CN112750030B - Risk pattern recognition method, apparatus, device and computer readable storage medium - Google Patents
Risk pattern recognition method, apparatus, device and computer readable storage medium Download PDFInfo
- Publication number
- CN112750030B CN112750030B CN202110032721.0A CN202110032721A CN112750030B CN 112750030 B CN112750030 B CN 112750030B CN 202110032721 A CN202110032721 A CN 202110032721A CN 112750030 B CN112750030 B CN 112750030B
- Authority
- CN
- China
- Prior art keywords
- target
- risk
- user
- pattern recognition
- heterogeneous
- Prior art date
- Legal status (The legal status 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 status listed.)
- Active
Links
- 238000012567 pattern recognition method Methods 0.000 title claims abstract description 33
- 230000006399 behavior Effects 0.000 claims abstract description 90
- 238000000034 method Methods 0.000 claims abstract description 26
- 238000012216 screening Methods 0.000 claims abstract description 19
- 238000003909 pattern recognition Methods 0.000 claims description 32
- 238000012549 training Methods 0.000 claims description 20
- 230000002452 interceptive effect Effects 0.000 claims 2
- 238000005516 engineering process Methods 0.000 abstract description 6
- 238000005065 mining Methods 0.000 abstract description 4
- 238000004891 communication Methods 0.000 description 5
- 238000010586 diagram Methods 0.000 description 4
- 238000009434 installation Methods 0.000 description 4
- 230000004931 aggregating effect Effects 0.000 description 2
- 238000004422 calculation algorithm Methods 0.000 description 2
- 238000010276 construction Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000013473 artificial intelligence Methods 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 229910003460 diamond Inorganic materials 0.000 description 1
- 239000010432 diamond Substances 0.000 description 1
- 238000000605 extraction Methods 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q40/00—Finance; Insurance; Tax strategies; Processing of corporate or income taxes
- G06Q40/03—Credit; Loans; Processing thereof
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
Abstract
The invention relates to the technical field of financial science and technology (Fintech). The invention discloses a risk pattern recognition method, a device, equipment and a readable storage medium, wherein the method and the device are used for solving the problems that in the prior art, the behavior information of a target user is acquired, a target risk pattern is determined according to actual recognition requirements, and a target element path set connected with multiple association relations is set according to the target risk pattern, so that the simple statistics of individual behaviors is avoided; screening out a target association relation set in the user behavior information through the target element path set, so that irrelevant information can be filtered out; by constructing a heterogeneous relationship network by utilizing association relations among different users, user-risk entities and/or risk entity-risk entities, a target risk mode is extracted by utilizing a target element path set on the heterogeneous relationship network so as to obtain a risk variable of a target user in the target risk mode, thereby realizing deeper mining on the behavior of the target user.
Description
Technical Field
The present invention relates to the technical field of financial science and technology (Fintech), and in particular, to a risk pattern recognition method, apparatus, device, and computer-readable storage medium.
Background
With the development of computer technology, more and more technologies (big data, distributed, blockchain Blockchain, artificial intelligence, etc.) are applied in the financial field, and the traditional financial industry is gradually changing to financial technology (Fintech), but due to the requirements of security and real-time performance of the financial industry, higher requirements are also put forward on the technologies. In the field of credit risk management, risk pattern identification of customers is one of the major importance. The existing risk pattern recognition usually aims at statistics of behaviors of individual individuals, namely statistics of simple relations among clients and risk entities, and the descriptions of the clients are often not deep enough, so that the effectiveness of recognition results finally obtained is low.
Disclosure of Invention
The invention mainly aims to provide a risk pattern recognition method, a risk pattern recognition device, risk pattern recognition equipment and a computer readable storage medium, and aims to solve the technical problem that the effectiveness of recognition results obtained by an existing risk pattern recognition mode is poor.
In order to achieve the above object, the present invention provides a risk pattern recognition method, including:
acquiring user behavior information of a target user, and determining a target risk mode to be identified, wherein the user behavior information comprises association relations between the user and risk entities, different users and/or different risk entities;
Setting a target meta-path set according to the target risk mode to obtain a target association relation set from the user behavior information based on the target meta-path set;
and constructing a heterogeneous relationship network according to the target association relationship set so as to identify a risk variable of the target user in the target risk mode by utilizing the target meta-path set on the heterogeneous relationship network.
Optionally, the user behavior information includes a list of user behaviors,
The step of setting a target meta-path set according to the target risk mode to obtain a target association relation set from the user behavior information based on the target meta-path set comprises the following steps:
Determining a target element path set according to the target risk mode, and extracting two adjacent path node pairs in the target element path set, wherein each path node is a user or a risk entity;
And screening an association relation table conforming to the path node pair from the user behavior list to serve as the target association relation set.
Optionally, the step of constructing a heterogeneous relationship network according to the target association relationship set includes:
The association relation table is associated according to the node sequence of the target element path set, and an initial relation network formed by a plurality of directional connection side lengths is obtained, wherein each connection side length corresponds to an association relation in the association relation table;
And constructing the heterogeneous relation network on the initial relation network based on a preset weight rule.
Optionally, the step of constructing the heterogeneous relationship network on the initial relationship network based on a preset weight rule includes:
and calculating the side length weight of each connection side length in the initial relation network according to the weight rule, and assigning the side length weight to the corresponding connection side length until all the connection side lengths are traversed, so as to obtain the heterogeneous relation network.
Optionally, the target meta-path set comprises a plurality of target meta-paths,
The step of identifying risk variables of the target user in the target risk mode by using the target meta-path set on the heterogeneous relation network comprises the following steps:
screening an assigned side length set corresponding to each target element path from the heterogeneous relation network, and determining an end variable value corresponding to each target element path;
calculating a weight continuous product of each assigned side length set, and multiplying each weight continuous product by a corresponding end variable value to obtain a weighted variable value corresponding to each target element path;
And adding the weighted variable values corresponding to each target element path to obtain the risk variable.
Optionally, the step of acquiring user behavior information of the target user and determining the target risk pattern to be identified includes:
Receiving a risk pattern recognition instruction, and acquiring the user behavior information from the risk pattern recognition instruction;
And acquiring the risk pattern specified by the risk pattern identification instruction to serve as the target risk pattern.
Optionally, after the step of identifying the risk variable of the target user in the target risk mode by using the target meta-path set on the heterogeneous relational network, the method further includes:
And acquiring an initial training data set, and adding the risk variable into the initial training data set to obtain a target training data set so as to perform risk modeling based on the target training data set.
In addition, to achieve the above object, the present invention also provides a risk pattern recognition apparatus, including:
The system comprises an incidence relation acquisition module, a target risk module and a target risk module, wherein the incidence relation acquisition module is used for acquiring user behavior information of a target user and determining a target risk mode to be identified, and the user behavior information comprises incidence relations among users, risk entities, different users and/or different risk entities;
The target relation determining module is used for setting a target meta-path set according to the target risk mode so as to obtain a target association relation set from the user behavior information based on the target meta-path set;
And the risk variable identification module is used for constructing a heterogeneous relationship network according to the target association relationship set so as to identify the risk variable of the target user in the target risk mode by utilizing the target meta-path set on the heterogeneous relationship network.
Optionally, the user behavior information includes a list of user behaviors,
The target relation determining module comprises:
the path node extraction unit is used for determining a target element path set according to the target risk mode and extracting two adjacent path node pairs in the target element path set, wherein each path node is a user or a risk entity;
And the association relation screening unit is used for screening an association relation table conforming to the path node pair from the user behavior list to be used as the target association relation set.
Optionally, the risk variable identification module includes:
a node order association unit, configured to associate the association relationship tables according to a node order of the target element path set, to obtain an initial relationship network formed by a plurality of connection side lengths with directionality, where each connection side length corresponds to an association relationship in the association relationship tables;
The heterogeneous network construction unit is used for constructing the heterogeneous relationship network on the initial relationship network based on a preset weight rule.
Optionally, the heterogeneous network construction unit is further configured to:
and calculating the side length weight of each connection side length in the initial relation network according to the weight rule, and assigning the side length weight to the corresponding connection side length until all the connection side lengths are traversed, so as to obtain the heterogeneous relation network.
Optionally, the target meta-path set comprises a plurality of target meta-paths,
The risk variable identification module comprises:
the amplitude side length screening unit is used for screening an assigned side length set corresponding to each target element path from the heterogeneous relation network and determining an end point variable value corresponding to each target element path;
the weighted variable obtaining unit is used for calculating the weight continuous product of each assigned side length set and multiplying each weight continuous product by a corresponding end variable value to obtain a weighted variable value corresponding to each target element path;
And the risk variable obtaining unit is used for adding the weighted variable values corresponding to each target element path to obtain the risk variable.
Optionally, the association relation obtaining module includes:
the behavior information acquisition unit is used for receiving a risk pattern recognition instruction and acquiring the user behavior information from the risk pattern recognition instruction;
And the target mode determining unit is used for acquiring the risk mode specified by the risk mode identification instruction to serve as the target risk mode.
Optionally, the risk pattern recognition apparatus further includes:
And the variable risk modeling module is used for acquiring an initial training data set, adding the risk variable into the initial training data set to obtain a target training data set, and performing risk modeling based on the target training data set.
In addition, to achieve the above object, the present invention also provides a risk pattern recognition apparatus including: the risk pattern recognition system comprises a memory, a processor and a risk pattern recognition program stored on the memory and capable of running on the processor, wherein the risk pattern recognition program realizes the steps of the risk pattern recognition method when being executed by the processor.
In addition, in order to achieve the above object, the present invention also provides a computer-readable storage medium having stored thereon a risk pattern recognition program which, when executed by a processor, implements the steps of the risk pattern recognition method as described above.
The invention provides a risk pattern recognition method, a risk pattern recognition device, risk pattern recognition equipment and a computer readable storage medium. According to the invention, the user behavior information of the target user is acquired, the target risk mode is determined according to the actual identification requirement, and the target element path set connected with multiple association relations is set according to the target risk mode, so that the simple statistics of individual behaviors in the prior art is avoided; screening out a target association relation set in the user behavior information through the currently determined target element path set, so that other irrelevant information can be filtered out; by constructing a heterogeneous relationship network by utilizing association relations among different users, risk entities and/or risk entities, extracting a target risk mode by utilizing a target element path set on the heterogeneous relationship network so as to obtain a risk variable of the target user in the target risk mode, deeper mining of the behavior of the target user is realized, the effectiveness of the finally obtained risk variable is improved, and the technical problem that the effectiveness of an identification result obtained by the existing risk mode identification mode is poor is solved.
Drawings
FIG. 1 is a schematic diagram of a device architecture of a hardware operating environment according to an embodiment of the present invention;
FIG. 2 is a flowchart of a risk pattern recognition method according to a first embodiment of the present invention;
FIG. 3 is a schematic diagram of an initial relationship network according to a second embodiment of the risk pattern recognition method of the present invention;
FIG. 4 is a schematic diagram of a heterogeneous network according to a second embodiment of the risk pattern recognition method of the present invention;
FIG. 5 is a flowchart illustrating the algorithm principle of a second embodiment of the risk pattern recognition method according to the present invention;
Fig. 6 is a schematic functional block diagram of a risk pattern recognition apparatus according to the present invention.
The achievement of the objects, functional features and advantages of the present invention will be further described with reference to the accompanying drawings, in conjunction with the embodiments.
Detailed Description
It should be understood that the specific embodiments described herein are for purposes of illustration only and are not intended to limit the scope of the invention.
Referring to fig. 1, fig. 1 is a schematic device structure of a hardware running environment according to an embodiment of the present invention.
As shown in fig. 1, the risk pattern recognition apparatus may include: a processor 1001, such as a CPU, a user interface 1003, a network interface 1004, a memory 1005, a communication bus 1002. Wherein the communication bus 1002 is used to enable connected communication between these components. The user interface 1003 may include a Display, an input unit such as a Keyboard (Keyboard), and the optional user interface 1003 may further include a standard wired interface, a wireless interface. The network interface 1004 may optionally include a standard wired interface, a wireless interface (e.g., WI-FI interface). The memory 1005 may be a high-speed RAM memory or a stable memory (non-volatile memory), such as a disk memory. The memory 1005 may also optionally be a storage device separate from the processor 1001 described above.
It will be appreciated by those skilled in the art that the device structure shown in fig. 1 is not limiting of the device and may include more or fewer components than shown, or may combine certain components, or a different arrangement of components.
As shown in fig. 1, an operating system, a network communication module, a user interface module, and a risk pattern recognition program may be included in a memory 1005 as one type of computer storage medium.
In the device shown in fig. 1, the network interface 1004 is mainly used for connecting to a background server, and performing data communication with the background server; the user interface 1003 is mainly used for connecting a client (programmer end) and communicating data with the client; and the processor 1001 may be configured to call a risk pattern recognition program stored in the memory 1005 and perform operations in the following risk pattern recognition method:
acquiring user behavior information of a target user, and determining a target risk mode to be identified, wherein the user behavior information comprises association relations between the user and risk entities, different users and/or different risk entities;
Setting a target meta-path set according to the target risk mode to obtain a target association relation set from the user behavior information based on the target meta-path set;
and constructing a heterogeneous relationship network according to the target association relationship set so as to identify a risk variable of the target user in the target risk mode by utilizing the target meta-path set on the heterogeneous relationship network.
Further, the user behavior information includes a user behavior list,
The step of setting a target meta-path set according to the target risk mode to obtain a target association relation set from the user behavior information based on the target meta-path set comprises the following steps:
Determining a target element path set according to the target risk mode, and extracting two adjacent path node pairs in the target element path set, wherein each path node is a user or a risk entity;
And screening an association relation table conforming to the path node pair from the user behavior list to serve as the target association relation set.
Further, the step of constructing a heterogeneous relationship network according to the target association relationship set includes:
The association relation table is associated according to the node sequence of the target element path set, and an initial relation network formed by a plurality of directional connection side lengths is obtained, wherein each connection side length corresponds to an association relation in the association relation table;
And constructing the heterogeneous relation network on the initial relation network based on a preset weight rule.
Further, the step of constructing the heterogeneous relationship network on the initial relationship network based on a preset weight rule includes:
and calculating the side length weight of each connection side length in the initial relation network according to the weight rule, and assigning the side length weight to the corresponding connection side length until all the connection side lengths are traversed, so as to obtain the heterogeneous relation network.
Further, the target element path set comprises a plurality of target element paths,
The step of identifying risk variables of the target user in the target risk mode by using the target meta-path set on the heterogeneous relation network comprises the following steps:
screening an assigned side length set corresponding to each target element path from the heterogeneous relation network, and determining an end variable value corresponding to each target element path;
calculating a weight continuous product of each assigned side length set, and multiplying each weight continuous product by a corresponding end variable value to obtain a weighted variable value corresponding to each target element path;
And adding the weighted variable values corresponding to each target element path to obtain the risk variable.
Further, the step of obtaining user behavior information of the target user and determining the target risk pattern to be identified includes:
Receiving a risk pattern recognition instruction, and acquiring the user behavior information from the risk pattern recognition instruction;
And acquiring the risk pattern specified by the risk pattern identification instruction to serve as the target risk pattern.
Further, after the step of identifying the risk variable of the target user in the target risk mode by using the target meta-path set on the heterogeneous relational network, the processor 1001 may be configured to invoke a risk mode identification program stored in the memory 1005 and perform operations in the following risk mode identification method:
And acquiring an initial training data set, and adding the risk variable into the initial training data set to obtain a target training data set so as to perform risk modeling based on the target training data set.
Based on the hardware structure, the embodiment of the risk pattern recognition method is provided.
In order to solve the problems, the invention provides a risk pattern recognition method, namely, a target user is aimed at obtaining user behavior information of the target user, determining a target risk pattern according to actual recognition requirements, and setting a target element path set connected with multiple association relations according to the target risk pattern, so that the method of simply counting individual behaviors in the prior art is avoided; screening out a target association relation set in the user behavior information through the currently determined target element path set, so that other irrelevant information can be filtered out; by constructing a heterogeneous relationship network by utilizing association relations among different users, risk entities and/or risk entities, extracting a target risk mode by utilizing a target element path set on the heterogeneous relationship network so as to obtain a risk variable of the target user in the target risk mode, deeper mining of the behavior of the target user is realized, the effectiveness of the finally obtained risk variable is improved, and the technical problem that the effectiveness of an identification result obtained by the existing risk mode identification mode is poor is solved.
Referring to fig. 2, fig. 2 is a flowchart of a first embodiment of a risk pattern recognition method according to the present invention. The risk pattern recognition method comprises the following steps of;
Step S10, obtaining user behavior information of a target user, and determining a target risk mode to be identified, wherein the user behavior information comprises association relations between the user and risk entities, different users and/or different risk entities;
In this embodiment, the method is applied to the terminal device. The method is mainly suitable for application scenes of credit risk management. And the credit risk management is used for grading, classifying, reporting and managing risks through the procedures of risk identification, metering, monitoring, control and the like, so that the balanced development of risks and benefits is maintained, and the economic benefits of loans are improved. For personal credit services, risk management is to combine various factors that cause individuals to violate, identify the risk of individuals to violate, and manage and control the whole process. The target user refers to the particular individual applying and using the loan, typically identified by some type of ID (e.g., identification number, cell phone number), etc. The user behavior information refers to information recorded about behaviors made by the user in a credit related business scene, and specifically may include one or more of association relations between a target user and a risk entity, between the target user and other users, and between different risk entities, and specifically may exist in a text format, a table format, a picture format, and the like. The target risk pattern refers to one or more risk patterns that need to be identified in the current identification task. The risk entity refers to an entity having a certain interaction relation with the user, and may be a specific article or a virtual concept, for example, the user opens an account at a certain bank, and the client logs in a certain APP.
And the terminal preferably determines the target user in a certain risk identification task. The number of the target users may be one or a plurality of. The terminal obtains user behavior information of the target user through searching or directly obtaining, and determines a target risk mode to be identified in the identification task through identifying one or more risk modes from preset related information or receiving one or more risk modes designated currently and the like.
Step S20, a target meta-path set is set according to the target risk mode, and a target association relation set is obtained from the user behavior information based on the target meta-path set;
In this embodiment, the target meta-path set includes one or more target meta-paths. The target element path refers to a continuous association relationship between a user and a risk entity, for example, a plurality of existing association relationship tables, namely, a user- (relationship a) -risk entity, a risk entity- (relationship b) -user, a user- (action c) -user and a user- (action d) -risk entity are continuously associated, and the tables can form a path of the user-risk entity-user-risk entity, which is a target element path. The path starting point of the target element path corresponds to an object expected to be calculated by the recognition task, and the end point corresponds to attribute information expected to be collected. The target association relation set refers to a set of some part of association relation needed for building a relation network in the user behavior information.
And the terminal determines one or more continuous association relations, namely one or more target element paths, according to the currently determined one or more target risk modes, and the target element paths are juxtaposed as a target element path set. And the terminal screens out related target association relations from the user behavior information of the target user according to the target element path set to serve as a target association relation set.
And step S30, constructing a heterogeneous relationship network according to the target association relationship set, so as to identify a risk variable of the target user in the target risk mode by utilizing the target meta-path set on the heterogeneous relationship network.
In this embodiment, the heterogeneous relationship network refers to that a user and a risk entity can be associated, the user and the risk entity are regarded as nodes, the association relationship is regarded as edges, and the behavior intensity (such as the number of times, the number of days, etc.) of the user on the risk entity is set as the weight of the edges, so that different graph structures G can be generated for different behavior scenes, and a large number of behavior patterns effective for risk identification are contained. The risk variable refers to a representation value of a target user in a target risk mode, and the larger the value of the risk variable is, the higher the corresponding risk is.
The terminal takes various target association relations in the target association relation set as edges of the relation network, and takes the user and the risk entity as endpoints of the edges so as to construct the heterogeneous relation network. And the terminal calculates the risk variable of the target user in the target risk mode by utilizing the target element path set and combining a corresponding calculation method on the heterogeneous relation network constructed at present.
The invention provides a risk pattern recognition method. The risk pattern recognition method comprises the steps of obtaining user behavior information of a target user and determining a target risk pattern to be recognized, wherein the user behavior information comprises association relations between the user and risk entities, different users and/or different risk entities; setting a target meta-path set according to the target risk mode to obtain a target association relation set from the user behavior information based on the target meta-path set; and constructing a heterogeneous relationship network according to the target association relationship set so as to identify a risk variable of the target user in the target risk mode by utilizing the target meta-path set on the heterogeneous relationship network. According to the invention, the user behavior information of the target user is acquired, the target risk mode is determined according to the actual identification requirement, and the target element path set connected with multiple association relations is set according to the target risk mode, so that the simple statistics of individual behaviors in the prior art is avoided; screening out a target association relation set in the user behavior information through the currently determined target element path set, so that other irrelevant information can be filtered out; by constructing a heterogeneous relationship network by utilizing association relations among different users, risk entities and/or risk entities, extracting a target risk mode by utilizing a target element path set on the heterogeneous relationship network so as to obtain a risk variable of the target user in the target risk mode, deeper mining of the behavior of the target user is realized, the effectiveness of the finally obtained risk variable is improved, and the technical problem that the effectiveness of an identification result obtained by the existing risk mode identification mode is poor is solved.
Further, based on the first embodiment shown in fig. 2, a second embodiment of the risk pattern recognition method of the present invention is proposed. In this embodiment, the user behavior information includes a user behavior list,
Step S20 includes:
Determining a target element path set according to the target risk mode, and extracting two adjacent path node pairs in the target element path set, wherein each path node is a user or a risk entity;
And screening an association relation table conforming to the path node pair from the user behavior list to serve as the target association relation set.
In this embodiment, the path node pair refers to a combination of two adjacent nodes in the form of a user-user, a user-risk entity, or a risk entity-risk entity.
As a specific embodiment, a target meta-path is first set, for example, a behavior desired to be described in the present recognition task, that is, the target risk mode is: customer A installs some kind of APP, and most of this kind of APP uses the customer to like to purchase dog food, then the corresponding target element route that should set is: user- (installation) -APP- (use) -user- (purchase) -commodity [ attribute: whether dog food ], the starting point of the target element path is a client, the end point is a commodity, and the collection attribute is whether dog food. For the target element path, the relation table which is required to be screened from the user behavior list by the terminal, namely the target association relation combination comprises: user-APP (installation table), user-APP (usage table), user-commodity (purchase), commodity-attribute [ whether dog food ], the corresponding user node pair includes four pairs: user-install APP, user-use APP, user-purchase merchandise, merchandise-whether the attribute is dog food.
Further, the step of constructing a heterogeneous relationship network according to the target association relationship set includes:
The association relation table is associated according to the node sequence of the target element path set, and an initial relation network formed by a plurality of directional connection side lengths is obtained, wherein each connection side length corresponds to an association relation in the association relation table;
And constructing the heterogeneous relation network on the initial relation network based on a preset weight rule.
In this embodiment, along the target element path set including one user- (installation) -APP- (use) -user- (purchase) -commodity [ attribute: whether dog food ] in the specific embodiment, the terminal correlates each relevant association relationship table according to the order, and the initial relationship network after correlation is shown in fig. 3. The single circular area in the first left column represents the target user (customer a in the figure), the first column side represents the installation behavior, the diamond shape in the second column represents the installed APP, the second column side represents the target user's usage behavior of the installed APP, the circles in the third column and their representing several other users, the third column side represents the purchasing behavior of several other users, and the rectangles in the fourth column represent the merchandise purchased by the tow bar other users.
After the terminal builds the initial relation network, the terminal converts the initial relation network into a heterogeneous relation network based on a preset weight rule.
Further, the step of constructing the heterogeneous relationship network on the initial relationship network based on a preset weight rule includes:
and calculating the side length weight of each connection side length in the initial relation network according to the weight rule, and assigning the side length weight to the corresponding connection side length until all the connection side lengths are traversed, so as to obtain the heterogeneous relation network.
In this embodiment, the terminal calculates the weight corresponding to each connection side according to a preset weight rule, for example, only the commodity with the attribute being dog food should be selected from the connection sides of commodity-attribute, and then the commodity with the attribute not being dog food can be removed; and selecting the connection side length of which the use times exceed a certain preset times threshold value from the connection side lengths of the user-commodity (purchase), and removing the rest connection side lengths. Similarly, the connection side length of the thickened part in the heterogeneous relation network shown in fig. 4 can be screened, two thickened paths connected by three connection sides are arranged in fig. 4, and the connection side length of each thickened part can be given with a weight according to corresponding times.
Further, the target element path set comprises a plurality of target element paths,
The step of identifying risk variables of the target user in the target risk mode by using the target meta-path set on the heterogeneous relation network comprises the following steps:
screening an assigned side length set corresponding to each target element path from the heterogeneous relation network, and determining an end variable value corresponding to each target element path;
calculating a weight continuous product of each assigned side length set, and multiplying each weight continuous product by a corresponding end variable value to obtain a weighted variable value corresponding to each target element path;
And adding the weighted variable values corresponding to each target element path to obtain the risk variable.
In this embodiment, the risk variable calculating method may specifically be, for example, a target element path after two weighted values exist in a heterogeneous relational network, where one of the paths has weights of 2, 3, and 4, and the other path has weights of 5, 6, and 7, and the end variable value corresponding to each path is 1, and then the risk variable of the target user in the target risk mode is 2×3×4×1+5×6×7×1=258.
As a specific embodiment, a flow chart of the algorithm principle of the method is shown in fig. 5. Firstly, setting a target element path, selecting an association relation table between a user and a risk entity related to the target element path, then associating all tables according to the target element path to construct an initial relation network, screening a plurality of connection side lengths in the initial relation network by using the target element path and screening conditions, calculating corresponding weights of the connection side lengths, and finally, aggregating variables corresponding to all the screened and weighted connection side length ends to a starting point according to the weights to obtain a final risk variable.
Further, in the embodiment, the initial relation network is firstly constructed, then the corresponding side lengths in the network are weighted according to the preset rule to construct the heterogeneous relation network, and finally the risk variable of the target user is obtained by aggregating the weights corresponding to the target element paths, so that the behavior of the target user is mined more deeply, and the effectiveness of the finally obtained risk variable is improved.
Further, based on the first embodiment shown in fig. 2, a third embodiment of the risk pattern recognition method of the present invention is proposed. In this embodiment, step S10 includes:
Receiving a risk pattern recognition instruction, and acquiring the user behavior information from the risk pattern recognition instruction;
And acquiring the risk pattern specified by the risk pattern identification instruction to serve as the target risk pattern.
In this embodiment, when receiving the risk pattern recognition instruction, the terminal obtains user behavior information of the target user from the risk pattern recognition instruction, and obtains the currently specified risk pattern according to the risk pattern which is the instruction, and uses the currently specified risk pattern as the target risk pattern of the present recognition task.
Further, after step S30, the method further includes:
And acquiring an initial training data set, and adding the risk variable into the initial training data set to obtain a target training data set so as to perform risk modeling based on the target training data set.
In this embodiment, since the existing risk pattern recognition identifies the user-risk entity and the user-user relationship separately, only information related to personal behaviors can be captured, and the relationship between behaviors is lacking, the user's description is often not deep enough. Such as the number of customers the user makes a telephone call to contact, there is not much information itself, but if the user is known to be of a particular identity, the assistance in identifying the risk is greater. Therefore, after the risk variable is obtained, the risk variable can be used as effective information of a risk modeling process so as to fully utilize the data validity of the risk variable.
Furthermore, in the embodiment, the risk variable is added into the subsequent risk modeling process, so that the data used for risk modeling are more effective, and the effectiveness of the finally obtained risk model is further improved.
As shown in fig. 6, the present invention further provides a risk pattern recognition apparatus, including:
The association relation acquisition module 10 is configured to acquire user behavior information of a target user, and determine a target risk pattern to be identified, where the user behavior information includes association relations between the user and a risk entity, different users, and/or different risk entities;
A target relationship determining module 20, configured to set a target meta-path set according to the target risk mode, so as to obtain a target association relationship set from the user behavior information based on the target meta-path set;
The risk variable identification module 30 is configured to construct a heterogeneous relationship network according to the target association relationship set, so as to identify, on the heterogeneous relationship network, a risk variable of the target user in the target risk mode by using the target meta path set.
The method executed by each program module may refer to each embodiment of the risk pattern recognition method of the present invention, and will not be described herein.
The invention also provides risk pattern recognition equipment.
The risk pattern recognition device comprises a processor, a memory and a risk pattern recognition program stored on the memory and capable of running on the processor, wherein the risk pattern recognition program realizes the steps of the risk pattern recognition method when being executed by the processor.
The method implemented when the risk pattern recognition program is executed may refer to various embodiments of the risk pattern recognition method of the present invention, which are not described herein.
The invention also provides a computer readable storage medium.
The computer-readable storage medium of the present invention has stored thereon a risk pattern recognition program which, when executed by a processor, implements the steps of the risk pattern recognition method as described above.
The method implemented when the risk pattern recognition program is executed may refer to various embodiments of the risk pattern recognition method of the present invention, and will not be described herein.
It should be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or system. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article, or system that comprises the element.
The foregoing embodiment numbers of the present invention are merely for the purpose of description, and do not represent the advantages or disadvantages of the embodiments.
From the above description of the embodiments, it will be clear to those skilled in the art that the above-described embodiment method may be implemented by means of software plus a necessary general hardware platform, but of course may also be implemented by means of hardware, but in many cases the former is a preferred embodiment. Based on such understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art in the form of a software product stored in a storage medium (e.g. ROM/RAM, magnetic disk, optical disk) as described above, comprising instructions for causing a terminal device (which may be a mobile phone, a computer, a server, an air conditioner, or a network device, etc.) to perform the method according to the embodiments of the present invention.
The foregoing description is only of the preferred embodiments of the present invention, and is not intended to limit the scope of the invention, but rather is intended to cover any equivalents of the structures or equivalent processes disclosed herein or in the alternative, which may be employed directly or indirectly in other related arts.
Claims (7)
1. A risk pattern recognition method, applied to a credit risk management scenario, the risk pattern recognition method comprising:
Acquiring user behavior information of a target user, and determining a target risk mode to be identified, wherein the user behavior information comprises association relations between the user and risk entities, different users and/or different risk entities, and the risk entities are entities with interactive action relations with the user;
Setting a target meta-path set according to the target risk mode to obtain a target association relation set from the user behavior information based on the target meta-path set;
Constructing a heterogeneous relationship network according to the target association relationship set, so as to identify a risk variable of the target user in the target risk mode by utilizing the target element path set on the heterogeneous relationship network, wherein nodes of the heterogeneous relationship network are users and risk entities, edges of the heterogeneous relationship network are the association relationship, and weights of the edges are the behavior intensity of the users on the risk entities;
The user behavior information comprises a list of user behaviors,
The step of setting a target meta-path set according to the target risk mode to obtain a target association relation set from the user behavior information based on the target meta-path set comprises the following steps:
Determining a target element path set according to the target risk mode, and extracting two adjacent path node pairs in the target element path set, wherein each path node is a user or a risk entity;
screening out an association relation table conforming to the path node pair from the user behavior list to be used as the target association relation set;
The step of constructing the heterogeneous relationship network according to the target association relationship set comprises the following steps:
The association relation table is associated according to the node sequence of the target element path set, and an initial relation network formed by a plurality of directional connection side lengths is obtained, wherein each connection side length corresponds to an association relation in the association relation table;
constructing the heterogeneous relationship network on the initial relationship network based on a preset weight rule;
the step of constructing the heterogeneous relationship network on the initial relationship network based on a preset weight rule comprises the following steps:
and calculating the side length weight of each connection side length in the initial relation network according to the weight rule, and assigning the side length weight to the corresponding connection side length until all the connection side lengths are traversed, so as to obtain the heterogeneous relation network.
2. The method of claim 1, wherein the set of target meta-paths includes a plurality of target meta-paths,
The step of identifying risk variables of the target user in the target risk mode by using the target meta-path set on the heterogeneous relation network comprises the following steps:
screening an assigned side length set corresponding to each target element path from the heterogeneous relation network, and determining an end variable value corresponding to each target element path;
calculating a weight continuous product of each assigned side length set, and multiplying each weight continuous product by a corresponding end variable value to obtain a weighted variable value corresponding to each target element path;
And adding the weighted variable values corresponding to each target element path to obtain the risk variable.
3. The risk pattern recognition method of claim 1, wherein the step of acquiring user behavior information of the target user and determining the target risk pattern to be recognized comprises:
Receiving a risk pattern recognition instruction, and acquiring the user behavior information from the risk pattern recognition instruction;
And acquiring the risk pattern specified by the risk pattern identification instruction to serve as the target risk pattern.
4. A risk pattern recognition method according to any one of claims 1-3, wherein after the step of recognizing risk variables of the target user in the target risk pattern using the target meta-path set on the heterogeneous relational network, further comprises:
And acquiring an initial training data set, and adding the risk variable into the initial training data set to obtain a target training data set so as to perform risk modeling based on the target training data set.
5. A risk pattern recognition apparatus, characterized in that the risk pattern recognition apparatus comprises:
The system comprises an incidence relation acquisition module, a target risk module and a target risk module, wherein the incidence relation acquisition module is used for acquiring user behavior information of a target user and determining a target risk mode to be identified, the user behavior information comprises incidence relations among users, risk entities, different users and/or different risk entities, the risk entities are entities with interactive action relations with the users, and the user behavior information comprises a user behavior list;
The target relation determining module is used for setting a target meta-path set according to the target risk mode so as to obtain a target association relation set from the user behavior information based on the target meta-path set;
The risk variable identification module is used for constructing a heterogeneous relation network according to the target association relation set so as to identify a risk variable of the target user in the target risk mode by utilizing the target meta-path set on the heterogeneous relation network, wherein nodes of the heterogeneous relation network are users and risk entities, edges of the heterogeneous relation network are association relations between the users and the risk entities, and weights of the edges are behavior intensity of the users on the risk entities;
the target relationship determination module is further configured to:
Determining a target element path set according to the target risk mode, and extracting two adjacent path node pairs in the target element path set, wherein each path node is a user or a risk entity;
screening out an association relation table conforming to the path node pair from the user behavior list to be used as the target association relation set;
the risk variable identification module is further configured to:
The association relation table is associated according to the node sequence of the target element path set, and an initial relation network formed by a plurality of directional connection side lengths is obtained, wherein each connection side length corresponds to an association relation in the association relation table;
constructing the heterogeneous relationship network on the initial relationship network based on a preset weight rule;
the risk variable identification module is further configured to:
and calculating the side length weight of each connection side length in the initial relation network according to the weight rule, and assigning the side length weight to the corresponding connection side length until all the connection side lengths are traversed, so as to obtain the heterogeneous relation network.
6. A risk pattern recognition apparatus, characterized in that the risk pattern recognition apparatus comprises: memory, a processor and a risk pattern recognition program stored on the memory and executable on the processor, which risk pattern recognition program when executed by the processor implements the steps of the risk pattern recognition method of any one of claims 1 to 4.
7. A computer-readable storage medium, on which a risk pattern recognition program is stored, which, when executed by a processor, implements the steps of the risk pattern recognition method according to any one of claims 1 to 4.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110032721.0A CN112750030B (en) | 2021-01-11 | 2021-01-11 | Risk pattern recognition method, apparatus, device and computer readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110032721.0A CN112750030B (en) | 2021-01-11 | 2021-01-11 | Risk pattern recognition method, apparatus, device and computer readable storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN112750030A CN112750030A (en) | 2021-05-04 |
| CN112750030B true CN112750030B (en) | 2024-04-26 |
Family
ID=75650630
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110032721.0A Active CN112750030B (en) | 2021-01-11 | 2021-01-11 | Risk pattern recognition method, apparatus, device and computer readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112750030B (en) |
Families Citing this family (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113628032B (en) * | 2021-08-12 | 2024-02-09 | 上海上湖信息技术有限公司 | Method and device for determining user relationship |
| CN113724073A (en) * | 2021-09-09 | 2021-11-30 | 支付宝(杭州)信息技术有限公司 | Risk identification and control method and device |
| CN113780827A (en) * | 2021-09-14 | 2021-12-10 | 北京沃东天骏信息技术有限公司 | Article screening method and device, electronic equipment and computer readable medium |
| CN113704566B (en) * | 2021-10-29 | 2022-01-18 | 贝壳技术有限公司 | Identification number body identification method, storage medium and electronic equipment |
Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017041541A1 (en) * | 2015-09-08 | 2017-03-16 | 北京邮电大学 | Method for pushing recommendation information, and server and storage medium |
| CN108734469A (en) * | 2018-03-22 | 2018-11-02 | 阿里巴巴集团控股有限公司 | The method and apparatus for determining consumer's risk label undetermined |
| CN109492945A (en) * | 2018-12-14 | 2019-03-19 | 深圳壹账通智能科技有限公司 | Business risk identifies monitoring method, device, equipment and storage medium |
| CN109935277A (en) * | 2019-02-25 | 2019-06-25 | 大连理工大学 | Abnormal die body querying method based on first path in a kind of heterogeneous network |
| CN109934706A (en) * | 2017-12-15 | 2019-06-25 | 阿里巴巴集团控股有限公司 | A kind of transaction risk control method, apparatus and equipment based on graph structure model |
| CN110322254A (en) * | 2019-07-04 | 2019-10-11 | 同盾控股有限公司 | Online fraud recognition methods, device, medium and electronic equipment |
| CN110852437A (en) * | 2019-10-29 | 2020-02-28 | 天津大学 | Representation learning method for heterogeneous information network |
-
2021
- 2021-01-11 CN CN202110032721.0A patent/CN112750030B/en active Active
Patent Citations (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| WO2017041541A1 (en) * | 2015-09-08 | 2017-03-16 | 北京邮电大学 | Method for pushing recommendation information, and server and storage medium |
| CN109934706A (en) * | 2017-12-15 | 2019-06-25 | 阿里巴巴集团控股有限公司 | A kind of transaction risk control method, apparatus and equipment based on graph structure model |
| CN108734469A (en) * | 2018-03-22 | 2018-11-02 | 阿里巴巴集团控股有限公司 | The method and apparatus for determining consumer's risk label undetermined |
| CN109492945A (en) * | 2018-12-14 | 2019-03-19 | 深圳壹账通智能科技有限公司 | Business risk identifies monitoring method, device, equipment and storage medium |
| CN109935277A (en) * | 2019-02-25 | 2019-06-25 | 大连理工大学 | Abnormal die body querying method based on first path in a kind of heterogeneous network |
| CN110322254A (en) * | 2019-07-04 | 2019-10-11 | 同盾控股有限公司 | Online fraud recognition methods, device, medium and electronic equipment |
| CN110852437A (en) * | 2019-10-29 | 2020-02-28 | 天津大学 | Representation learning method for heterogeneous information network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN112750030A (en) | 2021-05-04 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112750030B (en) | Risk pattern recognition method, apparatus, device and computer readable storage medium | |
| CN107609708B (en) | User loss prediction method and system based on mobile game shop | |
| CN108665159A (en) | A kind of methods of risk assessment, device, terminal device and storage medium | |
| CN105023165A (en) | Method, device and system for controlling release tasks in social networking platform | |
| CN108985954A (en) | A kind of method and relevant device of incidence relation that establishing each mark | |
| CN105376223A (en) | Network identity relationship reliability calculation method | |
| CN109743286A (en) | A kind of IP type mark method and apparatus based on figure convolutional neural networks | |
| CN110675252A (en) | Risk assessment method and device, electronic equipment and storage medium | |
| CN106897880B (en) | Account risk assessment method and equipment | |
| CN108197795A (en) | The account recognition methods of malice group, device, terminal and storage medium | |
| CN111325422A (en) | Work order distribution method and system | |
| CN112587932A (en) | Game plug-in detection method and device, electronic equipment and storage medium | |
| CN110909246A (en) | Resource delivery method and device, electronic equipment and readable storage medium | |
| CN110399818A (en) | A kind of method and apparatus of risk profile | |
| CN114529210A (en) | Evaluation method and device based on virtual reality | |
| CN111126503B (en) | Training sample generation method and device | |
| CN113935832A (en) | Abnormal behavior detection processing method and device | |
| CN110399399B (en) | User analysis method, device, electronic equipment and storage medium | |
| CN110738571A (en) | transaction risk control method and related device | |
| CN109559013A (en) | Method for testing risk and device, electronic equipment and readable storage medium storing program for executing based on trivial games | |
| CN112419050B (en) | Credit evaluation method and device based on telephone communication network and social behavior | |
| CN104885114A (en) | Method and device for optimizing information diffusion between communities linked by interaction similarities | |
| CN110046910B (en) | Method and equipment for judging validity of transaction performed by customer through electronic payment platform | |
| CN111652525B (en) | Method, device, equipment and computer storage medium for analyzing risk tail end customer | |
| CN116823480A (en) | Abnormality detection processing method and device |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |