIPO (PCT)
Prior art keywords
risk
correlation
data
user
node
Prior art date
Application number
PCT/CN2019/102962
Other languages
English (en)
French (fr)
Inventor
王丽杰
徐志成
Original Assignee
平安科技(深圳)有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 平安科技(深圳)有限公司 filed Critical 平安科技(深圳)有限公司
Publication of WO2020232879A1 publication Critical patent/WO2020232879A1/zh

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/24Querying
    • G06F16/245Query processing
    • G06F16/2458Special types of queries, e.g. statistical queries, fuzzy queries or distributed queries
    • G06F16/2465Query processing support for facilitating data mining operations in structured databases
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/20Information retrieval; Database structures therefor; File system structures therefor of structured data, e.g. relational data
    • G06F16/28Databases characterised by their database models, e.g. relational or object models
    • G06F16/284Relational databases
    • G06F16/288Entity relationship models
    • 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
    • YGENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02PCLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
    • Y02P90/00Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
    • Y02P90/30Computing systems specially adapted for manufacturing

Definitions

  • This application relates to a method, device, computer equipment, and storage medium for optimizing risk conduction association maps.
  • Knowledge graph also known as scientific knowledge graph, is a series of various graphs that show the development process and structural relationship of knowledge. It uses visualization technology to describe knowledge resources and their carriers, and mines, analyzes, constructs, draws and displays knowledge and their relationships. Of mutual connections. In a general sense, the main goal of the knowledge graph is to describe the various entities and concepts that exist in the real world and the relationships between them.
  • Traditional knowledge graphs can be constructed through entities and their association relationships, and can be used to conduct risk transmission between entities. In some application scenarios, it can be constructed based on certain factor information to establish association relationships between nodes to obtain risk transmission. Correlation graph, characterize the relationship network between nodes, conduct risk conduction analysis.
  • the inventor realizes that due to the diversity of node data dimensions in the knowledge graph constructed in the traditional way, to a certain extent, the risk conduction association graph is complicated.
  • the traditional risk conduction correlation map has huge basic node data, which affects the calculation complexity of relationship conduction to a certain extent.
  • determining the relationship conduction calculation between nodes requires sorting out the correlation between nodes and performing real-time The calculation results have led to the problem of low data analysis efficiency when conducting risk conduction analysis on this risk conduction correlation map. This risk conduction correlation map urgently needs to be optimized.
  • a method, device, computer equipment, and storage medium for optimizing a risk conduction correlation map are provided.
  • a method for optimizing the risk transmission correlation map includes:
  • association factor determines association factor between the directly associated risk nodes, the association factor combination between the indirectly associated risk nodes, and the association level of each association factor in the association factor combination, and according to the direct association
  • the device tag risk weight parameter, the reference weight parameter of the correlation factor, and the risk conduction weight parameter of the correlation factor at different correlation levels update the parameters of the initial risk conduction correlation map to obtain an optimized risk conduction correlation Atlas.
  • a risk transmission correlation map optimization device includes:
  • the risk node acquisition module is used to acquire the risk node in the initial risk conduction association graph, the user risk data of the risk node, and the device label information of the risk user corresponding to the risk node;
  • a device tag risk weight parameter determination module configured to determine the device tag risk weight parameter of the risk node according to the user risk data and the device tag information
  • the benchmark weight parameter determination module is used to determine the correlation factor between the directly related risk nodes, the correlation factor combination between the indirectly related risk nodes, and the correlation of the correlation factors in the correlation factor combination according to the correlation between the risk nodes Level, and determine the reference weight parameter of the correlation factor according to the user risk data of the directly correlated risk node;
  • the risk transmission weight parameter determination module is used to determine the risk transmission weight parameter of the correlation factor at each correlation level according to the reference weight parameter of each correlation factor in the correlation factor combination and the correlation level of the correlation factor;
  • the initial risk conduction correlation map optimization module is used to update the initial risk conduction correlation according to the device tag risk weight parameter, the reference weight parameter of the correlation factor, and the risk conduction weight parameter of the correlation factor at different correlation levels
  • the parameters of the map are used to obtain an optimized risk transmission correlation map.
  • a computer device including a memory and one or more processors, the memory stores computer readable instructions, when the computer readable instructions are executed by the processor, the one or more processors execute The following steps:
  • association factor determines association factor between the directly associated risk nodes, the association factor combination between the indirectly associated risk nodes, and the association level of each association factor in the association factor combination, and according to the direct association
  • the device tag risk weight parameter, the reference weight parameter of the correlation factor, and the risk conduction weight parameter of the correlation factor at different correlation levels update the parameters of the initial risk conduction correlation map to obtain an optimized risk conduction correlation Atlas.
  • One or more non-volatile computer-readable storage media storing computer-readable instructions.
  • the one or more processors execute the following steps:
  • association factor determines association factor between the directly associated risk nodes, the association factor combination between the indirectly associated risk nodes, and the association level of each association factor in the association factor combination, and according to the direct association
  • the device tag risk weight parameter, the reference weight parameter of the correlation factor, and the risk conduction weight parameter of the correlation factor at different correlation levels update the parameters of the initial risk conduction correlation map to obtain an optimized risk conduction correlation Atlas.
  • Fig. 1 is an application scenario diagram of a method for optimizing a risk conduction association map according to one or more embodiments.
  • Fig. 2 is a schematic flowchart of a method for optimizing a risk conduction association map according to one or more embodiments.
  • Fig. 3 is a schematic diagram of a process of constructing an initial risk conduction association map according to one or more embodiments.
  • Fig. 4 is a schematic flow chart of a method for optimizing a risk conduction correlation map in another embodiment.
  • Fig. 5 is a schematic flowchart of a method for optimizing risk conduction association maps in another embodiment.
  • Fig. 6 is a block diagram of an apparatus for optimizing a risk conduction association map according to one or more embodiments.
  • Figure 7 is a block diagram of a computer device according to one or more embodiments.
  • the method for optimizing the risk conduction correlation map provided by this application can be applied to the application environment as shown in FIG. 1.
  • the terminal 102 and the server 104 communicate through the network.
  • the server 104 obtains the risk node in the initial risk conduction association graph, the user risk data of the risk node, and the device label information of the risk node corresponding to the risk user, and determines the device label risk weight parameter of the risk node according to the device label information and user risk data,
  • User risk data determine the reference weight parameter of the correlation factor According to the reference weight parameter of each correlation factor in the correlation factor combination, and the correlation level of the correlation factor, determine the risk transmission weight parameter of the correlation factor at each correlation level; according to the equipment label risk weight Parameters, the benchmark weight parameters of the correlation factors and the risk transmission weight parameters of the correlation factors at different correlation levels
  • a method for optimizing a risk conduction association graph is provided. Taking the method applied to the server in FIG. 1 as an example, the method includes the following steps:
  • Step S300 Obtain the risk node in the initial risk conduction association graph, the user risk data of the risk node, and the device label information of the risk user corresponding to the risk node.
  • the initial risk conduction association graph refers to the connection between users with the same data as the nodes, and the data dimension corresponding to the same data as the correlation factor between the nodes, which is used to characterize the connections between users. Knowledge graph. According to the nodes that carry user risk data in each correlation spectrum node, by labeling the risk nodes, the initial risk conduction correlation map with the risk nodes can be obtained.
  • User risk data refers to risk behaviors that have occurred. Risk behaviors include default behaviors, overdue behaviors, etc. The risk assessment is performed according to the specific default situation of the risk behavior, and the score value data used to characterize the user's risk situation, the score value data The size of the score is related to the number of risky behaviors and the degree of risk of risky behaviors.
  • User risk data can be obtained by analyzing the risk behavior records of each user in a pre-built risk database.
  • the risk behavior records include various risk behaviors of the user and the risk degree of the risk behavior. In an embodiment, the more the number of risky behaviors, or the more serious the plot corresponding to the risky behavior, the greater the value of user risk data corresponding to the user.
  • a user who does not have risk behavior does not include user risk data.
  • the user is a normal user, and the corresponding node in the initial risk conduction association graph is a non-risk node. In the initial risk conduction correlation map, there are risk nodes and non-risk nodes. According to whether there is user risk data, it can be determined whether the users corresponding to each node of the initial risk conduction correlation map are risk users. Carry out risk node labeling.
  • Risk behaviors generated by risky users generally have related records in the mobile device used by the user. By performing risk analysis on each data record in the user's mobile device according to the set data dimensions, the device label information of each dimension is
  • Step S400 Determine the risk weight parameter of the device label of the risk node according to the user risk data and the device label information.
  • a risk node is a node that contains user risk data, so the user corresponding to the risk node is a risk user, and the risk behavior of a risk user has related data records in the mobile device used by the user, for example, the user's browsing history, search history , Bad APP use records, loan APP use records, device flashing records, multiple devices with the same mobile phone number, different mobile phone numbers on the same device, and other different data dimensions.
  • the data record of each data dimension in the user’s mobile device corresponds to a device tag.
  • the risk identification of each device label can be determined as risky or non-risk, according to the numerous risk nodes in the initial risk conduction map and the user risk data of each risk node , A combination of multiple device tags with risk identifications can be obtained, thereby obtaining the weight of each device tag's influence on user risk data, and by accumulating the impact weights, the risk weight parameter of the device tag of the risk node is obtained.
  • Step S500 According to the association relationship between the risk nodes, determine the association factor between the directly associated risk nodes, the association factor combination between the indirectly associated risk nodes, and the association level of each association factor in the association factor combination, and according to the directly associated The user risk data of the risk node determines the benchmark weight parameter of the correlation factor.
  • the association relationship refers to the connection between two nodes.
  • the association relationship includes direct association and indirect association.
  • user A, user B, and user C are all risky users.
  • the correlation factor of address is linked together, the correlation between user A and user B is a direct correlation.
  • user B and user C have the same work unit, they are linked together based on the correlation factor of work unit, then user B
  • the association relationship with the user C is also a direct association, and therefore, the association relationship between the user A and the user C is an indirect association through the user B.
  • the association relationship group of multiple correlation factors and the user risk data is constructed.
  • the user risk data is known data, it can be calculated through the data, And the data fitting processing of the data calculation result, the reference weight parameter of each correlation factor is obtained, and the reference weight parameter is used to characterize the strength of the conductivity of each correlation factor between each node.
  • user A, user D, and user E are risk users
  • user B and user C are normal users
  • user A and User B is directly related by the correlation factor of home address
  • user B and user C are directly related by the correlation factor of work unit
  • user B and user D are directly related by the same mobile phone number
  • user C and user E are directly related by the same software account
  • user A is indirectly associated with user E
  • user D is indirectly associated with user E.
  • the combination of association factors between risk users is different. According to the number of association factors, you can Determine the correlation level of each correlation factor in the correlation factor combination.
  • the correlation level of the correlation factor is related to the direction of correlation conduction. For example, user A—home address—user B—mobile phone number—user D, and user D—mobile phone number—user B—home address—user A.
  • the correlation factor corresponding to the mobile phone number The level is different, and the correlation level of the correlation factor of home address is also different.
  • Step S600 Determine the risk transmission weight parameter of the correlation factor at each correlation level according to the reference weight parameter of each correlation factor in the correlation factor combination and the correlation level of the correlation factor.
  • the benchmark weight parameters of each correlation factor, and the correlation factors between risk nodes with indirect correlations, the correlation relationship group of multiple correlation factor combinations and user risk data is constructed. Because the user risk data is With the known data, the risk transmission weight parameters of each correlation factor at different correlation levels can be obtained through data calculation and data fitting processing of the data calculation results.
  • step S700 the parameters of the initial risk conduction correlation map are updated according to the device tag risk weight parameters, the reference weight parameters of the correlation factors, and the risk conduction weight parameters of the correlation factors at different correlation levels to obtain an optimized risk conduction correlation profile.
  • the risk transmission is related
  • the map is optimized.
  • the probability data of risk transmission is derived from the difference information of the label and factor weight ratio and the degree of correlation.
  • the fraud risk probability of other nodes with unknown risk conditions related to it can be obtained, which improves the risk probability result Based on the accuracy of the risk probability, it can be evaluated whether it is a fraud, and then rejected or focused on the investigation, saving manpower and time.
  • the above risk conduction correlation map optimization method based on the initial risk conduction correlation map, analyzes the device label information of the risk node corresponding to the risk user through the correlation factors between the directly related risk nodes, and obtains the device label risk weight parameters.
  • the correlation factor between directly related risk nodes determines the reference weight parameter of the correlation factor, and the risk conduction of each correlation factor under each correlation level is determined by the correlation level of each correlation factor in the correlation factor combination between the indirect correlation risk nodes
  • the weight parameters are used to optimize the initial risk transmission correlation map, fully excavate the deep connections between the nodes of the correlation map, and clarify the specific transmission parameters between risk nodes.
  • step S300 before obtaining the risk nodes in the initial risk conduction association map and the user risk data of the risk nodes, it also includes the construction of the initial risk conduction association map.
  • the construction process includes step S120 to step S180.
  • Step S120 Obtain user data information of each user used to construct the initial risk conduction association map, mark each user as an association map node, screen risk users carrying user risk data among users, and assign the association map corresponding to the risk user The nodes are marked as risk nodes.
  • Step S140 Obtain a preset data dimension, and extract data values corresponding to the data dimension from the user data information.
  • Step S160 Search for each feature association graph node containing the same data value, and mark the data dimension corresponding to the data value as the association factor between the feature association graph nodes.
  • Step S180 construct the initial risk conduction correlation map according to the correlation map node and the correlation factor, and the initial risk conduction correlation map is marked with risk nodes.
  • the initial risk conduction association map is used to display the association relationship between users.
  • the users containing user data information are marked as associated map nodes.
  • the user data information can be data information of various dimensions including risk data for users who have breached the contract, or It is the data information of each dimension of the non-default user in the normal state.
  • the data information extraction message is used to extract the relevant data value of the required data dimension from the user data information according to the data dimension it carries.
  • the preset data dimension is Refers to the data dimensions that are analyzed in advance based on actual needs and used to characterize the strength of connections between users. By setting data dimensions, the initial risk transmission model can be more organized and standardized.
  • the node can be determined as a feature correlation map node, the risk users who carry user risk data among the users are screened, and the correlation map node corresponding to the risk user is marked as a risk node.
  • the data value establishes the association relationship between the two feature association map nodes, and the data dimension corresponding to the data value is marked as the association factor between the feature association map nodes. Based on the association map node and the association factor, the initial node marked with risk is constructed. Risk transmission correlation map.
  • step S140 extracting multiple data dimension information carried in the message according to the acquired data information, and before extracting the data value corresponding to each data dimension in the user data information, further includes the step S132 and step S134.
  • Step S132 Obtain sample data of existing risk users, and label the risk data in the sample data with data dimensions.
  • Step S134 the co-occurrence frequency of the dimensional data in the sample data is counted, and the preset data dimension is the data dimension whose co-occurrence frequency meets a set threshold.
  • a data information extraction message carrying preset data dimension information may be generated first according to the data dimensions obtained by filtering.
  • the preset data dimension may be obtained by extracting the message from the extracted data information.
  • Co-occurrence refers to the phenomenon that the same field appears in different data at the same time.
  • the co-occurrence frequency of the same dimension data refers to the frequency of the specific field of the dimension label appearing in different data at the same time according to the dimension label of the same field.
  • the data information extraction message carrying the data dimension information is generated according to the filtered data dimension to determine the correlation factor between the nodes of the initial correlation map.
  • the set threshold of the co-occurrence frequency can be set according to actual needs.
  • step S400 acquiring the device label information of the risk node corresponding to the risk user, and determining the device label risk weight parameter of the risk node according to the device label information and user risk data includes step S420 to step S460.
  • Step S420 Obtain the device label information of the risk node corresponding to the risk user, and extract the risk identifier carried by each device label in the device label information.
  • the risk identifier includes risky and non-risky.
  • Step S440 According to the combination result corresponding to the risk identification carried by each risk label and the user risk data of the risk node, the weight of the influence of each risk label on the user risk data is obtained.
  • Step S460 Determine the device tag risk weight parameter of the risk node according to the risk label carrying the risk identifier in the risk node and the influence weight of each risk label.
  • Device tag information is a collection of multiple device tags.
  • Device tags specifically include device fingerprints, boot time, simulators, bad apps, loan apps, browser records, search records, proxy IP, call records, address books, and WeChat data , Device flashing, device blacklist, multiple devices with the same mobile phone number, different mobile phone numbers on the same device, and frequently used locations of the device, etc.
  • Each equipment label searches for the corresponding risk database according to its specific data records. When there is a matching data record, the risk identification of the equipment label is marked as risky, otherwise, it is marked as no risk. Combine the risk identifiers corresponding to each device tag in the tag set to obtain the risk identifier combination result.
  • the risk identifier combination result corresponding to each risk node and the user risk data of each risk node multiple risk identifier combinations and user risks can be constructed
  • the relational expression group of the data By solving the relational expression group, multiple sets of calculation results can be obtained.
  • the corresponding value of each risk label in the calculation result is used to perform data fitting processing.
  • the fitting result the risk of each risk label to the user is obtained
  • the influence weight of the data is calculated according to the risk label carrying the risk identifier in each risk node and the influence weight of each risk label, and the device label risk weight parameter of the risk node is obtained by cumulative calculation.
  • step S600 according to the reference weight parameter of each correlation factor in the correlation factor combination and the correlation level of the correlation factor, determining the risk conduction weight parameter of the correlation factor at each correlation level includes Step S620 to step S640.
  • Step S620 According to the correlation level of each correlation factor in the correlation factor combination, and the user risk data of the corresponding risk node in the correlation factor combination, a combination weight parameter corresponding to the correlation factor combination is obtained.
  • Step S640 according to the combination weight parameter and the reference weight parameter of each correlation factor in the correlation factor combination, obtain the risk conduction weight parameter of the correlation factor at each correlation level.
  • the combination of correlation factors refers to the multiple correlation factors existing between two risk nodes.
  • the combination weight parameter corresponding to the combination of correlation factors refers to the overall transmission influence proportion from one risk user to another risk user. For example, risk user A passes through normal users B is indirectly associated with risk user C, and there are multiple associations between user A and user B, and there are also multiple associations between user B and user C, so there are multiple association factor combinations between user A and user C , But the corresponding combination weight parameters of different correlation factor combinations are the same.
  • the risk of the correlation factor at each correlation level can be obtained by solving the relational expression group and data fitting processing Transmission weight parameter.
  • the parameters of the initial risk transmission correlation map are updated to obtain the optimized risk transmission correlation map After that, it also includes:
  • the device tag information of the target user is obtained.
  • the first risk data of the target user is obtained.
  • the transmission risk probability of the associated risk user to the target user is obtained, and the second risk data is obtained.
  • the risk probability of the target user is obtained.
  • the user can conduct risk analysis on the target user.
  • the risk identification and device label risk weight parameters of each device label in the device label information can be passed through the proportion of each device label
  • the first risk data of the target user is obtained by cumulative calculation.
  • the associated risk users of the target user can be searched, including directly associated risk users and indirectly associated risk users.
  • the benchmark weight parameter and the risk transmission weight parameter of the correlation factor at different correlation levels, as well as the user risk data of the associated risk user with the target user, can obtain the transmission risk probability of the associated risk user to the target user through risk transmission, and get the second
  • the risk data is calculated by obtaining the weighting parameters of the first risk data and the second risk data to obtain the risk probability of the target user.
  • the weighting parameters of the first risk data and the second risk data can be obtained by analyzing the risk weight parameters of the device tag of each risk node, the risk transmission result of each risk node, and the user risk data of the risk node.
  • the method further includes:
  • the risk level is divided, and the service suggestion corresponding to each risk level is determined.
  • the risk probability of the target user find the risk level of the target user's risk probability, and output the service suggestion corresponding to the risk level.
  • the fraud risk probability of other nodes associated with it can be calculated through risk conduction, which can be drilled according to the graph. Effect, define the threshold of fraud probability. For example, risk probability (corresponding percentile data) [0, 40): low risk; [40, 80): fraud risk; [80, +] suspected fraud. Find the risk level of the target user's risk probability, and output the service suggestions corresponding to the risk level.
  • the risk node transmission map can be marked according to different colors, so that the investigator can directly give the basis for rejection or follow-up key investigation.
  • a device for optimizing risk transmission correlation graphs which includes a risk node acquisition module 300, a device tag risk weight parameter determination module 400, a reference weight parameter determination module 500, and a risk transmission weight
  • the parameter determination module 600 and the initial risk conduction correlation map optimization module 700 are provided, which includes a risk node acquisition module 300, a device tag risk weight parameter determination module 400, a reference weight parameter determination module 500, and a risk transmission weight. among them:
  • the risk node obtaining module 300 is used to obtain the risk node in the initial risk conduction association graph, the user risk data of the risk node, and the device label information of the risk user corresponding to the risk node.
  • the device tag risk weight parameter determination module 400 is configured to determine the device tag risk weight parameter of the risk node according to the user risk data and the device tag information.
  • the benchmark weight parameter determination module 500 is used to determine the correlation factor between the directly related risk nodes, the correlation factor combination between the indirectly related risk nodes, and the correlation factor of each correlation factor in the correlation factor combination according to the correlation between the risk nodes.
  • the correlation level is determined, and the reference weight parameter of the correlation factor is determined according to the user risk data of the directly correlated risk node.
  • the risk conduction weight parameter determination module 600 is configured to determine the risk conduction weight parameter of the correlation factor at each correlation level according to the reference weight parameter of each correlation factor in the correlation factor combination and the correlation level of the correlation factor.
  • the initial risk conduction correlation map optimization module 700 is configured to update the initial risk conduction according to the device tag risk weight parameter, the reference weight parameter of the correlation factor, and the risk conduction weight parameter of the correlation factor at different correlation levels.
  • the parameters of the correlation map are used to obtain an optimized risk conduction correlation map.
  • the above-mentioned risk conduction correlation map optimization device analyzes the equipment label information of risk nodes corresponding to risk users on the basis of the initial risk conduction correlation map through the correlation factors between directly related risk nodes, and obtains the equipment label risk weight parameters.
  • the correlation factor between directly related risk nodes determines the reference weight parameter of the correlation factor, and the risk conduction of each correlation factor under each correlation level is determined by the correlation level of each correlation factor in the correlation factor combination between the indirect correlation risk nodes
  • the weight parameters are used to optimize the initial risk transmission correlation map, fully excavate the deep connections between the nodes of the correlation map, and clarify the specific transmission parameters between risk nodes.
  • the risk conduction correlation map optimization device further includes an initial risk conduction correlation map construction module, which is used to obtain user data information of each user used to construct the initial risk conduction correlation map, and label each user
  • an initial risk conduction correlation map construction module which is used to obtain user data information of each user used to construct the initial risk conduction correlation map, and label each user
  • screen risk users who carry user risk data among the users mark the associated graph nodes corresponding to the risk users as risk nodes; obtain preset data dimensions, and extract from the user data information The data value corresponding to the data dimension; find the feature association graph node that contains the same data value, and mark the data dimension corresponding to the data value as the correlation factor between the feature association graph nodes; according to the association
  • the map node and the correlation factor construct the initial risk conduction correlation map, and the initial risk conduction correlation map is marked with risk nodes.
  • the initial risk conduction association map building module is also used to:
  • the device tag risk weight parameter determination module 400 is also used for the device tag risk weight parameter determination module 400 .
  • the risk transmission weight parameter determination module 600 is further used for:
  • the combination weight parameter corresponding to the correlation factor combination is obtained; according to the combination weight parameter and the user risk data
  • the reference weight parameter of each correlation factor in the correlation factor combination is used to obtain the risk transmission weight parameter of the correlation factor at each correlation level.
  • the risk transmission correlation map optimization device further includes a target user risk analysis module for:
  • the target user When a risk analysis request from a target user is received, the target user’s device label information is obtained; the target user’s first risk data is obtained according to the device label information; the target user’s first risk data is obtained according to the optimized risk conduction association map, searching for the The associated risk user of the target user; according to the reference weight parameter of the correlation factor and the risk transmission weight parameter of the correlation factor at different correlation levels, the transmission risk probability of the associated risk user to the target user is obtained, and the first 2. Risk data; obtaining the risk probability of the target user according to the first risk data and the second risk data.
  • the target user risk analysis module is also used for:
  • the risk conduction correlation map optimization device further includes a risk analysis result output module for:
  • the risk level is divided, and the service suggestion corresponding to each risk level is determined; according to the risk probability of the target user, the target user is searched The risk level of the risk probability of which is located, and output the service proposal corresponding to the risk level.
  • Each module in the above risk conduction correlation map optimization device can be implemented in whole or in part by software, hardware, and a combination thereof.
  • the foregoing modules may be embedded in the form of hardware or independent of the processor in the computer device, or may be stored in the memory of the computer device in the form of software, so that the processor can call and execute the operations corresponding to the foregoing modules.
  • a computer device is provided.
  • the computer device may be a server, and its internal structure diagram may be as shown in FIG. 7.
  • the computer equipment includes a processor, a memory, a network interface and a database connected through a system bus. Among them, the processor of the computer device is used to provide calculation and control capabilities.
  • the memory of the computer device includes a non-volatile storage medium and an internal memory.
  • the non-volatile storage medium stores an operating system, computer readable instructions, and a database.
  • the internal memory provides an environment for the operation of the operating system and computer-readable instructions in the non-volatile storage medium.
  • the database of the computer equipment is used to store the optimization data of the risk transmission correlation map.
  • the network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer-readable instructions are executed by the processor, a method for optimizing the risk conduction correlation map is realized.
  • FIG. 7 is only a block diagram of part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied.
  • the specific computer device may Including more or fewer parts than shown in the figure, or combining some parts, or having a different arrangement of parts.
  • a computer device includes a memory and one or more processors.
  • the memory stores computer-readable instructions.
  • the one or more processors implement the methods provided in any of the embodiments of the present application. The steps of the optimization method of risk transmission correlation map.
  • One or more non-volatile computer-readable storage media storing computer-readable instructions.
  • the one or more processors implement any one of the embodiments of the present application. Provide the steps of the optimization method of risk transmission correlation map.
  • Non-volatile memory may include read only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory.
  • ROM read only memory
  • PROM programmable ROM
  • EPROM electrically programmable ROM
  • EEPROM electrically erasable programmable ROM
  • Volatile memory may include random access memory (RAM) or external cache memory.
  • RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous chain Channel (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

Landscapes

  • Engineering & Computer Science (AREA)
  • Databases & Information Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Computational Linguistics (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Animal Behavior & Ethology (AREA)
  • Fuzzy Systems (AREA)
  • Mathematical Physics (AREA)
  • Probability & Statistics with Applications (AREA)
  • Software Systems (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

一种风险传导关联图谱优化方法,包括:获取初始风险传导关联图谱中的风险节点及其对应的用户风险数据与设备标签信息,进而确定风险节点的设备标签风险权重参数,根据直接关联的风险节点间的关联因子和对应的用户风险数据,确定关联因子的基准权重参数,根据间接关联的风险节点间的关联因子组合,以及组合中各关联因子的关联等级,结合基准权重参数,确定关联因子在各关联等级下的风险传导权重参数,更新初始风险传导关联图谱的参数,优化风险传导关联图谱,充分挖掘关联图谱各节点之间的深度联系,来实现对初始风险传导关联图谱的优化,有利于提高风险分析效率。

Description

风险传导关联图谱优化方法、装置、计算机设备和存储介质
相关申请的交叉引用
本申请要求于2019年05月20日提交中国专利局,申请号为2019104197238,申请名称为“风险传导关联图谱优化方法、装置、计算机设备和存储介质”的中国专利申请的优先权,其全部内容通过引用结合在本申请中。
技术领域
本申请涉及一种风险传导关联图谱优化方法、装置、计算机设备和存储介质。
背景技术
知识图谱又称为科学知识图谱,是显示知识发展进程与结构关系的一系列各种不同的图形,用可视化技术描述知识资源及其载体,挖掘、分析、构建、绘制和显示知识及它们之间的相互联系。在通用意义上,知识图谱主要的目标是用来描述真实世界中存在的各种实体和概念,以及它们之间的关联关系。
传统的知识图谱,可以通过实体以及实体间的关联关系来构建得到,可用于进行实体间的风险传导,在一些应用场景中,可以基于某些因子信息建立节点间的关联关系来构建得到风险传导关联图谱,表征节点之间的关系网络,进行风险传导分析。
然而,发明人意识到,由于传统方式构建的知识图谱中,节点数据维度的多样性,在一定程度上导致了风险传导关联图谱关联关系的错综复杂。一方面,传统的风险传导关联图谱有着巨大的节点基础数据,一定程度上影响了关系传导的计算复杂程度,另一方面,确定节点间的关系传导计算需要通过梳理节点间关联关系,并进行实时计算得到,进而导致了这种风险传导关联图谱在进行风险传导分析时,存在着数据分析效率不高的问题,这种风险传导关联图谱亟待优化。
发明内容
根据本申请公开的各种实施例,提供一种风险传导关联图谱优化方法、装置、计算机设备和存储介质。
一种风险传导关联图谱优化方法包括:
获取初始风险传导关联图谱中的风险节点、所述风险节点的用户风险数据、以及与所述风险节点对应风险用户的设备标签信息;
根据所述用户风险数据和所述设备标签信息,确定所述风险节点的设备标签风险权重参数;
根据风险节点间的关联关系,确定直接关联的风险节点间的关联因子、间接关联的风 险节点间的关联因子组合,以及所述关联因子组合中各关联因子的关联等级,并根据所述直接关联的风险节点的用户风险数据,确定所述关联因子的基准权重参数;
根据所述关联因子组合中各关联因子的基准权重参数,以及所述关联因子的关联等级,确定所述关联因子在各关联等级下的风险传导权重参数;及
根据所述设备标签风险权重参数、所述关联因子的基准权重参数以及所述关联因子在不同关联等级下的风险传导权重参数,更新所述初始风险传导关联图谱的参数,得到优化的风险传导关联图谱。
一种风险传导关联图谱优化装置包括:
风险节点获取模块,用于获取初始风险传导关联图谱中的风险节点、所述风险节点的用户风险数据、以及与所述风险节点对应风险用户的设备标签信息;
设备标签风险权重参数确定模块,用于根据所述用户风险数据和所述设备标签信息,确定所述风险节点的设备标签风险权重参数;
基准权重参数确定模块,用于根据风险节点间的关联关系,确定直接关联的风险节点间的关联因子、间接关联的风险节点间的关联因子组合,以及所述关联因子组合中各关联因子的关联等级,并根据所述直接关联的风险节点的用户风险数据,确定所述关联因子的基准权重参数;
风险传导权重参数确定模块,用于根据所述关联因子组合中各关联因子的基准权重参数,以及所述关联因子的关联等级,确定所述关联因子在各关联等级下的风险传导权重参数;及
初始风险传导关联图谱优化模块,用于根据所述设备标签风险权重参数、所述关联因子的基准权重参数以及所述关联因子在不同关联等级下的风险传导权重参数,更新所述初始风险传导关联图谱的参数,得到优化的风险传导关联图谱。
一种计算机设备,包括存储器和一个或多个处理器,所述存储器中储存有计算机可读指令,所述计算机可读指令被所述处理器执行时,使得所述一个或多个处理器执行以下步骤:
获取初始风险传导关联图谱中的风险节点、所述风险节点的用户风险数据、以及与所述风险节点对应风险用户的设备标签信息;
根据所述用户风险数据和所述设备标签信息,确定所述风险节点的设备标签风险权重参数;
根据风险节点间的关联关系,确定直接关联的风险节点间的关联因子、间接关联的风险节点间的关联因子组合,以及所述关联因子组合中各关联因子的关联等级,并根据所述直接关联的风险节点的用户风险数据,确定所述关联因子的基准权重参数;
根据所述关联因子组合中各关联因子的基准权重参数,以及所述关联因子的关联等级,确定所述关联因子在各关联等级下的风险传导权重参数;及
根据所述设备标签风险权重参数、所述关联因子的基准权重参数以及所述关联因子在 不同关联等级下的风险传导权重参数,更新所述初始风险传导关联图谱的参数,得到优化的风险传导关联图谱。
一个或多个存储有计算机可读指令的非易失性计算机可读存储介质,计算机可读指令被一个或多个处理器执行时,使得一个或多个处理器执行以下步骤:
获取初始风险传导关联图谱中的风险节点、所述风险节点的用户风险数据、以及与所述风险节点对应风险用户的设备标签信息;
根据所述用户风险数据和所述设备标签信息,确定所述风险节点的设备标签风险权重参数;
根据风险节点间的关联关系,确定直接关联的风险节点间的关联因子、间接关联的风险节点间的关联因子组合,以及所述关联因子组合中各关联因子的关联等级,并根据所述直接关联的风险节点的用户风险数据,确定所述关联因子的基准权重参数;
根据所述关联因子组合中各关联因子的基准权重参数,以及所述关联因子的关联等级,确定所述关联因子在各关联等级下的风险传导权重参数;及
根据所述设备标签风险权重参数、所述关联因子的基准权重参数以及所述关联因子在不同关联等级下的风险传导权重参数,更新所述初始风险传导关联图谱的参数,得到优化的风险传导关联图谱。
本申请的一个或多个实施例的细节在下面的附图和描述中提出。本申请的其它特征和优点将从说明书、附图以及权利要求书变得明显。
附图说明
为了更清楚地说明本申请实施例中的技术方案,下面将对实施例中所需要使用的附图作简单地介绍,显而易见地,下面描述中的附图仅仅是本申请的一些实施例,对于本领域普通技术人员来讲,在不付出创造性劳动的前提下,还可以根据这些附图获得其它的附图。
图1为根据一个或多个实施例中风险传导关联图谱优化方法的应用场景图。
图2为根据一个或多个实施例中风险传导关联图谱优化方法的流程示意图。
图3为根据一个或多个实施例中构建初始风险传导关联图谱的流程示意图。
图4为再一个实施例中风险传导关联图谱优化方法的流程示意图。
图5为又一个实施例中风险传导关联图谱优化方法的流程示意图。
图6为根据一个或多个实施例中风险传导关联图谱优化装置的框图。
图7为根据一个或多个实施例中计算机设备的框图。
具体实施方式
为了使本申请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,此处描述的具体实施例仅仅用以解释本申请,并不用于限定本申请。
本申请提供的风险传导关联图谱优化方法,可以应用于如图1所示的应用环境中。终端102与服务器104通过网络进行通信。服务器104获取初始风险传导关联图谱中的风险节点、风险节点的用户风险数据、以及风险节点对应风险用户的设备标签信息,根据设备标签信息和用户风险数据,确定风险节点的设备标签风险权重参数,根据风险节点间的关联关系,确定直接关联的风险节点间的关联因子、间接关联的风险节点间的关联因子组合,以及关联因子组合中各关联因子的关联等级,并根据直接关联的风险节点的用户风险数据,确定关联因子的基准权重参数根据关联因子组合中各关联因子的基准权重参数,以及关联因子的关联等级,确定关联因子在各关联等级下的风险传导权重参数;根据设备标签风险权重参数、关联因子的基准权重参数以及关联因子在不同关联等级下的风险传导权重参数,更新初始风险传导关联图谱的参数,得到优化的风险传导关联图谱,并将优化的风险传导关联图谱推送至终端102。终端102可以但不限于是各种个人计算机、笔记本电脑、智能手机、平板电脑和便携式可穿戴设备,服务器104可以用独立的服务器或者是多个服务器组成的服务器集群来实现。
在一个实施例中,如图2所示,提供了一种风险传导关联图谱优化方法,以该方法应用于图1中的服务器为例进行说明,包括以下步骤:
步骤S300,获取初始风险传导关联图谱中的风险节点、风险节点的用户风险数据、以及与风险节点对应风险用户的设备标签信息。
初始风险传导关联图谱是指以各用户为节点,以用户之间存在相同数据为连接关系,该相同数据对应的数据维度作为节点间的关联因子而构建的用于表征各用户之间的联系的知识图谱。根据各关联谱图节点中携带有用户风险数据的节点,通过进行风险节点标注,可以得到标注有风险节点的初始风险传导关联图谱。用户风险数据是指已经产生了风险行为,风险行为包括违约行为、逾期行为等,根据风险行为具体对应的违约情况进行风险评估,得到的用于表征用户风险情况的评分值数据,评分值数据的分值大小与风险行为数量以及风险行为的风险程度相关。用户风险数据可以通过预先构建的风险数据库中各用户的风险行为记录分析得到,风险行为记录中包括用户的各项风险行为,以及风险行为的风险程度。在实施例中,风险行为次数越多,或风险行为对应的情节越严重,该用户对应的用户风险数据数值越大。未产生风险行为的用户不包含用户风险数据,该用户为正常用户,在初始风险传导关联图谱中对应节点为非风险节点。在初始风险传导关联图谱中,存在风险节点也存在非风险节点,根据是否存在用户风险数据,可以确定初始风险传导关联图谱的各节点对应的用户是否为风险用户,从而对初始风险传导关联图谱的进行风险节点标注。风险用户产生的风险行为一般在用户使用的移动设备中存在着相关的记录,通过对用户移动设备中的各数据记录按设定的数据维度进行风险分析,得到各维度的设备标签信息。
步骤S400,根据用户风险数据和设备标签信息,确定风险节点的设备标签风险权重参数。
风险节点是包含有用户风险数据的节点,故风险节点对应的用户为风险用户,风险用户产生的风险行为在用户使用的移动设备中存在着相关的数据记录,例如,用户的浏览记录,搜索记录,不良APP使用记录,贷款APP使用记录,设备刷机记录,同手机号多设备、同设备不同手机号数等不同的数据维度,用户移动设备中的每个数据维度的数据记录对应一个设备标签,根据具体的数据记录的数据内容与风险数据库的匹配,可以确定各设备标签的风险标识为有风险或是无风险,根据初始风险传导关联图谱中众多的风险节点,以及各风险节点的用户风险数据,可以得到多个携带有风险标识的设备标签组合,从而得到各设备标签对用户风险数据影响权重,通过对影响权重的累加计算,得到风险节点的设备标签风险权重参数。
步骤S500,根据风险节点间的关联关系,确定直接关联的风险节点间的关联因子、间接关联的风险节点间的关联因子组合,以及关联因子组合中各关联因子的关联等级,并根据直接关联的风险节点的用户风险数据,确定关联因子的基准权重参数。
关联关系是指两个节点间的联系,关联关系包括直接关联和间接关联,举例来说,用户A、用户B以及用户C均为风险用户,当用户A与用户B在家庭住址相同,依据家庭住址这一关联因子关联到一起,则用户A与用户B之间的关联关系为直接关联,当用户B与用户C的工作单位相同,则依据工作单位这一关联因子关联到一起,则用户B与用户C之间的关联关系也为直接关联,从而,用户A与用户C之间的关联关系为通过用户B间接关联。根据风险用户的用户风险数据,以及存在直接关联关系的风险节点间的关联因子,构建多个关联因子与用户风险数据的关联关系式组,由于用户风险数据为已知数据,可以通过数据计算,以及数据计算结果的数据拟合处理,得到各关联因子的基准权重参数,基准权重参数用于表征各关联因子在各节点间的传导性的强弱。
间接关联的风险节点间可能存在一个或多个中间节点,从而对应存在多个关联因子,例如,用户A,用户D、用户E为风险用户,用户B和用户C为正常用户时,用户A与用户B通过家庭住址这一关联因子直接关联,用户B与用户C通过工作单位这一关联因子直接关联,用户B与用户D通过同一手机号直接关联,用户C与用户E通过同一软件账号直接关联,根据上述直接关联关系,可以确定用户A与用户D间接关联,用户A与用户E间接关联,用户D与用户E间接关联,各风险用户间的关联因子组合不同,根据关联因子的数量,可以确定关联因子组合中各关联因子的关联等级。其中,关联因子的关联等级与关联传导方向有关。例如,用户A—家庭住址—用户B—手机号—用户D,以及用户D—手机号—用户B—家庭住址—用户A,这两组关联因子组合中,手机号这一关联因子对应的关联等级是不同的,家庭住址这一关联因子的关联等级也是不同的。
步骤S600,根据关联因子组合中各关联因子的基准权重参数,以及关联因子的关联等级,确定关联因子在各关联等级下的风险传导权重参数。
根据各风险用户的用户风险数据,各关联因子的基准权重参数以及存在间接关联关系的风险节点间的关联因子,构建多个关联因子组合与用户风险数据的关联关系式组,由于 用户风险数据为已知数据,可以通过数据计算,以及数据计算结果的数据拟合处理,得到各关联因子在不同关联等级下的风险传导权重参数。
步骤S700,根据设备标签风险权重参数、关联因子的基准权重参数以及关联因子在不同关联等级下的风险传导权重参数,更新初始风险传导关联图谱的参数,得到优化的风险传导关联图谱。
在原有初始风险传导关联图谱的基础上,通过分析各设备标签风险权重参数、节点之间的关联因子的基准权重参数,以及各关联因子在不同关联等级下的风险传导权重参数,对风险传导关联图谱进行优化,根据优化的风险传导关联图谱,结合标签和因子权重比及关联度的差异信息推导出风险传导的概率数据。在优化的风险传导关联图谱中,如果某一节点已可明确标识出是存在欺诈的风险节点,就可以得出跟其关联项的其他未知风险情况的节点的欺诈风险概率,提高了风险概率结果的准确性,进而评估出基于风险概率,可评估是否为欺诈者,进而予以拒绝或是重点调查,节省人力和时间。
上述风险传导关联图谱优化方法,在初始风险传导关联图谱的基础上,通过存在直接关联的风险节点间的关联因子,分析风险节点对应风险用户的设备标签信息,得到设备标签风险权重参数,通过存在直接关联的风险节点间的关联因子,确定关联因子的基准权重参数,通过存在间接关联的风险节点间的关联因子组合中各关联因子的关联等级,确定各关联因子在各关联等级下的风险传导权重参数,从而对初始风险传导关联图谱进行优化,充分挖掘了关联图谱各节点之间的深度联系,明确了风险节点间的具体的传导参数,利用参数优化的风险传导关联图谱,可直接进行风险的传导,简化了节点间关联关系的梳理与实时计算过程,通过分析得到的设备标签风险权重参数,作为节点数据的评估参数,简化了节点基础数据量,便于进行风险传导分析,提高了风险分析效率。
在另一个实施例中,如图3所示,步骤S300,获取初始风险传导关联图谱中的风险节点以及风险节点的用户风险数据之前,还包括初始风险传导关联图谱的构建,初始风险传导关联图谱的构建过程包括步骤S120至步骤S180。
步骤S120,获取用于构建初始风险传导关联图谱的各用户的用户数据信息,并将各用户标注为关联图谱节点,筛选各用户中携带有用户风险数据的风险用户,将风险用户对应的关联图谱节点标注为风险节点。
步骤S140,获取预设的数据维度,从用户数据信息中提取与数据维度对应的数据值。
步骤S160,查找包含相同数据值的各特征关联图谱节点,并将数据值对应的数据维度标记为特征关联图谱节点间的关联因子。
步骤S180,根据所述关联图谱节点和所述关联因子,构建所述初始风险传导关联图谱,所述初始风险传导关联图谱标注有风险节点。
初始风险传导关联图谱用于展示用户间的关联关系,将包含用户数据信息的用户标记为标注为关联图谱节点,用户数据信息可以是包括风险数据的已违约用户的各维度的数据信息,也可以是处于正常状态的未违约用户的各维度的数据信息,数据信息提取消息用于 根据其携带的数据维度,从用户数据信息中提取所需要的数据维度的相关数据值,预设的数据维度是指根据实际需求,预先进行分析得到的用于表征用户间联系强弱的各数据维度。通过设定数据维度,可以是初始风险传导模型更有条理更加规范。若两个节点间存在相同的数据值,则可以将该节点确定为特征关联图谱节点,筛选各用户中携带有用户风险数据的风险用户,将风险用户对应的关联图谱节点标注为风险节点,根据数据值建立两个特征关联图谱节点间的关联关系,并将该数据值对应的数据维度标记为特征关联图谱节点间的关联因子,基于关联图谱节点和关联因子,构建得到标注有风险节点的初始风险传导关联图谱。
在其中一个实施例中,如图3所示,步骤S140,根据获取的数据信息提取消息中携带的多个数据维度信息,提取用户数据信息中与各数据维度对应的数据值之前,还包括步骤S132和步骤S134。
步骤S132,获取已有风险用户的样本数据,对样本数据中的风险数据进行数据维度标注。
步骤S134,统计维度数据在样本数据中的共现频率,预设的数据维度为共现频率满足设定阈值的数据维度。
在实施例中,可以先根据筛选得到的数据维度,生成携带有预设的数据维度信息的数据信息提取消息,在应用时,预设数据维度可以通过从提取数据信息提取消息中提取得到。共现是指在不同数据中同时出现相同字段的现象,同一维度数据的共现频率是指,根据相同字段的维度标签,确定该维度标签的具体字段在不同数据中同时出现的频率,获取大量已有风险用户的样本数据,对样本数据进行清洗,去除干扰信息,并对完成数据清洗的样本数据进行数据维度标注,统计各样本间的相同维度的数据的共现频率,筛选出共现频率满足设定阈值的数据维度,根据筛选得到的数据维度,生成携带有数据维度信息的数据信息提取消息,以确定初始关联图谱节点间的关联因子。共现频率的设定阈值可以根据实际需求进行设定。
在其中一个实施例中,如图4所示,步骤S400,获取风险节点对应风险用户的设备标签信息,根据设备标签信息和用户风险数据,确定风险节点的设备标签风险权重参数包括步骤S420至步骤S460。
步骤S420,获取风险节点对应风险用户的设备标签信息,提取设备标签信息中各设备标签携带的风险标识,风险标识包括有风险和无风险。
步骤S440,根据各风险标签携带的风险标识对应的组合结果,以及风险节点的用户风险数据,得到各风险标签对用户风险数据的影响权重。
步骤S460,根据风险节点中的携带有风险标识的风险标签,以及各风险标签影响权重,确定风险节点的设备标签风险权重参数。
设备标签信息为多个设备标签构成的标签集合,设备标签具体包括设备指纹、开机时长、模拟器、不良APP、贷款APP、浏览器记录、搜索记录、代理IP、通话记录、通讯 录、微信数据、设备刷机、设备黑名单、同手机号多设备、同设备不同手机号数、设备常用地等。各设备标签根据其具体数据记录,查找对应的风险数据库,当存在匹配的数据记录时,将该设备标签的风险标识标注为有风险,反之,则标注为无风险。根据标签集合中各设备标签对应的风险标识进行组合,得到风险标识组合结果,根据各风险节点对应的风险标识组合结果,以及各风险节点的用户风险数据,可以构建多个风险标识组合与用户风险数据的关系式组,通过对关系式组进行求解,可以得到多组计算结果,通过计算结果中的各个风险标签的对应数值,进行数据拟合处理,根据拟合结果得到各风险标签对用户风险数据的影响权重,根据各风险节点中的携带有风险标识的风险标签,以及各风险标签影响权重,通过累加计算得到该风险节点的设备标签风险权重参数。
在其中一个实施例中,如图5所示,步骤S600,根据关联因子组合中各关联因子的基准权重参数,以及关联因子的关联等级,确定关联因子在各关联等级下的风险传导权重参数包括步骤S620至步骤S640。
步骤S620,根据关联因子组合中各关联因子的关联等级,以及关联因子组合中对应的风险节点的用户风险数据,获得关联因子组合对应的组合权重参数。
步骤S640,根据组合权重参数以及关联因子组合中各关联因子的基准权重参数,获取关联因子在各关联等级下的风险传导权重参数。
关联因子组合是指两个风险节点间存在的多个关联因子,关联因子组合对应的组合权重参数是指从一个风险用户到另一个风险用户的整体传导影响比重,例如,风险用户A通过正常用户B与风险用户C间接关联,而用户A与用户B之间存在多种关联关系,用户B与用户C之间也存在多种关联关系,故用户A与用户C之间存在多个关联因子组合,但不同关联因子组合其对应的组合权重参数是相同的。通过多个组合权重参数,以及关联因子组合中各关联因子的基准权重参数,通过构建关系式组,可以通过对关系式组的求解与数据拟合处理,得到关联因子在各关联等级下的风险传导权重参数。
在其中一个实施例中,根据设备标签风险权重参数、关联因子的基准权重参数以及关联因子在不同关联等级下的风险传导权重参数,更新初始风险传导关联图谱的参数,得到优化的风险传导关联图谱之后,还包括:
当接收到目标用户的风险分析请求时,获取目标用户的设备标签信息。
根据设备标签信息,获得目标用户的第一风险数据。
根据优化的风险传导关联图谱,查找目标用户的关联风险用户。
根据关联因子的基准权重参数以及关联因子在不同关联等级下的风险传导权重参数,获得关联风险用户对目标用户的传导风险概率,得到第二风险数据。
根据第一风险数据和第二风险数据,获得目标用户的风险概率。
当风险传导关联图谱优化之后,可以用户对目标用户进行风险分析,通过获取用户的设备标签信息,可以根据设备标签信息中各设备标签的风险标识和设备标签风险权重参数,通过各设备标签的比重累加计算得到目标用户的第一风险数据,根据用户的用户信息 和优化的风险传导关联图谱,可以查找目标用户的关联风险用户,包括直接关联的风险用户和间接关联的风险用户,根据关联因子的基准权重参数以及关联因子在不同关联等级下的风险传导权重参数,以及与目标用户的关联风险用户的用户风险数据,可以通过风险传导,获得关联风险用户对目标用户的传导风险概率,得到第二风险数据,通过获取第一风险数据与第二风险数据的加权参数,计算得到目标用户的风险概率。其中,第一风险数据与第二风险数据的加权参数可以根据各风险节点的设备标签风险权重参数、各风险节点的风险传导结果以及风险节点的用户风险数据分析得到。
在其中一个实施例中,根据第一风险数据和第二风险数据,获得目标用户的风险概率之后,还包括:
根据获取的样本数据中各用户的风险概率以及各用户的违约情况,划分风险等级,并确定各风险等级对应的服务建议。
根据目标用户的风险概率,查找目标用户的风险概率所在的风险等级,并输出与风险等级对应的服务建议。
在其中一个实施例中,如果风险节点传导图谱的某一节点已可明确标识出是存在欺诈的风险节点,可以通过风险传导计算得出跟其关联的其他节点的欺诈风险概率,可根据图谱演练效果,定义欺诈概率阈值。比如说风险概率(对应的百分制数据)[0,40):低风险;[40,80):有欺诈风险;[80,+]疑似欺诈。查找目标用户的风险概率所在的风险等级,并输出与风险等级对应的服务建议,风险节点传导图谱中可以根据不同的颜色标识出来,便于案调者直接给与拒绝或是后续重点调查的依据。
应该理解的是,虽然图2-5的流程图中的各个步骤按照箭头的指示依次显示,但是这些步骤并不是必然按照箭头指示的顺序依次执行。除非本文中有明确的说明,这些步骤的执行并没有严格的顺序限制,这些步骤可以以其它的顺序执行。而且,图2-5中的至少一部分步骤可以包括多个子步骤或者多个阶段,这些子步骤或者阶段并不必然是在同一时刻执行完成,而是可以在不同的时刻执行,这些子步骤或者阶段的执行顺序也不必然是依次进行,而是可以与其它步骤或者其它步骤的子步骤或者阶段的至少一部分轮流或者交替地执行。
在其中一个实施例中,如图6所示,提供了一种风险传导关联图谱优化装置,包括风险节点获取模块300、设备标签风险权重参数确定模块400、基准权重参数确定模块500、风险传导权重参数确定模块600以及初始风险传导关联图谱优化模块700。其中:
风险节点获取模块300,用于获取初始风险传导关联图谱中的风险节点、所述风险节点的用户风险数据、以及与所述风险节点对应风险用户的设备标签信息。
设备标签风险权重参数确定模块400,用于根据所述用户风险数据和所述设备标签信息,确定所述风险节点的设备标签风险权重参数。
基准权重参数确定模块500,用于根据风险节点间的关联关系,确定直接关联的风险节点间的关联因子、间接关联的风险节点间的关联因子组合,以及所述关联因子组合中各 关联因子的关联等级,并根据所述直接关联的风险节点的用户风险数据,确定所述关联因子的基准权重参数。
风险传导权重参数确定模块600,用于根据所述关联因子组合中各关联因子的基准权重参数,以及所述关联因子的关联等级,确定所述关联因子在各关联等级下的风险传导权重参数。
初始风险传导关联图谱优化模块700,用于根据所述设备标签风险权重参数、所述关联因子的基准权重参数以及所述关联因子在不同关联等级下的风险传导权重参数,更新所述初始风险传导关联图谱的参数,得到优化的风险传导关联图谱。
上述风险传导关联图谱优化装置,在初始风险传导关联图谱的基础上,通过存在直接关联的风险节点间的关联因子,分析风险节点对应风险用户的设备标签信息,得到设备标签风险权重参数,通过存在直接关联的风险节点间的关联因子,确定关联因子的基准权重参数,通过存在间接关联的风险节点间的关联因子组合中各关联因子的关联等级,确定各关联因子在各关联等级下的风险传导权重参数,从而对初始风险传导关联图谱进行优化,充分挖掘了关联图谱各节点之间的深度联系,明确了风险节点间的具体的传导参数,利用参数优化的风险传导关联图谱,可直接进行风险的传导,简化了节点间关联关系的梳理与实时计算过程,通过分析得到的设备标签风险权重参数,作为节点数据的评估参数,简化了节点基础数据量,便于进行风险传导分析,提高了风险分析效率。
在其中一个实施例中,风险传导关联图谱优化装置还包括初始风险传导关联图谱构建模块,用于:获取用于构建初始风险传导关联图谱的各用户的用户数据信息,并将所述各用户标注为关联图谱节点,筛选各所述用户中携带有用户风险数据的风险用户,将所述风险用户对应的关联图谱节点标注为风险节点;获取预设的数据维度,从所述用户数据信息中提取与所述数据维度对应的数据值;查找包含相同所述数据值的特征关联图谱节点,并将所述数据值对应的数据维度标记为所述特征关联图谱节点间的关联因子;根据所述关联图谱节点和所述关联因子,构建所述初始风险传导关联图谱,所述初始风险传导关联图谱标注有风险节点。
在其中一个实施例中,初始风险传导关联图谱构建模块,还用于:
获取已有风险用户的样本数据,对所述样本数据中的风险数据进行数据维度标注;统计维度数据在所述样本数据中的共现频率,所述预设的数据维度为所述共现频率满足设定阈值的数据维度。
在其中一个实施例中,设备标签风险权重参数确定模块400,还用于
提取所述设备标签信息中设备标签携带的风险标识,所述风险标识包括有风险和无风险;根据所述风险标签携带的风险标识的组合结果,以及所述风险节点的用户风险数据,得到所述风险标签对所述用户风险数据的影响权重;根据所述风险节点中的携带有风险标识的风险标签,以及所述风险标签影响权重,确定所述风险节点的设备标签风险权重参数。
在其中一个实施例中,风险传导权重参数确定模块600,还用于:
根据所述关联因子组合中各关联因子的关联等级,以及所述关联因子组合中对应的风险节点的用户风险数据,获得所述关联因子组合对应的组合权重参数;根据所述组合权重参数以及所述关联因子组合中各关联因子的基准权重参数,获取所述关联因子在各关联等级下的风险传导权重参数。
在其中一个实施例中,风险传导关联图谱优化装置还包括目标用户风险分析模块,用于:
当接收到目标用户的风险分析请求时,获取所述目标用户的设备标签信息;根据所述设备标签信息,获得目标用户的第一风险数据;根据所述优化的风险传导关联图谱,查找所述目标用户的关联风险用户;根据所述关联因子的基准权重参数以及所述关联因子在不同关联等级下的风险传导权重参数,获得所述关联风险用户对所述目标用户的传导风险概率,得到第二风险数据;根据所述第一风险数据和所述第二风险数据,获得所述目标用户的风险概率。
在其中一个实施例中,目标用户风险分析模块,还用于:
根据风险节点的设备标签风险权重参数、风险传导结果以及用户风险数据,确定第一风险数据的加权参数与第二风险数据的加权参数;对所述第一风险数据和所述第二风险数据进行加权处理,得到所述目标用户的风险概率。
在其中一个实施例中,风险传导关联图谱优化装置还包括风险分析结果输出模块,用于:
根据获取的样本数据中各用户的风险概率以及所述各用户的违约情况,划分风险等级,并确定各所述风险等级对应的服务建议;根据所述目标用户的风险概率,查找所述目标用户的风险概率所在的风险等级,并输出与所述风险等级对应的服务建议。
关于风险传导关联图谱优化装置的具体限定可以参见上文中对于风险传导关联图谱优化方法的限定,在此不再赘述。上述风险传导关联图谱优化装置中的各个模块可全部或部分通过软件、硬件及其组合来实现。上述各模块可以硬件形式内嵌于或独立于计算机设备中的处理器中,也可以以软件形式存储于计算机设备中的存储器中,以便于处理器调用执行以上各个模块对应的操作。
在一个实施例中,提供了一种计算机设备,该计算机设备可以是服务器,其内部结构图可以如图7所示。该计算机设备包括通过系统总线连接的处理器、存储器、网络接口和数据库。其中,该计算机设备的处理器用于提供计算和控制能力。该计算机设备的存储器包括非易失性存储介质、内存储器。该非易失性存储介质存储有操作系统、计算机可读指令和数据库。该内存储器为非易失性存储介质中的操作系统和计算机可读指令的运行提供环境。该计算机设备的数据库用于存储风险传导关联图谱优化数据。该计算机设备的网络接口用于与外部的终端通过网络连接通信。该计算机可读指令被处理器执行时以实现一种风险传导关联图谱优化方法。
本领域技术人员可以理解,图7中示出的结构,仅仅是与本申请方案相关的部分结构 的框图,并不构成对本申请方案所应用于其上的计算机设备的限定,具体的计算机设备可以包括比图中所示更多或更少的部件,或者组合某些部件,或者具有不同的部件布置。
一种计算机设备,包括存储器和一个或多个处理器,存储器中储存有计算机可读指令,计算机可读指令被处理器执行时,使得一个或多个处理器实现本申请任意一个实施例中提供的风险传导关联图谱优化方法的步骤。
一个或多个存储有计算机可读指令的非易失性计算机可读存储介质,计算机可读指令被一个或多个处理器执行时,使得一个或多个处理器实现本申请任意一个实施例中提供的风险传导关联图谱优化方法的步骤。
本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流程,是可以通过计算机可读指令来指令相关的硬件来完成,所述的计算机可读指令可存储于一非易失性计算机可读取存储介质中,该计算机可读指令在执行时,可包括如上述各方法的实施例的流程。其中,本申请所提供的各实施例中所使用的对存储器、存储、数据库或其它介质的任何引用,均可包括非易失性和/或易失性存储器。非易失性存储器可包括只读存储器(ROM)、可编程ROM(PROM)、电可编程ROM(EPROM)、电可擦除可编程ROM(EEPROM)或闪存。易失性存储器可包括随机存取存储器(RAM)或者外部高速缓冲存储器。作为说明而非局限,RAM以多种形式可得,诸如静态RAM(SRAM)、动态RAM(DRAM)、同步DRAM(SDRAM)、双数据率SDRAM(DDRSDRAM)、增强型SDRAM(ESDRAM)、同步链路(Synchlink)DRAM(SLDRAM)、存储器总线(Rambus)直接RAM(RDRAM)、直接存储器总线动态RAM(DRDRAM)、以及存储器总线动态RAM(RDRAM)等。
以上实施例的各技术特征可以进行任意的组合,为使描述简洁,未对上述实施例中的各个技术特征所有可能的组合都进行描述,然而,只要这些技术特征的组合不存在矛盾,都应当认为是本说明书记载的范围。
以上所述实施例仅表达了本申请的几种实施方式,其描述较为具体和详细,但并不能因此而理解为对发明专利范围的限制。应当指出的是,对于本领域的普通技术人员来说,在不脱离本申请构思的前提下,还可以做出若干变形和改进,这些都属于本申请的保护范围。因此,本申请专利的保护范围应以所附权利要求为准。

Claims (20)

  1. 一种风险传导关联图谱优化方法,包括:
    获取初始风险传导关联图谱中的风险节点、所述风险节点的用户风险数据、以及与所述风险节点对应风险用户的设备标签信息;
    根据所述用户风险数据和所述设备标签信息,确定所述风险节点的设备标签风险权重参数;
    根据风险节点间的关联关系,确定直接关联的风险节点间的关联因子、间接关联的风险节点间的关联因子组合,以及所述关联因子组合中各关联因子的关联等级,并根据所述直接关联的风险节点的用户风险数据,确定所述关联因子的基准权重参数;
    根据所述关联因子组合中各关联因子的基准权重参数,以及所述关联因子的关联等级,确定所述关联因子在各关联等级下的风险传导权重参数;及
    根据所述设备标签风险权重参数、所述关联因子的基准权重参数以及所述关联因子在不同关联等级下的风险传导权重参数,更新所述初始风险传导关联图谱的参数,得到优化的风险传导关联图谱。
  2. 根据权利要求1所述的方法,其特征在于,在所述获取初始风险传导关联图谱中的风险节点、所述风险节点的用户风险数据、以及所述风险节点对应风险用户的设备标签信息之前,所述方法还包括:
    获取用于构建初始风险传导关联图谱的各用户的用户数据信息,并将所述各用户标注为关联图谱节点,筛选各所述用户中携带有用户风险数据的风险用户,将所述风险用户对应的关联图谱节点标注为风险节点;
    获取预设的数据维度,从所述用户数据信息中提取与所述数据维度对应的数据值;
    查找包含相同所述数据值的特征关联图谱节点,并将所述数据值对应的数据维度标记为所述特征关联图谱节点间的关联因子;及
    根据所述关联图谱节点和所述关联因子,构建所述初始风险传导关联图谱,所述初始风险传导关联图谱标注有风险节点。
  3. 根据权利要求2所述的方法,其特征在于,所述方法还包括:
    获取已有风险用户的样本数据,对所述样本数据中的风险数据进行数据维度标注;及统计维度数据在所述样本数据中的共现频率,所述预设的数据维度为所述共现频率满足设定阈值的数据维度。
  4. 根据权利要求1所述的方法,其特征在于,所述根据所述用户风险数据和所述设备标签信息,确定所述风险节点的设备标签风险权重参数包括:
    提取所述设备标签信息中设备标签携带的风险标识,所述风险标识包括有风险和无风险;
    根据所述风险标签携带的风险标识的组合结果,以及所述风险节点的用户风险数据,得到所述风险标签对所述用户风险数据的影响权重;及
    根据所述风险节点中的携带有风险标识的风险标签,以及所述风险标签影响权重,确定所述风险节点的设备标签风险权重参数。
  5. 根据权利要求1所述的方法,其特征在于,所述根据所述关联因子组合中各关联因子的基准权重参数,以及所述关联因子的关联等级,确定所述关联因子在各关联等级下的风险传导权重参数包括:
    根据所述关联因子组合中各关联因子的关联等级,以及所述关联因子组合中对应的风险节点的用户风险数据,获得所述关联因子组合对应的组合权重参数;及
    根据所述组合权重参数以及所述关联因子组合中各关联因子的基准权重参数,获取所述关联因子在各关联等级下的风险传导权重参数。
  6. 根据权利要求1所述的方法,其特征在于,所述方法还包括:
    当接收到目标用户的风险分析请求时,获取所述目标用户的设备标签信息;
    根据所述设备标签信息,获得目标用户的第一风险数据;
    根据所述优化的风险传导关联图谱,查找所述目标用户的关联风险用户;
    根据所述关联因子的基准权重参数以及所述关联因子在不同关联等级下的风险传导权重参数,获得所述关联风险用户对所述目标用户的传导风险概率,得到第二风险数据;及
    根据所述第一风险数据和所述第二风险数据,获得所述目标用户的风险概率。
  7. 根据权利要求6所述的方法,其特征在于,所述根据所述第一风险数据和所述第二风险数据,获得所述目标用户的风险概率包括:
    根据风险节点的设备标签风险权重参数、风险传导结果以及用户风险数据,确定第一风险数据的加权参数与第二风险数据的加权参数;及
    对所述第一风险数据和所述第二风险数据进行加权处理,得到所述目标用户的风险概率。
  8. 根据权利要求6所述的方法,其特征在于,所述根据所述第一风险数据和所述第二风险数据,获得所述目标用户的风险概率之后,还包括:
    根据获取的样本数据中各用户的风险概率以及所述各用户的违约情况,划分风险等级,并确定各所述风险等级对应的服务建议;及
    根据所述目标用户的风险概率,查找所述目标用户的风险概率所在的风险等级,并输出与所述风险等级对应的服务建议。
  9. 一种风险传导关联图谱优化装置,包括:
    风险节点获取模块,用于获取初始风险传导关联图谱中的风险节点、所述风险节点的用户风险数据、以及与所述风险节点对应风险用户的设备标签信息;
    设备标签风险权重参数确定模块,用于根据所述用户风险数据和所述设备标签信息,确定所述风险节点的设备标签风险权重参数;
    基准权重参数确定模块,用于根据风险节点间的关联关系,确定直接关联的风险节点 间的关联因子、间接关联的风险节点间的关联因子组合,以及所述关联因子组合中各关联因子的关联等级,并根据所述直接关联的风险节点的用户风险数据,确定所述关联因子的基准权重参数;
    风险传导权重参数确定模块,用于根据所述关联因子组合中各关联因子的基准权重参数,以及所述关联因子的关联等级,确定所述关联因子在各关联等级下的风险传导权重参数;及
    初始风险传导关联图谱优化模块,用于根据所述设备标签风险权重参数、所述关联因子的基准权重参数以及所述关联因子在不同关联等级下的风险传导权重参数,更新所述初始风险传导关联图谱的参数,得到优化的风险传导关联图谱。
  10. 根据权利要求9所述的装置,其特征在于,所述风险传导关联图谱优化装置还包括初始风险传导关联图谱构建模块,用于:
    获取用于构建初始风险传导关联图谱的各用户的用户数据信息,并将所述各用户标注为关联图谱节点,筛选各所述用户中携带有用户风险数据的风险用户,将所述风险用户对应的关联图谱节点标注为风险节点;
    获取预设的数据维度,从所述用户数据信息中提取与所述数据维度对应的数据值;
    查找包含相同所述数据值的特征关联图谱节点,并将所述数据值对应的数据维度标记为所述特征关联图谱节点间的关联因子;及
    根据所述关联图谱节点和所述关联因子,构建所述初始风险传导关联图谱,所述初始风险传导关联图谱标注有风险节点。
  11. 一种计算机设备,包括存储器及一个或多个处理器,所述存储器中储存有计算机可读指令,所述计算机可读指令被所述一个或多个处理器执行时,使得所述一个或多个处理器执行以下步骤:
    获取初始风险传导关联图谱中的风险节点、所述风险节点的用户风险数据、以及与所述风险节点对应风险用户的设备标签信息;
    根据所述用户风险数据和所述设备标签信息,确定所述风险节点的设备标签风险权重参数;
    根据风险节点间的关联关系,确定直接关联的风险节点间的关联因子、间接关联的风险节点间的关联因子组合,以及所述关联因子组合中各关联因子的关联等级,并根据所述直接关联的风险节点的用户风险数据,确定所述关联因子的基准权重参数;
    根据所述关联因子组合中各关联因子的基准权重参数,以及所述关联因子的关联等级,确定所述关联因子在各关联等级下的风险传导权重参数;及
    根据所述设备标签风险权重参数、所述关联因子的基准权重参数以及所述关联因子在不同关联等级下的风险传导权重参数,更新所述初始风险传导关联图谱的参数,得到优化的风险传导关联图谱。
  12. 根据权利要求11所述的计算机设备,其特征在于,所述处理器执行所述计算机 可读指令时还执行以下步骤:
    获取用于构建初始风险传导关联图谱的各用户的用户数据信息,并将所述各用户标注为关联图谱节点,筛选各所述用户中携带有用户风险数据的风险用户,将所述风险用户对应的关联图谱节点标注为风险节点;
    获取预设的数据维度,从所述用户数据信息中提取与所述数据维度对应的数据值;
    查找包含相同所述数据值的特征关联图谱节点,并将所述数据值对应的数据维度标记为所述特征关联图谱节点间的关联因子;及
    根据所述关联图谱节点和所述关联因子,构建所述初始风险传导关联图谱,所述初始风险传导关联图谱标注有风险节点。
  13. 根据权利要求12所述的计算机设备,其特征在于,所述处理器执行所述计算机可读指令时还执行以下步骤:
    获取已有风险用户的样本数据,对所述样本数据中的风险数据进行数据维度标注;及统计维度数据在所述样本数据中的共现频率,所述预设的数据维度为所述共现频率满足设定阈值的数据维度。
  14. 根据权利要求11所述的计算机设备,其特征在于,所述处理器执行所述计算机可读指令时还执行以下步骤:
    提取所述设备标签信息中设备标签携带的风险标识,所述风险标识包括有风险和无风险;
    根据所述风险标签携带的风险标识的组合结果,以及所述风险节点的用户风险数据,得到所述风险标签对所述用户风险数据的影响权重;及
    根据所述风险节点中的携带有风险标识的风险标签,以及所述风险标签影响权重,确定所述风险节点的设备标签风险权重参数。
  15. 根据权利要求11所述的计算机设备,其特征在于,所述处理器执行所述计算机可读指令时还执行以下步骤:
    根据所述关联因子组合中各关联因子的关联等级,以及所述关联因子组合中对应的风险节点的用户风险数据,获得所述关联因子组合对应的组合权重参数;及
    根据所述组合权重参数以及所述关联因子组合中各关联因子的基准权重参数,获取所述关联因子在各关联等级下的风险传导权重参数。
  16. 一个或多个存储有计算机可读指令的非易失性计算机可读存储介质,所述计算机可读指令被一个或多个处理器执行时,使得所述一个或多个处理器执行以下步骤:
    获取初始风险传导关联图谱中的风险节点、所述风险节点的用户风险数据、以及与所述风险节点对应风险用户的设备标签信息;
    根据所述用户风险数据和所述设备标签信息,确定所述风险节点的设备标签风险权重参数;
    根据风险节点间的关联关系,确定直接关联的风险节点间的关联因子、间接关联的风 险节点间的关联因子组合,以及所述关联因子组合中各关联因子的关联等级,并根据所述直接关联的风险节点的用户风险数据,确定所述关联因子的基准权重参数;
    根据所述关联因子组合中各关联因子的基准权重参数,以及所述关联因子的关联等级,确定所述关联因子在各关联等级下的风险传导权重参数;及
    根据所述设备标签风险权重参数、所述关联因子的基准权重参数以及所述关联因子在不同关联等级下的风险传导权重参数,更新所述初始风险传导关联图谱的参数,得到优化的风险传导关联图谱。
  17. 根据权利要求16所述的存储介质,其特征在于,所述计算机可读指令被所述处理器执行时还执行以下步骤:
    获取用于构建初始风险传导关联图谱的各用户的用户数据信息,并将所述各用户标注为关联图谱节点,筛选各所述用户中携带有用户风险数据的风险用户,将所述风险用户对应的关联图谱节点标注为风险节点;
    获取预设的数据维度,从所述用户数据信息中提取与所述数据维度对应的数据值;
    查找包含相同所述数据值的特征关联图谱节点,并将所述数据值对应的数据维度标记为所述特征关联图谱节点间的关联因子;及
    根据所述关联图谱节点和所述关联因子,构建所述初始风险传导关联图谱,所述初始风险传导关联图谱标注有风险节点。
  18. 根据权利要求17所述的存储介质,其特征在于,所述计算机可读指令被所述处理器执行时还执行以下步骤:
    获取已有风险用户的样本数据,对所述样本数据中的风险数据进行数据维度标注;及统计维度数据在所述样本数据中的共现频率,所述预设的数据维度为所述共现频率满足设定阈值的数据维度。
  19. 根据权利要求16所述的存储介质,其特征在于,所述计算机可读指令被所述处理器执行时还执行以下步骤:
    提取所述设备标签信息中设备标签携带的风险标识,所述风险标识包括有风险和无风险;
    根据所述风险标签携带的风险标识的组合结果,以及所述风险节点的用户风险数据,得到所述风险标签对所述用户风险数据的影响权重;及
    根据所述风险节点中的携带有风险标识的风险标签,以及所述风险标签影响权重,确定所述风险节点的设备标签风险权重参数。
  20. 根据权利要求16所述的存储介质,其特征在于,所述计算机可读指令被所述处理器执行时还执行以下步骤:
    根据所述关联因子组合中各关联因子的关联等级,以及所述关联因子组合中对应的风险节点的用户风险数据,获得所述关联因子组合对应的组合权重参数;及
    根据所述组合权重参数以及所述关联因子组合中各关联因子的基准权重参数,获取所 述关联因子在各关联等级下的风险传导权重参数。
PCT/CN2019/102962 2019-05-20 2019-08-28 风险传导关联图谱优化方法、装置、计算机设备和存储介质 WO2020232879A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN201910419723.8 2019-05-20
CN201910419723.8A CN110245165B (zh) 2019-05-20 2019-05-20 风险传导关联图谱优化方法、装置和计算机设备

Publications (1)

Publication Number Publication Date
WO2020232879A1 true WO2020232879A1 (zh) 2020-11-26

Family

ID=67884540

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2019/102962 WO2020232879A1 (zh) 2019-05-20 2019-08-28 风险传导关联图谱优化方法、装置、计算机设备和存储介质

Country Status (2)

Country Link
CN (1) CN110245165B (zh)
WO (1) WO2020232879A1 (zh)

Cited By (11)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112734270A (zh) * 2021-01-19 2021-04-30 中国科学院地理科学与资源研究所 一种能源风险传导的测度方法、系统及数据平台
CN112785320A (zh) * 2021-02-01 2021-05-11 北京互金新融科技有限公司 信用风险的确定方法及装置、存储介质和电子设备
CN113240304A (zh) * 2021-05-20 2021-08-10 北京百度网讯科技有限公司 特征构建方法、装置、设备以及存储介质
CN113344405A (zh) * 2021-06-18 2021-09-03 北京百度网讯科技有限公司 基于知识图谱生成信息的方法、装置、设备、介质和产品
CN113449921A (zh) * 2021-07-06 2021-09-28 中国工商银行股份有限公司 一种风险成因识别方法、装置及存储介质
CN113742495A (zh) * 2021-09-07 2021-12-03 平安科技(深圳)有限公司 基于预测模型的评级特征权重确定方法及装置、电子设备
CN114022270A (zh) * 2021-10-29 2022-02-08 平安银行股份有限公司 资产数据处理方法、相关设备及介质
CN114595909A (zh) * 2020-12-03 2022-06-07 国家电网有限公司大数据中心 一种电费回收风险评估方法和系统
CN115345727A (zh) * 2022-08-19 2022-11-15 中电金信软件有限公司 一种识别欺诈贷款申请的方法及装置
CN116094974A (zh) * 2023-01-28 2023-05-09 北京京航计算通讯研究所 网络通信参数配置测试用例生成方法、装置及测试方法
CN118337716A (zh) * 2024-06-12 2024-07-12 杭州我知科技有限公司 基于tsn的云化plc网络架构的安全保障方法

Families Citing this family (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP7298025B2 (ja) 2019-10-24 2023-06-26 先臨三維科技股▲ふん▼有限公司 三次元スキャナー及び三次元走査方法
CN110796394A (zh) * 2019-11-13 2020-02-14 北京顶象技术有限公司 待标定节点标记方法及装置
CN111091287A (zh) * 2019-12-13 2020-05-01 南京三百云信息科技有限公司 风险对象识别方法、装置以及计算机设备
CN111401700B (zh) * 2020-03-05 2023-09-19 平安科技(深圳)有限公司 一种数据分析方法、装置、计算机系统及可读存储介质
CN111753517A (zh) * 2020-06-30 2020-10-09 北京来也网络科技有限公司 基于rpa及ai的文档对比方法、装置、设备及介质
CN112395351A (zh) * 2020-11-19 2021-02-23 平安普惠企业管理有限公司 可视化的识别群诉风险方法、装置、计算机设备及介质
CN112668836B (zh) * 2020-12-07 2024-04-05 数据地平线(广州)科技有限公司 一种面向风险图谱的关联风险证据高效挖掘与监控方法和装置
CN113762684A (zh) * 2020-12-14 2021-12-07 北京沃东天骏信息技术有限公司 新用户风险评估方法、装置、电子设备及介质
CN117353966A (zh) * 2022-06-29 2024-01-05 华为技术有限公司 网络风险评估方法及相关装置
CN115034520B (zh) * 2022-08-09 2023-01-10 太平金融科技服务(上海)有限公司深圳分公司 风险预测方法、装置、设备及存储介质
CN116452130B (zh) * 2023-03-17 2023-09-05 国网山东省电力公司淄博供电公司 一种基于知识图谱的变电设备信息管理系统及方法
CN116934507B (zh) * 2023-09-19 2023-12-26 国任财产保险股份有限公司 一种基于大数据驱动智能理赔方法及系统

Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1816595A1 (en) * 2006-02-06 2007-08-08 MediaKey Ltd. A method and a system for identifying potentially fraudulent customers in relation to network based commerce activities, in particular involving payment, and a computer program for performing said method
CN109583620A (zh) * 2018-10-11 2019-04-05 平安科技(深圳)有限公司 企业潜在风险预警方法、装置、计算机设备和存储介质
CN109636565A (zh) * 2018-10-16 2019-04-16 深圳壹账通智能科技有限公司 风险数据的处理方法、装置、设备及计算机可读存储介质
CN109685355A (zh) * 2018-12-19 2019-04-26 重庆百行智能数据科技研究院有限公司 企业风险识别方法、装置和计算机可读存储介质

Family Cites Families (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104572904B (zh) * 2014-12-25 2017-12-22 微梦创科网络科技(中国)有限公司 一种标签关联程度的确定方法及装置
CN109657917B (zh) * 2018-11-19 2022-04-29 平安科技(深圳)有限公司 评估对象的风险预警方法、装置、计算机设备和存储介质
CN109753527A (zh) * 2019-01-03 2019-05-14 深圳壹账通智能科技有限公司 异常企业挖掘方法、装置、计算机设备和存储介质

Patent Citations (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
EP1816595A1 (en) * 2006-02-06 2007-08-08 MediaKey Ltd. A method and a system for identifying potentially fraudulent customers in relation to network based commerce activities, in particular involving payment, and a computer program for performing said method
CN109583620A (zh) * 2018-10-11 2019-04-05 平安科技(深圳)有限公司 企业潜在风险预警方法、装置、计算机设备和存储介质
CN109636565A (zh) * 2018-10-16 2019-04-16 深圳壹账通智能科技有限公司 风险数据的处理方法、装置、设备及计算机可读存储介质
CN109685355A (zh) * 2018-12-19 2019-04-26 重庆百行智能数据科技研究院有限公司 企业风险识别方法、装置和计算机可读存储介质

Cited By (16)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN114595909A (zh) * 2020-12-03 2022-06-07 国家电网有限公司大数据中心 一种电费回收风险评估方法和系统
CN112734270A (zh) * 2021-01-19 2021-04-30 中国科学院地理科学与资源研究所 一种能源风险传导的测度方法、系统及数据平台
CN112734270B (zh) * 2021-01-19 2024-01-23 中国科学院地理科学与资源研究所 一种能源风险传导的测度方法、系统及数据平台
CN112785320B (zh) * 2021-02-01 2023-09-19 北京互金新融科技有限公司 信用风险的确定方法及装置、存储介质和电子设备
CN112785320A (zh) * 2021-02-01 2021-05-11 北京互金新融科技有限公司 信用风险的确定方法及装置、存储介质和电子设备
CN113240304A (zh) * 2021-05-20 2021-08-10 北京百度网讯科技有限公司 特征构建方法、装置、设备以及存储介质
CN113344405A (zh) * 2021-06-18 2021-09-03 北京百度网讯科技有限公司 基于知识图谱生成信息的方法、装置、设备、介质和产品
CN113344405B (zh) * 2021-06-18 2024-05-17 北京百度网讯科技有限公司 基于知识图谱生成信息的方法、装置、设备、介质和产品
CN113449921A (zh) * 2021-07-06 2021-09-28 中国工商银行股份有限公司 一种风险成因识别方法、装置及存储介质
CN113742495B (zh) * 2021-09-07 2024-02-23 平安科技(深圳)有限公司 基于预测模型的评级特征权重确定方法及装置、电子设备
CN113742495A (zh) * 2021-09-07 2021-12-03 平安科技(深圳)有限公司 基于预测模型的评级特征权重确定方法及装置、电子设备
CN114022270A (zh) * 2021-10-29 2022-02-08 平安银行股份有限公司 资产数据处理方法、相关设备及介质
CN114022270B (zh) * 2021-10-29 2024-05-24 平安银行股份有限公司 资产数据处理方法、相关设备及介质
CN115345727A (zh) * 2022-08-19 2022-11-15 中电金信软件有限公司 一种识别欺诈贷款申请的方法及装置
CN116094974A (zh) * 2023-01-28 2023-05-09 北京京航计算通讯研究所 网络通信参数配置测试用例生成方法、装置及测试方法
CN118337716A (zh) * 2024-06-12 2024-07-12 杭州我知科技有限公司 基于tsn的云化plc网络架构的安全保障方法

Also Published As

Publication number Publication date
CN110245165A (zh) 2019-09-17
CN110245165B (zh) 2023-04-11

Similar Documents

Publication Publication Date Title
WO2020232879A1 (zh) 风险传导关联图谱优化方法、装置、计算机设备和存储介质
CN107818344B (zh) 用户行为进行分类和预测的方法和系统
CN109829628A (zh) 基于大数据的风险预警方法、装置和计算机设备
WO2021184571A1 (zh) 动态表单生成方法、装置、计算机设备和存储介质
WO2020037942A1 (zh) 风险预测处理方法、装置、计算机设备和介质
CN110020422B (zh) 特征词的确定方法、装置和服务器
CN109543925B (zh) 基于机器学习的风险预测方法、装置、计算机设备和存储介质
CN109272396A (zh) 客户风险预警方法、装置、计算机设备和介质
CN109670837A (zh) 债券违约风险的识别方法、装置、计算机设备和存储介质
CN110321437B (zh) 一种语料数据处理方法、装置、电子设备及介质
CN110674360B (zh) 一种用于数据的溯源方法和系统
CN110750694A (zh) 数据标注实现方法及装置、电子设备、存储介质
CN109800354B (zh) 一种基于区块链存储的简历修改意图识别方法及系统
CN105306495A (zh) 用户识别方法和装置
CN113139025B (zh) 一种威胁情报的评价方法、装置、设备及存储介质
CN103944887A (zh) 基于隐条件随机场的入侵事件检测方法
CN112148305A (zh) 一种应用检测方法、装置、计算机设备和可读存储介质
CN112667875A (zh) 一种数据获取、数据分析方法、装置、设备及存储介质
CN107943906A (zh) 一种信息的收藏、展示方法和装置
CN109995605B (zh) 一种流量识别方法、装置以及计算机可读存储介质
CN111460268A (zh) 数据库查询请求的确定方法、装置和计算机设备
CN109992960B (zh) 一种伪造参数检测方法、装置、电子设备及存储介质
CN108255888B (zh) 一种数据处理方法及系统
CN113254672B (zh) 异常账号的识别方法、系统、设备及可读存储介质
CN113364702B (zh) 广告流量异常检测方法、系统、存储介质及电子设备

Legal Events

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

Ref document number: 19929408

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 19929408

Country of ref document: EP

Kind code of ref document: A1