CN117787418A

CN117787418A - Risk identification method and device, storage medium and electronic equipment

Info

Publication number: CN117787418A
Application number: CN202311870367.4A
Authority: CN
Inventors: 陈中奇; 李怀松; 宋博文; 张天翼; 王维强; 靳雅; 张映
Original assignee: Alipay Hangzhou Information Technology Co Ltd
Current assignee: Alipay Hangzhou Information Technology Co Ltd
Priority date: 2023-12-29
Filing date: 2023-12-29
Publication date: 2024-03-29

Abstract

The specification discloses a risk identification method, a risk identification device, a storage medium and electronic equipment. Acquiring service data of a target user under a target service, and determining a target risk type required to be identified for the target user; according to the target risk type, determining a risk rule for identifying the target risk type in preset risk rules as a target rule; screening data matched with the target rule from the service data as available data; constructing prompt information according to the target risk type, the target rule and the available data; and inputting the prompt information into a pre-trained generation type model to obtain risk description of the target user, which is output by the generation type model.

Description

Risk identification method and device, storage medium and electronic equipment

Technical Field

The present disclosure relates to the field of computer technologies, and in particular, to a risk identification method, a risk identification device, a storage medium, and an electronic device.

Background

Nowadays, with the rapid development of artificial intelligence (Artificial Intelligence, AI) technology, content authoring based on AI has gradually begun to be applied in various fields. However, at present, AI authoring technology is still imperfect, and many disadvantages still exist.

Taking the field of wind control as an example, when analyzing whether a user has risks or not by means of AI, a common method is to input data of the user into a large language model constructed based on AI, and then ask a question to the large language model, so that the large language model answers to the risk description of the user. In the above process, since the large language model has strong randomness when outputting the content as the generative model, even if the same data and problems are input, the content output each time may have great difference, so it is difficult to ensure that the content generated by the AI is truly reliable and logically controllable at present. In practical application, uncertainty of generated content is likely to cause judgment errors, leakage of service data, privacy of users and the like.

Therefore, how to ensure the logic control of the content generated by AI is a urgent problem to be solved.

Disclosure of Invention

The present disclosure provides a risk identification method, apparatus, storage medium, and electronic device, so as to at least partially solve the foregoing problems of the prior art.

The technical scheme adopted in the specification is as follows:

the specification provides a risk identification method, which comprises the following steps:

acquiring service data of a target user under a target service, and determining a target risk type required to be identified for the target user;

According to the target risk type, determining a risk rule for identifying the target risk type in preset risk rules as a target rule;

screening data matched with the target rule from the service data as available data;

constructing prompt information according to the target risk type, the target rule and the available data;

and inputting the prompt information into a pre-trained generation type model to obtain risk description of the target user, which is output by the generation type model.

Optionally, according to the target risk type, determining, as a target rule, a risk rule for identifying the target risk type in preset risk rules, where the risk rule specifically includes:

inputting the business data into a pre-established map frame to obtain a logic map of the target user;

recalling a risk path matched with the target risk type in the logic map, wherein the risk path is used as a target risk path and consists of a risk type, a risk rule for obtaining the risk type and service data meeting the risk rule;

determining a risk rule contained in the target risk path as a target rule;

Screening the data matched with the target rule from the service data as available data, wherein the data specifically comprises:

and determining service data contained in the target risk path as available data.

Optionally, creating a map frame in advance specifically includes:

according to expert experience under the target service obtained from the historical information, determining each risk type existing in the target service and obtaining a risk rule of each risk type;

and taking each risk type as a first node, obtaining a risk rule of each risk type as a second node, taking the data frame as a third node, and constructing a map frame, wherein each second node is connected with one first node and one third node to serve as a risk path.

Optionally, determining the target risk type to be identified for the target user specifically includes:

in response to receiving query information, determining a target risk type contained in the query information;

recalling a risk path matched with the target risk type in the logic map as a target risk path, wherein the method specifically comprises the following steps:

determining a risk path including the target risk type as a candidate risk path in the logical graph;

Respectively extracting query text features of the query information and path text features of each candidate risk path;

for each candidate risk path, determining the similarity between the path text characteristics of the candidate risk path and the query text characteristics;

and determining the candidate risk paths with the similarity not smaller than a specified threshold as target risk paths.

Optionally, constructing prompt information according to the target risk type, the target rule and the available data, which specifically includes:

extracting keywords related to the target risk type from the target rule and the available data;

and constructing prompt information according to the target risk type and the keywords.

Optionally, the pre-training generative model specifically includes:

acquiring a sample wind control corpus;

masking the sample wind control corpus to obtain a wind control corpus with disturbance;

inputting the wind-controlled corpus with disturbance into a to-be-trained generation model to obtain the corpus to be optimized output by the generation model;

and training the generated model according to the difference between the corpus to be optimized and the sample wind control corpus.

Optionally, the method further comprises:

Acquiring a sample instruction and a labeling answer corresponding to the sample instruction;

inputting the sample instruction into the generated model to obtain an answer to be optimized output by the generated model;

and adjusting parameters of the generated model according to the difference between the answer to be optimized and the labeling answer.

The present specification provides a risk identification apparatus comprising:

the acquisition module is used for acquiring service data of a target user under a target service and determining a target risk type required to be identified for the target user;

the determining module is used for determining a risk rule for identifying the target risk type from preset risk rules according to the target risk type, and taking the risk rule as a target rule;

the matching module is used for screening the data matched with the target rule from the service data to be used as available data;

the construction module is used for constructing prompt information according to the target risk type, the target rule and the available data;

and the output module is used for inputting the prompt information into a pre-trained generation type model to obtain the risk description of the target user output by the generation type model.

Optionally, the determining module is specifically configured to input the service data into a pre-created spectrum frame to obtain a logic spectrum of the target user; recalling a risk path matched with the target risk type in the logic map, wherein the risk path is used as a target risk path and consists of a risk type, a risk rule for obtaining the risk type and service data meeting the risk rule; determining a risk rule contained in the target risk path as a target rule;

the matching module is specifically configured to determine service data included in the target risk path as available data.

Optionally, the device further includes a preset module, specifically configured to determine, according to expert experience under the target service obtained from the history information, each risk type existing in the target service and a risk rule for obtaining each risk type; and taking each risk type as a first node, obtaining a risk rule of each risk type as a second node, taking the data frame as a third node, and constructing a map frame, wherein each second node is connected with one first node and one third node to serve as a risk path.

Optionally, the acquiring module is specifically configured to determine, in response to receiving query information, a target risk type included in the query information;

the determining module is specifically configured to determine, in the logic atlas, a risk path including the target risk type as a candidate risk path; respectively extracting query text features of the query information and path text features of each candidate risk path; for each candidate risk path, determining the similarity between the path text characteristics of the candidate risk path and the query text characteristics; and determining the candidate risk paths with the similarity not smaller than a specified threshold as target risk paths.

Optionally, the construction module is specifically configured to extract, from the target rule and the available data, a keyword related to the target risk type; and constructing prompt information according to the target risk type and the keywords.

Optionally, the device further comprises a training module, specifically configured to obtain a sample wind control corpus; masking the sample wind control corpus to obtain a wind control corpus with disturbance; inputting the wind-controlled corpus with disturbance into a to-be-trained generation model to obtain the corpus to be optimized output by the generation model; and training the generated model according to the difference between the corpus to be optimized and the sample wind control corpus.

Optionally, the training module is further configured to obtain a sample instruction and a labeling answer corresponding to the sample instruction; inputting the sample instruction into the generated model to obtain an answer to be optimized output by the generated model; and adjusting parameters of the generated model according to the difference between the answer to be optimized and the labeling answer.

The present specification provides a computer readable storage medium storing a computer program which when executed by a processor implements the risk identification method described above.

The present specification provides an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the risk identification method described above when executing the program.

The above-mentioned at least one technical scheme that this specification adopted can reach following beneficial effect:

acquiring service data of a target user under a target service, and determining a target risk type required to be identified for the target user; according to the target risk type, determining a risk rule for identifying the target risk type in preset risk rules as a target rule; screening data matched with the target rule from the service data as available data; constructing prompt information according to the target risk type, the target rule and the available data; and inputting the prompt information into a pre-trained generation type model to obtain risk description of the target user, which is output by the generation type model.

When the risk identification method provided by the specification is adopted, after the target risk type required to be identified for the target user is determined, determining a target rule in preset risk rules, and determining available data in service data of the user according to the target rule; and constructing prompt information according to the target risk type, the target rule and the available data, and inputting the prompt information into the generative model to obtain the risk description of the target user output by the generative model. By adopting the method, the risk rule can be determined according to expert experience, effective available data can be screened out from a plurality of business data by combining the target risk types to be identified, prompt information is constructed according to the target risk types, the target rule and the available data, and the generated model is guided to output accurate and reliable risk description with controllable logic.

Drawings

The accompanying drawings, which are included to provide a further understanding of the specification, illustrate and explain the exemplary embodiments of the present specification and their description, are not intended to limit the specification unduly. Attached at

In the figure:

fig. 1 is a schematic flow chart of a risk identification method provided in the present specification;

FIG. 2 is a schematic diagram of a logic diagram provided in the present specification;

FIG. 3 is a schematic diagram of a risk identification apparatus provided in the present specification;

fig. 4 is a schematic view of an electronic device corresponding to fig. 1 provided in the present specification.

Detailed Description

At present, in the field of risk prevention and control, AI is used to analyze and describe risks possibly existing in users, which is not a very mature technology. The main reason for this is that any generative model built based on AI, such as a large language model, has too strong randomness of the output content, which results in uncontrollable output. In general, when data of the same user is input to the AI multiple times, the content of each output of the AI may be different, or even very different. The risk actually existing in the user is difficult to judge by the wind control personnel through the content output by the AI.

In order to solve the above problems, the present specification provides a risk identification method capable of guiding content that AI output is more accurate and logically controllable.

For the purposes of making the objects, technical solutions and advantages of the present specification more apparent, the technical solutions of the present specification will be clearly and completely described below with reference to specific embodiments of the present specification and corresponding drawings. It will be apparent that the described embodiments are only some, but not all, of the embodiments of the present specification. All other embodiments, which can be made by one of ordinary skill in the art without undue burden from the present disclosure, are intended to be within the scope of the present disclosure.

The following describes in detail the technical solutions provided by the embodiments of the present specification with reference to the accompanying drawings.

Fig. 1 is a schematic flow chart of a risk identification method provided in the present specification, including the following steps:

s100: and acquiring service data of a target user under a target service, and determining a target risk type required to be identified for the target user.

In this specification, an execution body for implementing a risk identification method may refer to a designated device such as a server provided on a service platform, and for convenience of description, only the execution body is taken as the server in this specification to describe a risk identification method provided in this specification.

The risk identification method provided by the specification aims to enable the AI to output accurate and logically controllable risk descriptions for users, so that possible risks of the users are judged. Based on this, in this step, it is necessary to first acquire service data of the target user under the target service. The target user is a user needing risk identification, the target service can be any service with risk, and the service data is various data generated when the target user executes the target service. The traffic data may also vary depending on the targeted traffic. In general, data generated when a target user executes a target service for a fixed period of time may be acquired as service data, and the end time of the fixed period of time may be the time of executing the method, and the time length may be, for example, two weeks, one month, etc., which is not particularly limited in this specification.

And the risk type which needs to be identified for the target user can be determined while the service data are acquired. Unlike general risk identification, when a large language model constructed based on AI is used to output a risk description of a target user, it is generally implemented in the form of a question-answer. That is, a question or instruction is input to the large language model, and an answer is made by the large language model. Based on this, the risk type, i.e., the target risk type, that requires a large language model to analyze the target user needs to be determined for use in subsequent steps.

There are a variety of ways to determine the target risk type, and this specification provides a specific embodiment for reference. Generally, the corresponding information may be preset or input by an air controller to acquire desired contents later. Specifically, in response to receiving query information, a target risk type included in the query information may be determined. The query information is preset or input by a person responsible for risk prevention and control, and various forms of the query information can exist as long as the type of the target risk to be identified can be reflected, and the specification does not limit the type of the target risk. For example, the query information may be "whether the target user is at risk of type a and type B".

S102: and determining a risk rule for identifying the target risk type from preset risk rules according to the target risk type, and taking the risk rule as a target rule.

After the target risk type that needs to be identified for the target user is obtained in step S100, in this step, a risk rule for identifying the target risk type by the user may be further determined based on the target risk type, as the target rule.

The risk rule is preset and is used for identifying whether the target user has a risk of a corresponding risk type based on the service data of the target user. Taking the target service as a transaction service as an example, assuming that one risk rule pointing to the risk type A is 'the maximum transaction amount of the target user in the early morning is greater than a specified value', when the service data of the target user meets the risk rule, the target user can be considered to have the risk corresponding to the risk type A. It should be noted that one risk type may correspond to one or more different risk rules, and that one risk rule may also point to one or more different risk types.

In this step, based on the determined target risk type, all risk rules corresponding to the target risk type may be determined as target rules for use in subsequent steps.

S104: and screening the data matched with the target rule from the service data as available data.

After the target rule is determined in step S102, data matching the target rule may be screened out from the service data as available data in this step. In general, each target rule may determine an available data, where the determined available data matches a target risk type to which the target rule points.

For example, still in the embodiment of step S102, when the target rule is "the maximum transaction amount of the target user in the early morning is greater than the specified value", then all the transactions occurring in the early morning need to be searched in the service data of the target user, and the maximum transaction amount of the transactions is determined and compared with the specified value. When the maximum transaction amount is greater than the specified value, this business data may be considered to satisfy the risk rule, which may be determined as available data.

More preferably, in order to make the above process of determining the target rule and determining the available data more intuitive, efficient and convenient to implement, the present specification additionally provides a manner of concatenating risk types, risk rules and business data in a manner of a knowledge graph. When determining a target rule and available data, specifically, inputting the business data into a pre-created map frame to obtain a logic map of the target user; recalling a risk path matched with the target risk type in the logic map, wherein the risk path is used as a target risk path and consists of a risk type, a risk rule for obtaining the risk type and service data meeting the risk rule; determining a risk rule contained in the target risk path as a target rule; and determining service data contained in the target risk path as available data.

When the map frame is created in advance, specifically, each risk type existing in the target service and a risk rule for obtaining each risk type can be determined according to expert experience under the target service obtained from historical information; and taking each risk type as a first node, obtaining a risk rule of each risk type as a second node, taking the data frame as a third node, and constructing a map frame, wherein each second node is connected with one first node and one third node to serve as a risk path.

Fig. 2 is a schematic diagram of one possible logic diagram provided in the present specification. As shown in fig. 2, the logic map of the target user includes a first node represented by a square, a second node represented by a diamond, and a third node represented by a circle. The first node is used for representing risk types, the second node is used for representing risk rules, and the third node is used for representing service data. Each second node is respectively connected with one first node and one third node to form a risk path; the first node and the third node are not connected.

When the map frame is created, the risk types and the corresponding risk rules can be obtained according to expert experience summarized when the target service is executed historically; when the method is executed, the business data of the user can be filled into the data frame in the map frame after being obtained, so as to form a logic map. In other words, only the data frame exists in the pre-created map frame, and no specific business data exists. After the business data is populated into the data framework, the graph framework is referred to as a logical graph. Therefore, for the same target service, the same spectrum frame can be adopted when the risks of different target users are identified, but the logic spectrum finally obtained is different due to different service data of different users.

When the business data is filled into the data frame, the business data can be filled into the data frame only when the business data can be connected with the risk rule of the data frame. For any risk rule in the map frame, when the business data which can meet the risk rule does not exist in the business data of the target user, then the data frame corresponding to the risk rule does not exist any data. Further, when the final logic map is obtained, a third node without any data, namely the data frame, can be deleted, and a second node connected with the third node is deleted, so that effective risk paths in the logic map are ensured. Preferably, when one business data of the target user can simultaneously meet a plurality of different risk rules, the data frames connected with the met different risk rules can be combined into one data frame.

The target rule and the available data are determined by taking fig. 2 as an example. In the logic diagram shown in fig. 2, a first node represented by a square represents different risk types A, B, C … …, a second node represented by a diamond represents different risk rules v1, v2, v3 … …, and a third node represented by a circle represents different business data f1, f2, f3 … …. Any path from the third node to the second node to the first node can be considered as a risk path, such as "f1→v1→a", "f6→v7→b" in fig. 2. Each risk path only contains one risk type, one risk rule and one service data, so after the target risk type is determined, all risk paths containing the target risk type can be determined as target risk paths, the risk rule contained in the target risk paths is determined as target rule, and the service data contained in the target risk paths is determined as available data. For example, assuming that the target risk type is B, the target risk paths are "f2→v3→b", "f3→v4→b", "f6→v7→b" may be determined in the logic diagram shown in fig. 2, and the target rule may be v3, v4, v7, and the available data may be f2, f3, and f6.

Additionally, when a mode of determining the target risk type according to the received query information is adopted, the point to be considered is that the query information is manually input text information and no standard format exists, so that the received query information does not necessarily point to a certain risk type accurately. At this time, one step of matching can be additionally added, so that the selected target risk path is more accurate and reliable. In particular, a risk path including the target risk type may be determined in the logical map as a candidate risk path; respectively extracting query text features of the query information and path text features of each candidate risk path; for each candidate risk path, determining the similarity between the path text characteristics of the candidate risk path and the query text characteristics; and determining the candidate risk paths with the similarity not smaller than a specified threshold as target risk paths.

Because the target risk type determined according to the query information is not necessarily completely trusted, the risk path found according to the target risk type may be initially determined as a candidate risk path. And respectively extracting the query text features of the query information and the path text features of each candidate risk path, determining the similarity between each path text feature and the query text feature, and finally determining the candidate risk path with higher similarity, namely not smaller than a specified threshold value, as the target risk path. There may be various ways of determining the similarity, such as a distance between two feature vectors, and the like, which is not particularly limited in this specification. Meanwhile, the number of the selected candidate risk paths and the target risk paths can be limited according to requirements.

S106: and constructing prompt information according to the target risk type, the target rule and the available data.

According to the target risk type, the target rule and the available data determined in the previous step, prompt information can be constructed in the step. The prompt information is used for being input into a generated model constructed based on the AI in the subsequent steps to guide the AI to output more accurately. There may be various ways to construct the hint information, for example, the target risk type, target rules, and available data itself may be directly used as hint information, as the simplest. The present specification provides one possible embodiment for reference. In particular, keywords related to the target risk type may be extracted from the target rule and the available data; and constructing prompt information according to the target risk type and the keywords.

In general, the available data may be represented as, for example, "maximum transaction amount: form of X "; the target rule may be expressed, for example, in a form that the maximum transaction amount is greater than Y, at this time, the available data and the keywords in the target rule may be extracted and integrated to obtain the keywords such as "maximum transaction amount", "X", "Y", "greater than" and the like, and a prompt message may be constructed in combination with the corresponding target risk types. The form of the prompt information may be various, for example, keywords and target risk types may be directly integrated together according to a specific sequence to obtain { "risk type a", "maximum transaction amount", "X", "Y", "greater than" }, which is not particularly limited in this specification.

S108: and inputting the prompt information into a pre-trained generation type model to obtain risk description of the target user, which is output by the generation type model.

In this step, the prompt information determined in step S106 may be input into a pre-trained generation model, so as to obtain a risk description of the target user output by the generation model, and complete risk identification of the target user. The generative model may be a model with natural language processing capability, such as a large language model.

When the generated model is trained in advance, a sample wind control corpus can be obtained; masking the sample wind control corpus to obtain a wind control corpus with disturbance; inputting the wind-controlled corpus with disturbance into a to-be-trained generation model to obtain the corpus to be optimized output by the generation model; and training the generated model according to the difference between the corpus to be optimized and the sample wind control corpus. The generated model is trained by adopting the special sample wind control corpus in the wind control field, so that the generated model can learn the expertise in the wind control field, and the understanding capability of the generated model in the risk identification process is improved.

In addition, the answer capabilities of the generative model may be trained even further. Specifically, a sample instruction and a labeling answer corresponding to the sample instruction can be obtained; inputting the sample instruction into the generated model to obtain an answer to be optimized output by the generated model; and adjusting parameters of the generated model according to the difference between the answer to be optimized and the labeling answer. The generated model is further adjusted through the preset questions and answers related to the wind control field, so that the output capability of the generated model on the content related to the wind control field is greatly improved on the basis of understanding the professional knowledge of the wind control field.

In the traditional method, after the generated model is trained, service data of the target user is directly input into the generated model, and output of the generated model to the target user is obtained. As mentioned in the background art, the description of AI generation in the conventional method is not stable due to the characteristic of the self-output of the generative model, and the correctness of the content and logic thereof cannot be ensured. The reason is that the generative model needs to process too much service data, which leads to the defect that the random output characteristic is amplified to be unstable. In the method, through combining the risk rule obtained by expert experience, the prompt information is constructed by adopting the service data of the target user, and then the simple and effective prompt information is input into the generating model, so that accurate and controllable output is obtained.

The foregoing describes one or more methods for implementing risk identification in the present specification, and based on the same ideas, the present specification further provides a corresponding risk identification device, as shown in fig. 3.

Fig. 3 is a schematic diagram of a risk identification device provided in the present specification, including:

the acquiring module 200 is configured to acquire service data of a target user under a target service, and determine a target risk type to be identified for the target user;

a determining module 202, configured to determine, according to the target risk type, a risk rule for identifying the target risk type from preset risk rules, as a target rule;

a matching module 204, configured to filter, from the service data, data that matches the target rule as available data;

a construction module 206, configured to construct prompt information according to the target risk type, the target rule, and the available data;

and the output module 208 is configured to input the prompt information into a pre-trained generation model, and obtain a risk description of the target user output by the generation model.

Optionally, the determining module 202 is specifically configured to input the service data into a pre-created spectrum frame to obtain a logic spectrum of the target user; recalling a risk path matched with the target risk type in the logic map, wherein the risk path is used as a target risk path and consists of a risk type, a risk rule for obtaining the risk type and service data meeting the risk rule; determining a risk rule contained in the target risk path as a target rule;

The matching module 204 is specifically configured to determine service data included in the target risk path as available data.

Optionally, the apparatus further includes a preset module 210, specifically configured to determine, according to expert experience under the target service obtained from the history information, each risk type existing in the target service and a risk rule for obtaining each risk type; and taking each risk type as a first node, obtaining a risk rule of each risk type as a second node, taking the data frame as a third node, and constructing a map frame, wherein each second node is connected with one first node and one third node to serve as a risk path.

Optionally, the obtaining module 200 is specifically configured to determine, in response to receiving query information, a target risk type included in the query information;

the determining module 202 is specifically configured to determine, in the logic atlas, a risk path including the target risk type as a candidate risk path; respectively extracting query text features of the query information and path text features of each candidate risk path; for each candidate risk path, determining the similarity between the path text characteristics of the candidate risk path and the query text characteristics; and determining the candidate risk paths with the similarity not smaller than a specified threshold as target risk paths.

Optionally, the constructing module 206 is specifically configured to extract, from the target rule and the available data, a keyword related to the target risk type; and constructing prompt information according to the target risk type and the keywords.

Optionally, the device further includes a training module 212, specifically configured to obtain a sample wind-controlled corpus; masking the sample wind control corpus to obtain a wind control corpus with disturbance; inputting the wind-controlled corpus with disturbance into a to-be-trained generation model to obtain the corpus to be optimized output by the generation model; and training the generated model according to the difference between the corpus to be optimized and the sample wind control corpus.

Optionally, the training module 212 is further configured to obtain a sample instruction and a labeling answer corresponding to the sample instruction; inputting the sample instruction into the generated model to obtain an answer to be optimized output by the generated model; and adjusting parameters of the generated model according to the difference between the answer to be optimized and the labeling answer.

The present specification also provides a computer readable storage medium storing a computer program operable to perform a risk identification method as provided in fig. 1 above.

The present specification also provides a schematic structural diagram of an electronic device corresponding to fig. 1 shown in fig. 4. At the hardware level, the electronic device includes a processor, an internal bus, a network interface, a memory, and a non-volatile storage, as described in fig. 4, although other hardware required by other services may be included. The processor reads the corresponding computer program from the non-volatile memory into the memory and then runs to implement the risk identification method described above with respect to fig. 1. Of course, other implementations, such as logic devices or combinations of hardware and software, are not excluded from the present description, that is, the execution subject of the following processing flows is not limited to each logic unit, but may be hardware or logic devices.

In the 90 s of the 20 th century, improvements to one technology could clearly be distinguished as improvements in hardware (e.g., improvements to circuit structures such as diodes, transistors, switches, etc.) or software (improvements to the process flow). However, with the development of technology, many improvements of the current method flows can be regarded as direct improvements of hardware circuit structures. Designers almost always obtain corresponding hardware circuit structures by programming improved method flows into hardware circuits. Therefore, an improvement of a method flow cannot be said to be realized by a hardware entity module. For example, a programmable logic device (Programmable Logic Device, PLD) (e.g., field programmable gate array (Field Programmable Gate Array, FPGA)) is an integrated circuit whose logic function is determined by the programming of the device by a user. A designer programs to "integrate" a digital system onto a PLD without requiring the chip manufacturer to design and fabricate application-specific integrated circuit chips. Moreover, nowadays, instead of manually manufacturing integrated circuit chips, such programming is mostly implemented by using "logic compiler" software, which is similar to the software compiler used in program development and writing, and the original code before the compiling is also written in a specific programming language, which is called hardware description language (Hardware Description Language, HDL), but not just one of the hdds, but a plurality of kinds, such as ABEL (Advanced Boolean Expression Language), AHDL (Altera Hardware Description Language), confluence, CUPL (Cornell University Programming Language), HDCal, JHDL (Java Hardware Description Language), lava, lola, myHDL, PALASM, RHDL (Ruby Hardware Description Language), etc., VHDL (Very-High-Speed Integrated Circuit Hardware Description Language) and Verilog are currently most commonly used. It will also be apparent to those skilled in the art that a hardware circuit implementing the logic method flow can be readily obtained by merely slightly programming the method flow into an integrated circuit using several of the hardware description languages described above.

The controller may be implemented in any suitable manner, for example, the controller may take the form of, for example, a microprocessor or processor and a computer readable medium storing computer readable program code (e.g., software or firmware) executable by the (micro) processor, logic gates, switches, application specific integrated circuits (Application Specific Integrated Circuit, ASIC), programmable logic controllers, and embedded microcontrollers, examples of which include, but are not limited to, the following microcontrollers: ARC 625D, atmel AT91SAM, microchip PIC18F26K20, and Silicone Labs C8051F320, the memory controller may also be implemented as part of the control logic of the memory. Those skilled in the art will also appreciate that, in addition to implementing the controller in a pure computer readable program code, it is well possible to implement the same functionality by logically programming the method steps such that the controller is in the form of logic gates, switches, application specific integrated circuits, programmable logic controllers, embedded microcontrollers, etc. Such a controller may thus be regarded as a kind of hardware component, and means for performing various functions included therein may also be regarded as structures within the hardware component. Or even means for achieving the various functions may be regarded as either software modules implementing the methods or structures within hardware components.

The system, apparatus, module or unit set forth in the above embodiments may be implemented in particular by a computer chip or entity, or by a product having a certain function. One typical implementation is a computer. In particular, the computer may be, for example, a personal computer, a laptop computer, a cellular telephone, a camera phone, a smart phone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or a combination of any of these devices.

For convenience of description, the above devices are described as being functionally divided into various units, respectively. Of course, the functions of each element may be implemented in one or more software and/or hardware elements when implemented in the present specification.

It will be appreciated by those skilled in the art that embodiments of the present description may be provided as a method, system, or computer program product. Accordingly, the present specification may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present description can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein.

The present description is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems) and computer program products according to embodiments of the specification. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

In one typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.

The memory may include volatile memory in a computer-readable medium, random Access Memory (RAM) and/or nonvolatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM). Memory is an example of computer-readable media.

Computer readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of storage media for a computer include, but are not limited to, phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium, which can be used to store information that can be accessed by a computing device. Computer-readable media, as defined herein, does not include transitory computer-readable media (transmission media), such as modulated data signals and carrier waves.

It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus 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 apparatus. 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 apparatus that comprises the element.

It will be appreciated by those skilled in the art that embodiments of the present description may be provided as a method, system, or computer program product. Accordingly, the present specification may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present description can take the form of a computer program product on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) having computer-usable program code embodied therein.

The description may be described in the general context of computer-executable instructions, such as program modules, being executed by a computer. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform particular tasks or implement particular abstract data types. The specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices that are linked through a communications network. In a distributed computing environment, program modules may be located in both local and remote computer storage media including memory storage devices.

In this specification, each embodiment is described in a progressive manner, and identical and similar parts of each embodiment are all referred to each other, and each embodiment mainly describes differences from other embodiments. In particular, for system embodiments, since they are substantially similar to method embodiments, the description is relatively simple, as relevant to see a section of the description of method embodiments.

The foregoing is merely exemplary of the present disclosure and is not intended to limit the disclosure. Various modifications and alterations to this specification will become apparent to those skilled in the art. Any modifications, equivalent substitutions, improvements, or the like, which are within the spirit and principles of the present description, are intended to be included within the scope of the claims of the present description.

Claims

1. A risk identification method, comprising:

2. The method according to claim 1, according to the target risk type, determining, as a target rule, a risk rule for identifying the target risk type from preset risk rules, specifically including:

determining a risk rule contained in the target risk path as a target rule;

3. The method of claim 2, wherein creating a map frame in advance comprises:

4. The method of claim 2, determining a target risk type to be identified for the target user, comprising in particular:

5. The method of claim 1, constructing a hint message according to the target risk type, the target rule, and the available data, specifically comprising:

6. The method of claim 1, the pre-training generative model, in particular comprising:

acquiring a sample wind control corpus;

7. The method of claim 6, the method further comprising:

8. A risk identification device comprising:

9. The apparatus of claim 8, wherein the determining module is specifically configured to input the service data into a pre-created spectrum frame to obtain a logic spectrum of the target user; recalling a risk path matched with the target risk type in the logic map, wherein the risk path is used as a target risk path and consists of a risk type, a risk rule for obtaining the risk type and service data meeting the risk rule; determining a risk rule contained in the target risk path as a target rule;

10. The apparatus of claim 9, further comprising a preset module, specifically configured to determine each risk type existing in the target service and a risk rule for obtaining each risk type according to expert experience under the target service obtained from history information; and taking each risk type as a first node, obtaining a risk rule of each risk type as a second node, taking the data frame as a third node, and constructing a map frame, wherein each second node is connected with one first node and one third node to serve as a risk path.

11. The apparatus of claim 9, wherein the obtaining module is specifically configured to determine a target risk type included in query information in response to receiving the query information;

12. The apparatus of claim 8, the construction module being configured to extract keywords related to the target risk type from the target rule and the available data; and constructing prompt information according to the target risk type and the keywords.

13. The device of claim 8, further comprising a training module, in particular for obtaining a sample wind-controlled corpus; masking the sample wind control corpus to obtain a wind control corpus with disturbance; inputting the wind-controlled corpus with disturbance into a to-be-trained generation model to obtain the corpus to be optimized output by the generation model; and training the generated model according to the difference between the corpus to be optimized and the sample wind control corpus.

14. The apparatus of claim 13, the training module further to obtain a sample instruction and a labeling answer corresponding to the sample instruction; inputting the sample instruction into the generated model to obtain an answer to be optimized output by the generated model; and adjusting parameters of the generated model according to the difference between the answer to be optimized and the labeling answer.

15. A computer readable storage medium storing a computer program which, when executed by a processor, implements the method of any of the preceding claims 1-7.

16. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the method of any of the preceding claims 1-7 when the program is executed.