Risk processing method, system and equipment of service order and medium

Abstract

The invention relates to the field of risk processing, in particular to a risk processing method, system and device for a service order and a medium, and the method comprises the steps: obtaining the service order information of a target service order; based on a large language model, extracting risk entity types and corresponding risk features from the service order information, and association degrees between the risk features corresponding to different risk entity types, and constructing a knowledge graph according to an output result of the large model; performing order risk prediction on the knowledge graph by using a pre-trained order risk identification model to obtain a risk type of the target service order and a risk probability corresponding to the risk type; and carrying out risk prompting on the target service order according to the risk type of the target service order and the risk probability corresponding to the risk type. According to the method, the risk identification efficiency is improved, and the risk identification flexibility is enhanced.

Description

Technical Field

[0001] The present invention relates to the field of risk processing, and particularly to a method, system, device and medium for risk processing of service orders. Background Art

[0002] Service orders, such as online passenger transportation service orders or online freight transportation service orders, are an important part of the modern Internet economy, where the order provider (such as a driver) and the order recipient (such as a user) achieve service transactions through a network platform. Risk identification and processing of service orders are crucial for ensuring service security and reliability, which can help prevent and reduce fraud, violations and other potential risks.

[0003] In current practice, service order risk identification mainly relies on manual and preset rules. Specifically, analysts need to manually collect and collate a large amount of order data, summarize risk judgment rules from this order data, and use these rules to identify and process risks for new service orders. However, this method has obvious shortcomings: First, the process of manually collecting and processing data is both time-consuming and laborious, especially when the order volume is large, this problem is particularly prominent; Second, preset rules often fail to cover all risk scenarios. When facing complex and variable risk identification scenarios, the existing rules and systems need to be repeatedly deliberated and modified, which undoubtedly increases the complexity of risk identification and reduces its flexibility. Summary of the Invention

[0004] To solve the above problems, the present invention provides a method, system, device and medium for risk processing of service orders.

[0005] The first aspect of the present invention discloses a method for risk processing of service orders, the method comprising:

[0006] Obtaining service order information of a target service order;

[0007] Based on a large language model that has completed context learning in advance, extracting risk entity types and their corresponding risk features from the service order information, as well as the association degrees between risk features corresponding to different risk entity types, and constructing a knowledge graph according to the output result of the large model;

[0008] Using a pre-trained order risk identification model to perform order risk prediction on the knowledge graph, obtaining the risk type of the target service order and the risk probability corresponding to this risk type;

[0009] Performing risk prompt on the target service order according to the risk type of the target service order and the risk probability corresponding to this risk type.

[0010] Further, the step of obtaining the service order information of the target service order includes:

[0011] Obtain the target service order;

[0012] Extract the order basic information from the target service order; wherein, the order basic information includes the order provider, the order recipient, and the order content;

[0013] Retrieve the order provider characteristics according to the order provider;

[0014] Retrieve the order recipient characteristics according to the order recipient;

[0015] Retrieve the real-time environment information and the order basic information according to the order content.

[0016] Further, based on a large language model that has pre-completed context learning, the steps of extracting the risk entity types and their corresponding risk characteristics, and the correlation degrees between the risk characteristics corresponding to different risk entity types from the service order information, and constructing a knowledge graph according to the output result of the large model include:

[0017] Take the order provider, the order recipient, and the order as the risk entity types;

[0018] Based on the large language model, according to the order provider characteristics, the order recipient characteristics, the order basic information, and the real-time environment information, extract the risk characteristics of the order provider related to risks, the risk characteristics of the order payer, and the risk characteristics of the order, and calculate the correlation degrees between the risk characteristics of different risk entity types;

[0019] Take the risk entity types as the first-level nodes, take the order payer risk characteristics, the order provider risk characteristics, and the order risk characteristics as the second-level nodes, and take the correlation degrees as undirected edges to construct a knowledge graph.

[0020] Further, the step of obtaining the service order information of the target service order includes:

[0021] Judge the status of the target service order:

[0022] When the target service order is in an unprocessed state, obtain the service order information according to the target service order;

[0023] When the target service order is in a processing state, obtain the service order information according to the target service order, and also obtain the communication data between the order provider and the order recipient corresponding to the target service order.

[0024] Further, based on a large language model that has completed context learning in advance, extracting the risk entity types and their corresponding risk characteristics from the service order information, as well as the correlation degrees between the risk characteristics corresponding to different risk entity types, the steps of constructing a knowledge graph according to the output results of the large model include:

[0025] Regarding the order provider, the order recipient, and the order as the risk entity types;

[0026] Based on the large language model, according to the order provider characteristics, the order recipient characteristics, the order basic information, the real-time environment information, and the communication data, extracting the risk characteristics of the order provider related to risks, the risk characteristics of the order payer, and the risk characteristics of the order, and calculating the correlation degrees between the risk characteristics of different risk entity types;

[0027] Regarding the risk entity types as the first-level nodes, the order payer risk characteristics, the order provider risk characteristics, and the order risk characteristics as the second-level nodes, and using the correlation degrees as undirected edges to construct a knowledge graph.

[0028] Further, the large language model is obtained through context learning with multiple pre-input examples; wherein, each example includes service order information and its corresponding communication data, as well as the risk entity types, risk characteristics, and the correlation degrees between the risk characteristics of different risk entity types annotated for the service order.

[0029] Further, the steps of giving a risk prompt for the target service order according to the risk type of the target service order and the risk probability corresponding to the risk type include:

[0030] Judging the status of the target service order:

[0031] When the target service order is in an unprocessed state, judging whether the risk probability exceeds the first threshold corresponding to the risk type, and if it exceeds, sending a first warning notice to the order provider;

[0032] When the target service order is in a processing state, judging whether the risk probability exceeds the second threshold corresponding to the risk type, and if it exceeds, sending a second warning message to the order provider and the order recipient.

[0033] A second aspect of the present invention discloses a risk processing system for service orders, and the system includes:

[0034] An acquisition module, used to acquire the service order information of the target service order;

[0035] A building module is used to extract risk entity types and their corresponding risk features, as well as the correlation degrees between risk features corresponding to different risk entity types, from the service order information based on a large language model that has completed context learning in advance, and construct a knowledge graph according to the output result of the large model;

[0036] A prediction module is used to perform order risk prediction on the knowledge graph using a pre-trained order risk identification model, and obtain the risk type of the target service order and the risk probability corresponding to this risk type;

[0037] A prompt module is used to give risk prompts for the target service order according to the risk type of the target service order and the risk probability corresponding to this risk type.

[0038] A third aspect of the present invention discloses an electronic device, which includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of any one of the service order risk processing methods disclosed in the first aspect of the present invention.

[0039] A fourth aspect of the present invention discloses a storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the steps of any one of the service order risk processing methods disclosed in the first aspect of the present invention.

[0040] The present invention adopts a pre-constructed undirected knowledge graph and a pre-trained risk identification model. By automatically extracting risk features from a large amount of order data, the overhead of manual data collection and processing is greatly reduced, thus significantly improving the efficiency of risk identification; at the same time, since the undirected knowledge graph can accurately depict the risk correlation degrees between various features, and the pre-trained risk identification model can adapt to various complex and changeable risk identification scenarios, the flexibility of risk identification is significantly improved. Therefore, the method of the present invention improves the efficiency of risk identification while enhancing the flexibility of risk identification. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for description in the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0042] Figure 1 is a flowchart of a service order risk processing method disclosed in an embodiment of the present invention;

[0043] Figure 2 is a schematic diagram of a knowledge graph disclosed in an embodiment of the present invention;

[0044] Figure 3 It is a schematic structural diagram of a risk processing system for service orders disclosed in an embodiment of the present invention;

[0045] Figure 4 It is a schematic structural diagram of an electronic device disclosed in an embodiment of the present invention. Detailed implementation manners

[0046] In order to enable those skilled in the art to better understand the solution of the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0047] The terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, device, or product end that includes a series of steps or units is not limited to the listed steps or units, but optionally further includes steps or units not listed, or optionally further includes other steps or units inherent to these processes, methods, devices, or product ends.

[0048] Referring to "embodiments" herein means that the specific features, structures, or characteristics described in connection with the embodiments may be included in at least one embodiment of the present invention. The phrase appears in various places in the specification and does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art will explicitly and implicitly understand that the embodiments described herein may be combined with other embodiments.

[0049] Please refer to Figure 1 as shown in Figure 1 It is a schematic flowchart of a risk processing method for service orders disclosed in an embodiment of the present invention. As Figure 1 shown, the risk processing method for service orders may include the following operations:

[0050] S101. Obtain the service order information of the target service order;

[0051] In this optional embodiment, the service order information may include: order basic information, order provider characteristics, order recipient characteristics, order basic information, and real-time environment information. This embodiment explains the present invention by taking an online passenger transportation service order or an online freight transportation service order as an example. An online passenger transportation service order refers to an order generated through an Internet platform (such as a ride-hailing platform, a carpooling platform, etc.) with transporting passengers as the core service content. For example, a passenger places an order through a ride-hailing platform to reserve a vehicle for a passenger-carrying service from place A to place B. The platform matches a driver to receive the order and complete the transportation, and the passenger pays the fee. An online freight transportation service order refers to an order generated through an Internet platform with transporting goods as the core service content. The shipper publishes the goods transportation demand (such as furniture moving, bulk commodity transportation) through the platform. The platform matches a truck driver or a logistics company (service provider) to receive the order and complete the transportation, and the shipper pays the freight.

[0052] The service order corresponds to the order provider and the order recipient. The order provider refers to the entity that actually provides the service in the service order, that is, the party that undertakes the order and fulfills the service through the platform. In this embodiment, the order provider is the driver; the order recipient refers to the entity that initiates the demand and receives the service in the service order, that is, the party that places an order and enjoys the service through the platform. In this embodiment, the order recipient is the passenger or the shipper.

[0053] In an optional embodiment, the steps of obtaining the service order information of the target service order include:

[0054] Obtain the target service order;

[0055] Extract the order basic information from the target service order; wherein, the order basic information includes the order provider, the order recipient, and the order content;

[0056] Retrieve the order provider characteristics according to the order provider;

[0057] Retrieve the order recipient characteristics according to the order recipient;

[0058] Retrieve the real-time environment information and the order basic information according to the order content.

[0059] In this optional embodiment, the order provider refers to the unique identifier of the service provider, the order recipient refers to the unique identifier of the service demander, and the order content refers to the service details and execution rules agreed upon in the order. Taking an online passenger transportation service order or an online freight transportation service order as an example, the order content may include: departure place, destination, number of passengers, reservation time, whether a child seat is required, type of goods (such as furniture, fresh food), weight / volume, loading and unloading requirements (such as need for handling), insured amount, etc.

[0060] The characteristics of the order provider are represented based on the attribute data related to the service provider extracted by the order provider from the database, and may include: driving experience, service rating, historical cancellation rate, order acceptance response time, vehicle load capacity, cargo damage complaint rate, frequently traveled route area, lateness rate, complaint rate, overtime rate, etc.

[0061] The characteristics of the order recipient are represented based on the attribute data related to the service demander extracted by the order recipient from the database, and may include: historical order frequency, average payment method, complaint rate, average historical order amount, shipment frequency, cargo type preference, lateness rate, etc.

[0062] The basic order information represents the necessary data relied on during the order execution process, the service details and execution rules agreed upon in the order. In this embodiment, the basic order information includes the order content and the geographical path and transportation planning data involved in the order execution process, such as: starting point coordinates, ending point coordinates, driving route, congested sections along the way, estimated driving distance, estimated driving duration, loading location coordinates, unloading location coordinates, highway / national road preference, avoidance of prohibited areas, etc.

[0063] The real-time environment information represents the status data of the external environment during the order execution, and may include: real-time weather, real-time traffic conditions, time factors, etc. Among them, the real-time weather may include heavy rain, high temperature, haze; the real-time traffic conditions may include: road congestion index, accident warning; the time factors may include: whether it is a holiday, peak hours in the morning and evening, etc.

[0064] In another optional embodiment, the step of obtaining the service order information of the target service order includes:

[0065] Judge the status of the target service order:

[0066] When the target service order is in the unexecuted state, obtain the service order information according to the target service order;

[0067] When the target service order is in the in-progress state, obtain the service order information according to the target service order, and also obtain the communication data between the order provider and the order recipient corresponding to the target service order.

[0068] In this optional embodiment, the unexecuted state means that the target service order is in the preparatory stage before starting to be executed, the in-progress state means that the target service order is in the stage where it has been accepted and the service is being executed, and the communication data refers to the text format communication records generated during the service interaction between the order provider and the order recipient.

[0069] The communication data may include: text interaction records generated by both parties through the built-in instant messaging tool of the platform, translated text generated by voice recognition after a call using a virtual number, text data converted from the audio collected in real time by in-vehicle devices during the service execution process, etc.

[0070] S102. Based on a large language model that has completed context learning in advance, extract the risk entity types and their corresponding risk features from the service order information, as well as the correlation degrees between the risk features corresponding to different risk entity types, and construct a knowledge graph according to the output result of the large model;

[0071] In an optional embodiment, the risk entity types include: the order provider, the order recipient, and the order; the steps of extracting the risk entity types and their corresponding risk features from the service order information, as well as the correlation degrees between the risk features corresponding to different risk entity types, and constructing a knowledge graph according to the output result of the large model based on a large language model that has completed context learning in advance include:

[0072] Take the order provider, the order recipient, and the order as the risk entity types;

[0073] Based on the large language model, according to the order provider features, the order recipient features, the order basic information, and the real-time environment information, extract the risk features of the order provider related to risks, the risk features of the order payer, and the risk features of the order, and calculate the correlation degrees between the risk features of different risk entity types;

[0074] Take the risk entity types as the first-level nodes, take the order payer risk features, the order provider risk features, and the order risk features as the second-level nodes, and take the correlation degrees as undirected edges to construct a knowledge graph.

[0075] In this optional embodiment, context learning means that the large language model directly uses the implicit rules in the input text to complete specific tasks (such as entity extraction, relationship modeling, etc.) based on the general semantic understanding, logical association, and pattern recognition capabilities accumulated during the pre-training stage.

[0076] It can be seen that in this alternative embodiment, through the context learning ability of the large language model, the unstructured service order information (such as communication data, background features) is automatically transformed into a structured knowledge graph, realizing the dynamic modeling of risk entity types and their associated relationships. Compared with traditional rule engines or manual annotation methods, this solution significantly reduces the dependence on domain annotation data, improves the mining efficiency of complex risk patterns (such as the implicit association between driver behavior and user complaints), and at the same time supports real-time updating of the knowledge graph to cope with the dynamic changes in business scenarios (such as the impact of sudden weather on order fulfillment). Finally, risk reasoning based on the knowledge graph can optimize the platform's risk warning, resource scheduling, and decision-making response capabilities.

[0077] In another alternative embodiment, based on a large language model that has completed context learning in advance, extracting risk entity types and their corresponding risk features, as well as the degree of association between risk features corresponding to different risk entity types from the service order information, the steps of constructing a knowledge graph according to the output result of the large model include:

[0078] Regarding the order provider, the order recipient, and the order as the risk entity types;

[0079] Based on the large language model, according to the order provider features, the order recipient features, the order basic information, the real-time environment information, and the communication data, extract the risk features of the order provider related to risks, the risk features of the order payer, and the risk features of the order, and calculate the degree of association between the risk features of different risk entity types;

[0080] Regarding the risk entity types as the first-level nodes, the order payer risk features, the order provider risk features, and the order risk features as the second-level nodes, and the degree of association as the undirected edges to construct a knowledge graph.

[0081] It can be seen that in this alternative embodiment, by incorporating communication data into the input of the large language model, it is possible to more comprehensively capture the dynamic risk signals in the service interaction process (such as communication conflicts, abnormal route negotiations, or driver behavior deviations), thereby enhancing the mining depth of the knowledge graph for implicit risk associations (such as the causal relationship between user complaints and driver detour behavior), and further improving the real-time performance and accuracy of risk prediction. Especially in complex interpersonal interaction scenarios during service fulfillment (such as dispute prediction, trust assessment), it has a more fine-grained decision-making support ability.

[0082] In yet another alternative embodiment, the large language model is obtained through context learning with multiple pre-input examples; where each example includes service order information and its corresponding communication data, as well as the risk entity types, risk features, and the degree of association between risk features of different risk entity types annotated for the service order.

[0083] For example, an example may include:

[0084] Service order information: Order ID - 20230901; Route information: Planned distance 15 km, real-time congestion level (high); Real-time environmental information: Heavy rain, visibility 200 m; Characteristics of the order provider: Punctuality rate 85%, number of complaints in the last 30 days 2 times, detour frequency 30%; Characteristics of the order recipient: Historical complaint rate 10%, recent frequency of urging orders (3 times / week);

[0085] Communication data: The user sent "Please arrive as soon as possible, in a hurry" 3 times; The driver replied "Heavy rain and slippery road, need to detour 3 km".

[0086] Annotations for the service order: Risk entity types: Driver (order provider), User (order recipient), Order; Risk characteristics: Driver: Low punctuality rate (85%), high complaint rate (2 times / 30 days), high detour frequency (30%); User: High complaint tendency (10%), high urging frequency (3 times / week); Order: High congestion risk (high congestion level), bad weather risk (heavy rain), low visibility risk (200 m); The bad weather risk of the order and the high detour frequency of the driver (correlation degree 0.85), the low punctuality rate of the driver and the high urging frequency of the user (correlation degree 0.95).

[0087] Figure 2 An example of a knowledge graph obtained based on a large language model is shown, and each node in the graph does not represent real characteristics.

[0088] It can be seen that this optional embodiment enables the large language model to quickly understand the risk logic of the business scenario by providing context learning examples with clear annotations, significantly improving the feature extraction accuracy and correlation inference consistency of the model under few-shot conditions. Compared with the solution that only relies on the general knowledge of the model, this method can reduce misjudgments caused by semantic ambiguity, and at the same time enhance the interpretability of the knowledge graph, especially suitable for the standardized modeling of complex risk patterns.

[0089] S103. Use a pre-trained order risk identification model to predict the order risk of the knowledge graph, and obtain the risk type of the target service order and the risk probability corresponding to this risk type;

[0090] In this optional embodiment, the order risk identification model is a graph neural network model.

[0091] S104. According to the risk type of the target service order and the risk probability corresponding to this risk type, give a risk prompt for the target service order.

[0092] In yet another alternative embodiment, the steps of performing risk prompt for the target service order according to the risk type of the target service order and the risk probability corresponding to the risk type include:

[0093] Judge the status of the target service order:

[0094] When the target service order is in the non - processed state, judge whether the risk probability exceeds the first threshold corresponding to the risk type. If it exceeds, send a first warning notice to the order provider;

[0095] When the target service order is in the processing state, judge whether the risk probability exceeds the second threshold corresponding to the risk type. If it exceeds, send a second warning message to the order provider and the order recipient.

[0096] It can be seen that in this alternative embodiment, through a differential early - warning mechanism, the risk judgment thresholds (the first threshold focuses on preventive intervention, and the second threshold focuses on real - time risk control) and the notification objects (single - party or two - way warning) are dynamically adjusted according to the order status, significantly improving the accuracy and timeliness of risk prompt: early warning in the non - processed state can avoid potential risks (such as pre - screening high - risk driver resources), and two - way synchronous warning in the processing state (such as notifying the driver to adjust the route and the user to adjust expectations when sudden congestion causes delays) can not only reduce the interference of redundant information to irrelevant parties, but also improve the fault tolerance rate in complex scenarios through collaborative response, ultimately achieving double optimization of platform risk control and user experience.

[0097] Please refer to Figure 3 as shown Figure 3 is a schematic structural diagram of a risk processing system for service orders disclosed in an embodiment of the present invention. As Figure 3 shown, the system includes:

[0098] An acquisition module 301, configured to acquire service order information of a target service order;

[0099] A construction module 302, configured to extract risk entity types and their corresponding risk features, as well as the correlation degree between risk features corresponding to different risk entity types from the service order information based on a large - language model that has completed context learning in advance, and construct a knowledge graph according to the output result of the large model;

[0100] A prediction module 303, configured to perform order risk prediction on the knowledge graph using a pre - trained order risk identification model to obtain the risk type of the target service order and the risk probability corresponding to the risk type;

[0101] A prompt module 304 is configured to perform a risk prompt on the target service order according to the risk type of the target service order and the risk probability corresponding to the risk type.

[0102] For the specific limitations of the risk processing system for service orders, reference can be made to the limitations of the risk processing method for service orders in the foregoing text, which will not be elaborated herein. Each module in the above-mentioned risk processing system for service orders can be implemented in whole or in part by software, hardware, and their combination. The above-mentioned modules can be embedded in or independent of the processor in the electronic device in a hardware format, or stored in the memory in the electronic device in a software format, so as to facilitate the processor to call the operations corresponding to the above-mentioned modules.

[0103] It should be noted that, in order to highlight the innovative part of the present invention, modules that are not closely related to solving the technical problems proposed by the present invention are not introduced in this embodiment, but this does not mean that there are no other modules in this embodiment.

[0104] As Figure 3 shown, the electronic device 1 provided by the present invention may include a memory 11, a processor 12, and a bus, and may further include a computer program stored in the memory 11 and executable on the processor 12, such as a risk processing program for service orders.

[0105] Among them, the memory 11 includes at least one type of readable storage medium, and the readable storage medium includes flash memory, mobile hard disk, multimedia card, card-type memory (such as: SD or DX memory, etc.), magnetic memory, magnetic disk, optical disk, etc. The memory 11 may be an internal storage unit of the electronic device 1 in some embodiments, such as the mobile hard disk of the electronic device 1. The memory 11 may also be an external storage device of the electronic device 1 in other embodiments, such as a plug-in mobile hard disk, a Smart Media Card (SMC), a Secure Digital (SD) card, a Flash Card, etc. equipped on the electronic device 1. Further, the memory 11 may include both the internal storage unit and the external storage device of the electronic device 1. The memory 11 can be used not only to store application software installed in the electronic device 1 and various types of data, such as the code for risk processing of service orders, but also to temporarily store data that has been output or will be output.

[0106] In some embodiments, the processor 12 may be composed of an integrated circuit. For example, it may be composed of a single packaged integrated circuit, or may be composed of multiple packaged integrated circuits with the same or different functions, including a combination of one or more central processing units (CPUs), microprocessors, digital processing chips, graphics processors, and various control chips, etc. The processor 12 is the control core (Control Unit) of the electronic device 1, connecting various components of the entire electronic device 1 through various interfaces and circuits. By running or executing programs or modules stored in the memory 11 (such as the risk processing program of the service order, etc.), and by invoking the data stored in the memory 11, it performs various functions of the electronic device 1 and processes data.

[0107] The processor 12 executes the operating system of the electronic device 1 and various installed application programs. The processor 12 executes the application programs to implement the steps in the above-mentioned risk processing method for service orders.

[0108] Exemplarily, the computer program may be divided into one or more modules. The one or more modules are stored in the memory 11 and executed by the processor 12 to complete this application. The one or more modules may be a series of computer program instruction segments capable of completing specific functions, and these instruction segments are used to describe the execution process of the computer program in the electronic device 1. For example, the computer program may be divided into an acquisition module 301, a construction module 302, a prediction module 303, and a prompt module 304.

[0109] The above-mentioned integrated unit implemented in the form of software function modules may be stored in a computer-readable storage medium. The storage medium may be non-volatile or volatile. The above-mentioned software function modules are stored in a storage medium, including several instructions to enable a computer device (which may be a personal computer, a computer device, or a network device, etc.) or a processor to execute some functions of the risk processing method for service orders described in various embodiments of this application.

[0110] In summary, a risk handling method, system, device, and medium for service orders disclosed by the present invention adopt a pre-constructed undirected knowledge graph and a pre-trained risk identification model. By automatically extracting risk features from a large amount of order data, the overhead of manual data collection and processing is greatly reduced, thus significantly improving the efficiency of risk identification. At the same time, since the undirected knowledge graph can accurately depict the risk correlation degree between various features, and the pre-trained risk identification model can adapt to various complex and changeable risk identification scenarios, the flexibility of risk identification is significantly improved. Therefore, the present invention effectively overcomes various disadvantages in the prior art and has high industrial utilization value.

[0111] The above embodiments are only used to exemplarily illustrate the principles and effects of the present invention, rather than to limit the present invention. Any person familiar with this technology can modify or change the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or changes completed by those with ordinary knowledge in the technical field without departing from the spirit and technical idea disclosed by the present invention should still be covered by the claims of the present invention.

Claims

1. A risk management method for a service order, characterized in that: The method comprises: Obtain service order information of the target service order; Based on a large language model that has completed context learning in advance, risk entity types and their corresponding risk features, as well as the correlation between risk features corresponding to different risk entity types, are extracted from the service order information, and a knowledge graph is constructed according to the output results of the large model; Use the pre-trained order risk identification model to predict order risks on the knowledge graph to obtain the risk type of the target service order and the risk probability corresponding to the risk type; A risk warning is provided for the target service order according to the risk type of the target service order and the risk probability corresponding to the risk type.

2. A risk management method for a service order according to claim 1, characterized in that: The step of obtaining the service order information of the target service order includes: Obtaining the target service order; Extracting the basic order information from the target service order; wherein the basic order information includes the order provider, the order recipient and the order content; According to the order provider, retrieve the order provider characteristics; According to the order recipient, retrieve the order recipient's characteristics; According to the order content, the real-time environment information and the basic order information are retrieved.

3. A risk management method for a service order according to claim 2, characterized in that: Based on a large language model that has completed context learning in advance, the risk entity types and their corresponding risk features, as well as the correlation between the risk features corresponding to different risk entity types are extracted from the service order information, and the steps of constructing a knowledge graph according to the output results of the large model include: The order provider, the order recipient and the order are used as the risk entity types; Based on the large language model, according to the order provider characteristics, the order recipient characteristics, the order basic information and the real-time environment information, the risk characteristics of the order provider, the order payer and the order related to the risk are extracted, and the correlation between the risk characteristics of different risk entity types is calculated; The risk entity type is used as the first-level node, the risk characteristics of the order payer, the risk characteristics of the order provider and the order risk characteristics are used as the second nodes, and the association degree is used as the undirected edge to construct a knowledge graph.

4. A risk management method for a service order according to claim 2, characterized in that: The step of obtaining service order information of the target service order includes: Determine the status of the target service order: When the target service order is in an unprocessed state, obtaining service order information according to the target service order; When the target service order is in progress, service order information is acquired according to the target service order, and communication data between the order provider and the order recipient corresponding to the target service order is also acquired.

5. A risk management method for a service order according to claim 4, characterized in that: Based on a large language model that has completed context learning in advance, the risk entity types and their corresponding risk features, as well as the correlation between the risk features corresponding to different risk entity types are extracted from the service order information, and the steps of constructing a knowledge graph according to the output results of the large model include: The order provider, the order recipient and the order are used as the risk entity types; Based on the large language model, according to the order provider characteristics, the order recipient characteristics, the order basic information, the real-time environment information and the communication data, the risk characteristics of the order provider, the risk characteristics of the order payer and the risk characteristics of the order related to the risk are extracted, and the correlation between the risk characteristics of different risk entity types is calculated; The risk entity type is used as the first-level node, the risk characteristics of the order payer, the risk characteristics of the order provider and the order risk characteristics are used as the second nodes, and the association degree is used as the undirected edge to construct a knowledge graph.

6. A risk management method for a service order according to claim 1, characterized in that: The large language model is obtained by contextual learning based on multiple pre-input examples; wherein each example includes service order information and its corresponding communication data, as well as the risk entity type and risk characteristics annotated on the service order, and the correlation between the risk characteristics of different risk entity types.

7. A risk management method for a service order according to claim 1, characterized in that: The step of providing risk warning to the target service order according to the risk type of the target service order and the risk probability corresponding to the risk type includes: Determine the status of the target service order: When the target service order is in an unprocessed state, determining whether the risk probability exceeds a first threshold corresponding to the risk type, and if so, issuing a first warning notification to the order provider; When the target service order is in progress, it is determined whether the risk probability exceeds a second threshold corresponding to the risk type. If so, a second warning message is issued to the order provider and the order recipient.

8. A risk management system for service orders, characterized in that: The system comprises: An acquisition module, used to acquire service order information of a target service order; A construction module is used to extract risk entity types and their corresponding risk features, as well as the correlation between risk features corresponding to different risk entity types from the service order information based on a large language model that has completed context learning in advance, and to construct a knowledge graph according to the output result of the large model; A prediction module, used to use a pre-trained order risk identification model to perform order risk prediction on the knowledge graph to obtain the risk type of the target service order and the risk probability corresponding to the risk type; The prompt module is used to provide risk prompts for the target service order according to the risk type of the target service order and the risk probability corresponding to the risk type.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the risk handling method for a service order as claimed in any one of claims 1 to 7 are implemented.

10. A storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the risk handling method for a service order as claimed in any one of claims 1 to 7 are implemented.

Images