Method and system for evaluating service, electronic equipment and storage medium

Abstract

The embodiment of the invention discloses a service evaluation method and system, electronic equipment and a storage medium. The method comprises the steps of obtaining to-be-evaluated service data; and based on the to-be-evaluated service data, determining a standard service path corresponding to the to-be-evaluated service path from the standard service map. And determining a node deviation degree score of each service node in the to-be-evaluated service path based on the to-be-evaluated service path and the standard service path. And determining a service deviation degree score of the to-be-evaluated service data based on the node deviation degree score of each service node in the to-be-evaluated service path. Through the mode, the knowledge graph can be utilized to perform multi-dimensional correlation analysis on the to-be-evaluated service path so as to guarantee the accuracy and objectivity of the result, and the service transparency and the user credibility can be remarkably improved by obtaining the service deviation degree score of the to-be-evaluated service data.

Description

Technical Field

[0001] The embodiments in this specification belong to the field of data processing technology, and in particular relate to a method, system, electronic device, and storage medium for evaluating services. Background Technology

[0002] As living standards continue to improve, more and more service platforms are providing users with diversified service information to meet their various needs in life.

[0003] When faced with various service information, users often find it difficult to judge the rationality of the information due to a mismatch between professional information and actual conditions. Existing information evaluation methods mainly rely on manual spot checks and static rule reviews, which result in inaccurate evaluation results and poor objectivity. Summary of the Invention

[0004] Embodiments of this disclosure provide a method, system, electronic device, and storage medium for evaluating services.

[0005] In a first aspect of the embodiments of this disclosure, a method for evaluating a service is provided. The method includes acquiring service data to be evaluated, the service data having corresponding user information, the service data including a service path to be evaluated, the service path including one or more service nodes and one or more service edges, each service edge having a corresponding pair of service nodes. Based on the service data to be evaluated, determining a standard service path corresponding to the service path to be evaluated from a standard service graph, the standard service path including one or more standard service nodes and one or more standard service edges, each standard service edge having a corresponding pair of standard service nodes. Based on the service path to be evaluated and the standard service path, determining a node deviation score for each service node in the service path to be evaluated. Furthermore, the method also includes determining a service deviation score for the service data to be evaluated based on the node deviation scores of each service node in the service path to be evaluated.

[0006] In a second aspect of the embodiments of this disclosure, a system for evaluating services is provided. The system includes a data acquisition module configured to acquire service data to be evaluated, the service data having corresponding user information, the service data including a service path to be evaluated, the service path including one or more service nodes and one or more service edges, each service edge having a corresponding pair of service nodes. A path determination module configured to determine a standard service path corresponding to the service path to be evaluated from a standard service graph based on the service data to be evaluated, the standard service path including one or more standard service nodes and one or more standard service edges, each standard service edge having a corresponding pair of standard service nodes. A scoring calculation module configured to determine a node deviation score for each service node in the service path to be evaluated based on the service path to be evaluated and the standard service path. Furthermore, the system also includes a score generation module configured to determine a service deviation score for the service data to be evaluated based on the node deviation scores of each service node in the service path to be evaluated.

[0007] In a third aspect of the embodiments of this disclosure, a computer program product is provided, including a computer program that, when executed by a processor, implements the method provided according to the first aspect.

[0008] In a fourth aspect of the embodiments of this disclosure, an electronic device is provided, including one or more processors and a memory associated with the one or more processors, the memory being used to store program instructions that, when read and executed by the one or more processors, perform a method provided according to a first scheme.

[0009] It should be understood that the description in the Summary of the Invention section is not intended to limit the key or essential features of the embodiments of this disclosure, nor is it intended to restrict the scope of this disclosure. Other features of this disclosure will become readily apparent from the following description. Attached Figure Description

[0010] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. In the drawings, the same or similar reference numerals denote the same or similar elements, wherein: Figure 1 A schematic diagram of an example environment in which several embodiments of the present disclosure may be implemented is shown; Figure 2 This disclosure illustrates schematic diagrams of interfaces for obtaining service data to be evaluated, representing some embodiments of the present disclosure. Figure 3 A flowchart of a method for evaluating a service, according to some embodiments of this disclosure, is shown; Figure 4 A schematic diagram of a standard service map of some embodiments of this disclosure is shown; Figure 5 This disclosure illustrates an example process for determining the node deviation score of each service node in a service path to be evaluated, according to some embodiments of the present disclosure. Figure 6 This illustration shows an example process for determining a third score for the data deviation of a service node, according to some embodiments of this disclosure. Figure 7 An example process for determining a node deviation score for a service node, as shown in some embodiments of this disclosure, is illustrated. Figure 8 A schematic diagram of a spectral library representing some embodiments of this disclosure is shown; Figure 9 A flowchart illustrating yet another method for evaluating a service, according to some embodiments of the present disclosure, is shown. Figure 10 A schematic diagram of an interface showing the evaluation path for displaying service data to be evaluated, according to some embodiments of this disclosure; Figure 11 Example block diagrams of systems for evaluating services, representing some embodiments of this disclosure, are shown. Figure 12 A schematic block diagram of an electronic device according to some embodiments of the present disclosure is shown. Detailed Implementation

[0011] To make the objectives, technical solutions, and advantages of this application clearer, the technical solutions in the embodiments of this specification will be clearly and completely described below with reference to the corresponding drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0012] The terms “comprising” and “having”, and any variations thereof, in this specification, claims, and the foregoing drawings are intended to cover a non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the steps or units listed, but may optionally include steps or units not listed, or may optionally include other steps or units inherent to such process, method, product, or apparatus. Depending on the context, the word “if” as it applies herein may be interpreted as “when”, “when”, “in response to determination”, or “in response to detection”.

[0013] As mentioned above, when faced with various service information, users often find it difficult to judge the reasonableness of the information due to a mismatch between professional information and the actual situation. Here, service information can be understood as information provided by the service provider to the user in various service scenarios. For example, when the service scenario is a medical service scenario, the service information may be a diagnostic report provided by a hospital to the user, which includes one or more of the following: the user's symptom type, examination items, prescribed medications, and billing data. Similarly, when the service scenario is a car repair service scenario, the service information may be a repair order provided by a car repair shop to the user, which includes one or more of the following: the user's vehicle fault codes, vehicle inspection items, names of replaced parts, and billing data. Likewise, when the service scenario is a housekeeping service scenario, the service information may be a service order provided by a housekeeping agency to the user, which includes one or more of the following: the user's house area, house cleaning items, cleaning consumables usage, and billing data. In some embodiments of this disclosure, the service scenarios may also include medical aesthetic service scenarios and education and training service scenarios, etc., and are not limited thereto.

[0014] Taking the medical service scenario as an example, users often cannot make a reasonable judgment on the diagnostic reports provided by hospitals. For example, they cannot judge whether there are any unnecessary items in the examination or whether there is an excessive amount of medicine in the prescription. As a result, problems such as excessive service or excessive consumption in the diagnostic report cannot be identified.

[0015] Existing information assessment methods primarily focus on population-based healthcare service scenarios, relying on manual sampling and static rule-based review to determine the reasonableness of service information. Manual sampling, which relies on the experience of medical experts to judge the reasonableness of service information, is not only costly in terms of manpower and has low coverage, but is also affected by individual user differences, regional consumption levels, or emerging therapies, leading to inaccuracies and objectivity in the assessment results. Static rule-based review, on the other hand, judges the reasonableness of service information based on fixed thresholds or decision tree models. For example, it compares the differences between consumption data or other individual indicators in diagnostic reports and fixed thresholds, lacking multi-dimensional correlation analysis of service information. Furthermore, this static rule-based review method cannot handle service information with special characteristics, such as identifying disease types involving complications, further contributing to inaccuracies and objectivity in the assessment results.

[0016] Furthermore, other service scenarios lack complete information assessment methods, and the information assessment methods mentioned above, which are concentrated in the population-based medical service scenario, cannot be applied to other service scenarios. For example, in the pet medical service scenario, due to incomplete laws and regulations, the lack of unified industry standards and treatment guidelines, and the influence of inconsistent qualifications of pet medical institutions, as well as the varying quality of medicines and equipment, manual sampling and static rule-based review methods are prone to leading to poor accuracy and objectivity in the assessment results.

[0017] Therefore, embodiments of this disclosure propose a method for evaluating services. In embodiments of this disclosure, service data to be evaluated is obtained. The service data to be evaluated includes corresponding user information and a service path to be evaluated. The service path to be evaluated includes one or more service nodes and one or more service edges, each service edge having a corresponding pair of service nodes. Based on the service data to be evaluated, a standard service path corresponding to the service path to be evaluated is determined from a standard service graph. The standard service path includes one or more standard service nodes and one or more standard service edges, each standard service edge having a corresponding pair of standard service nodes. Based on the service path to be evaluated and the standard service path, a node deviation score for each service node in the service path to be evaluated is determined. Furthermore, the method also includes determining a service deviation score for the service data to be evaluated based on the node deviation scores of each service node in the service path to be evaluated.

[0018] In this way, standard service paths corresponding to the service paths to be evaluated in the service data to be evaluated can be determined based on standard service graphs. Node deviation scores and service deviation scores of each service node in the service path to be evaluated can be determined based on the standard service paths. This not only utilizes knowledge graphs to conduct multi-dimensional correlation analysis of the service paths to be evaluated to ensure the accuracy and objectivity of the results, but also significantly improves service transparency and user trust by obtaining the service deviation scores of the service data to be evaluated.

[0019] Figure 1 Schematic diagrams are shown illustrating example environments in which various embodiments of this disclosure may be implemented. For example... Figure 1As shown, the example environment 100 includes a processing terminal 101 for acquiring service data to be evaluated input by a user. In one example, the processing terminal 101 has a third-party application installed for evaluating services, allowing the user to input the service data to be evaluated on the input interface of the third-party application. In another example, the processing terminal 101 has a third-party application installed containing a mini-program for evaluating services, allowing the user to search for and access the mini-program for evaluating services on the third-party application and input the service data to be evaluated on the corresponding input interface. Here, the service data to be evaluated includes user information corresponding to the service information provided by the service provider. This user information may include one or more types of information such as user name, service recipient, service item type, material consumption, and consumption indicators. The service recipient, service item type, material consumption, and consumption indicators can all be determined by the user based on the service information provided by the service provider and are applicable to the service information provided by the service provider in various service scenarios. For example, taking a diagnostic report provided by a hospital as an example, the service recipient in the user information could be the type of disease in the diagnostic report; the service item type in the user information could be the examination items in the diagnostic report; the material consumption in the user information could be the prescribed medications in the diagnostic report; and the consumption index in the user information could be the billing data in the diagnostic report. Alternatively, the user information can also be obtained by the processing terminal 101 based on the electronic text uploaded by the user on the input interface, corresponding to the service information provided by the service provider, and then confirmed by the user.

[0020] In addition, the service data to be evaluated may also include service scenario information to indicate the service scenario of the service data to be evaluated, regional information to indicate the area where the user is located, and the name of the service provider to indicate the service provider. The service scenario information may be any one of the following, but is not limited to: medical service scenario, car repair service scenario, housekeeping service scenario, medical aesthetic service scenario, and education and training service scenario.

[0021] In the input interface for obtaining service data to be evaluated in some embodiments of this disclosure, users can also upload multimodal data corresponding to service information provided by the service provider. This multimodal data can be provided to the user by the service provider and includes at least two of the following: electronic text data, image data, audio data, video data, or wearable device data. Each modal data corresponds to one or more pieces of information in the user information. For example, taking user information obtained based on a diagnostic report provided by a hospital as an example, when the user uploads electronic text, X-ray images, and physiological indicators, since these three modal data are all related to the stage of diagnosis and treatment of the disease, these three modal data correspond to the examination items and prescribed medications in the user information.

[0022] Figure 2 A schematic diagram of an interface for obtaining service data to be evaluated, according to some embodiments of this disclosure, is shown. For example... Figure 2 As shown, the interface 200 for obtaining service data to be evaluated has multiple input boxes corresponding to user information, specifically including a user name input box, a region information input box, a service scenario type input box, a service provider name input box, a service recipient input box, a service item type input box, a material consumption input box, and a consumption index input box. Users can enter the corresponding information in each input box. In addition, the interface 200 for obtaining service data to be evaluated also includes an upload data control. After the user selects the upload data control, the processing terminal 101 can pop up an upload data interface on the interface 200 for obtaining service data to be evaluated, so that the user can upload multimodal data corresponding to the service information provided by the service provider. Furthermore, the interface 200 for obtaining service data to be evaluated also includes a confirmation control. After the user selects the confirmation control, the processing terminal 101 can process the service data to be evaluated and the multimodal data entered by the user.

[0023] Example environment 100 also includes a first server 102, which is used to collect sample service datasets corresponding to each service scenario and send each sample service dataset to processing terminal 101. Here, the sample service dataset includes multiple service data samples, and each service data sample includes one or more modal data. In one example, taking a medical service scenario as an example, the first server 102 can collect multiple diagnostic reports from various hospitals, and can also collect at least one modal data from the electronic text data, image data, audio data, video data, drug batch numbers, or wearable device data (such as pet smart collars or smart bracelets) corresponding to each diagnostic report. In another example, taking an auto repair service scenario as an example, the first server 102 can collect multiple repair work orders from various auto repair shops, and can also collect at least one modal data from the OBD fault codes, parts replacement photos, or parts supply chain barcodes corresponding to each repair work order. In another example, taking a domestic service scenario as an example, the first server 102 can collect multiple service work orders from various domestic service companies, and can also collect at least one modal data from the service duration detection data or cleaning agent usage detection data corresponding to each service report.

[0024] The example environment also includes a second server 103, which collects preset rule bases corresponding to each service scenario and sends these preset rule bases to the processing terminal 101. Here, the preset rule base can be understood as the industry standard rule base for the service scenario. In one example, taking a medical service scenario, the second server 103 can collect industry standard rule bases for medical service scenarios, such as disease classification libraries, observation indicator identifier naming and coding libraries, drug coding libraries, surgical operation classification libraries, general medical and health care procedure coding libraries, and unified medical language systems. In another example, taking an automotive repair service scenario, the second server 103 can collect industry standard rule bases for automotive repair service scenarios, such as automotive OEM repair manuals, automotive maintenance process specifications, accident vehicle repair technical specifications, automotive repair industry association standards, automotive repair electronic health record systems, and comprehensive repair information databases. In yet another example, taking a domestic service scenario, the second server 103 can collect industry standard rule bases for domestic service scenarios, such as domestic service quality specifications, domestic service cleaning service specifications, and domestic service training system guidelines.

[0025] Some embodiments of this disclosure involve a first server 102 and a second server 103, which may be a hardware server, a virtual server, a cloud server, etc. Furthermore, the first server 102 and the second server 103 may also be configured as the same server, which can be used to collect sample service datasets corresponding to each service scenario and preset rule bases corresponding to each service scenario, and send the sample service datasets and preset rule bases corresponding to each service scenario to the processing terminal 101, but is not limited to this.

[0026] Furthermore, after acquiring the service data to be evaluated input by the user, the processing terminal 101 can also determine the service path to be evaluated based on the user information in the service data. In some implementations, one or more specified types of information can be extracted from the user information in the service data to be evaluated as service nodes. One or more service edges can be determined based on the specified logical relationship between any two specified types of information, and the service path to be evaluated can be obtained based on all service nodes and all service edges. Here, the service path to be evaluated includes one or more service nodes and one or more service edges. Each service edge has a corresponding pair of service nodes. A service node can be understood as specified types of information in the user information, such as one or more types of information including the service object, service item type, material consumption, and consumption index of the user information. A service edge can be understood as a specified logical relationship between any two specified types of information in the user information. For example, there is a directed line between the service object and the service item type representing the selection of the service item type based on the service object; there is a directed line between the service item type and the material consumption representing the determination of the material consumption based on the service item type; and there is a directed line between the material consumption and the consumption index representing the determination of the consumption index based on the material consumption.

[0027] In one example, taking user information obtained from a diagnostic report provided by a pet hospital, the service path to be evaluated can be represented as: pet vomiting → X-ray examination → antibiotic prescription → cost of 800 yuan. The service path to be evaluated includes four service nodes: pet vomiting, X-ray examination, antibiotic prescription, and cost of 800 yuan. Pet vomiting corresponds to the service recipient in the user information; X-ray examination corresponds to the service type in the user information; antibiotic prescription corresponds to the material consumption in the user information; and cost of 800 yuan corresponds to the consumption indicator in the user information. There is a directed line representing the selection of X-ray examination based on pet vomiting, a directed line representing the determination of antibiotic prescription based on X-ray examination, and a directed line representing the determination of cost of 800 yuan based on antibiotic prescription.

[0028] Furthermore, after obtaining the sample service datasets corresponding to each service scenario collected by the first server 102 and the preset rule bases corresponding to each service scenario collected by the second server 103, the processing terminal 101 can establish a standard service graph based on the sample service datasets corresponding to each service scenario and the preset rule bases, and determine the standard service path corresponding to the service path to be evaluated from the standard knowledge graph. Here, the standard service graph can be understood as a knowledge graph that conforms to the industry standards of service scenarios, used to provide standardized data for the rationality assessment of services. It includes multiple standard service paths, each standard service path including one or more standard service nodes and one or more standard service edges. Each standard service node can be understood as the standard information corresponding to any one of the information types mentioned above, such as service object, service project type, material consumption, and consumption index. Each standard service edge can be understood as the standard logical relationship between the standard information corresponding to any two of the information types mentioned above, such as service object, service project type, material consumption, and consumption index. This standard logical relationship is also assigned a corresponding association weight value, and all standard service nodes, all standard service edges, and all association weight values ​​can be determined based on the sample service datasets corresponding to the service scenarios and the preset rule bases.

[0029] Subsequently, the processing terminal 101 can determine the node deviation score of each service node in the service path to be evaluated, based on the service path to be evaluated and the standard service path, and determine the service deviation score of the service data to be evaluated based on the node deviation scores of each service node in the service path to be evaluated. Here, the node deviation score of each service node can be understood as being calculated based on the deviation between each service node and the corresponding standard service node in the standard service path. The higher the node deviation score, the greater the difference between the service node and the corresponding standard service node in the standard service path; the lower the node deviation score, the smaller the difference between the service node and the corresponding standard service node in the standard service path. It is understood that the service deviation score of the service data to be evaluated can be the sum of the node deviation scores of each service node, but is not limited to this.

[0030] The processing terminal 101 involved in some embodiments of this disclosure may be a smartphone, tablet computer, desktop computer, laptop computer, notebook computer, ultra-mobile personal computer (UMPC), handheld computer, PC device, personal digital assistant (PDA), routing device, virtual reality device, etc.

[0031] In this way, standard service paths corresponding to the service paths to be evaluated in the service data to be evaluated can be determined based on standard service graphs. Node deviation scores and service deviation scores of each service node in the service path to be evaluated can be determined based on the standard service paths. This not only utilizes knowledge graphs to conduct multi-dimensional correlation analysis of the service paths to be evaluated to ensure the accuracy and objectivity of the results, but also significantly improves service transparency and user trust by obtaining the service deviation scores of the service data to be evaluated.

[0032] Figure 3 A flowchart of a method for evaluating a service according to some embodiments of the present disclosure is shown. Method 300 may, for example, be derived by... Figure 1 The processing terminal 101 in the example environment 100 shown executes. For example... Figure 3 As shown in box 302, method 300 can obtain service data to be evaluated. Here, the processing terminal can obtain the service data to be evaluated input by the user through the input interface of a third-party application used for evaluating the service or the input interface of a mini-program used for evaluating the service. The service data to be evaluated has user information corresponding to the service information provided by the service provider. This user information may include one or more types of information such as user name, service recipient, service item type, material consumption, and consumption index. The service recipient, service item type, material consumption, and consumption index can all be determined by the user based on the service information provided by the service provider, and are applicable to the service information provided by the service provider in various service scenarios. In one example, taking a diagnostic report provided by a hospital as an example, the service recipient in the user information can be the disease type in the diagnostic report, the service item type in the user information can be the examination item in the diagnostic report, the material consumption in the user information can be the prescribed medication in the diagnostic report, and the consumption index in the user information can be the billing data in the diagnostic report. Of course, user information can also be obtained by the processing terminal based on the electronic text uploaded by the user on the input interface, which corresponds to the service information provided by the service provider, and then confirmed by the user.

[0033] Understandably, to improve service evaluation efficiency, after acquiring the service data to be evaluated, the processing terminal can extract one or more specified types of information as service nodes from the user information of the service data, and determine one or more service edges based on the specified logical relationship between any two specified types of information. Finally, it obtains the service path to be evaluated based on all service nodes and all service edges. Here, the service path to be evaluated includes one or more service nodes and one or more service edges. Each service edge has a corresponding pair of service nodes. A service node can be understood as specified types of information in the user information, such as one or more types of information including the service object, service item type, material consumption, and consumption index. A service edge can be understood as a specified logical relationship between any two specified types of information in the user information. For example, there is a directed line between the service object and the service item type representing the selection of the service item type based on the service object; a directed line between the service item type and the material consumption type representing the determination of the material consumption based on the service item type; and a directed line between the material consumption type and the consumption index representing the determination of the consumption index based on the material consumption.

[0034] In box 304, method 300 can determine the standard service path corresponding to the service path to be evaluated from the standard service graph based on the service data to be evaluated. Here, the standard service graph can be understood as a knowledge graph that conforms to the industry standards of the service scenario, used to provide standardized data for the rationality assessment of the service. It is established based on the sample service dataset corresponding to the service scenario of the service data to be evaluated and a preset rule base, including multiple standard service paths. Each standard service path includes one or more standard service nodes and one or more standard service edges. Among them, each standard service node can be understood as the standard information corresponding to any one of the information types mentioned above, namely service object, service project type, material consumption, and consumption indicator. Each standard service edge can be understood as the standard logical relationship between the standard information corresponding to any two of the information types mentioned above, namely service object, service project type, material consumption, and consumption indicator. This standard logical relationship is also assigned a corresponding association weight value, and all standard service nodes, all standard service edges, and all association weight values ​​can be determined based on the sample service dataset corresponding to the service scenario and the preset rule base.

[0035] In one example, taking the service scenario of the service data to be evaluated as a medical service scenario, the standard service graph can be a four-dimensional knowledge graph containing disease types, examination items, prescribed medications, and billing data. This standard service graph includes multiple standard service paths with standard disease types, standard examination items, standard prescribed medications, and standard billing data. Within each standard service path, there are directed lines representing the standard logical relationship between the standard disease type and the standard examination item, and corresponding association weight values. Similarly, within each standard service path, there are directed lines representing the standard logical relationship between the standard examination item and the standard prescribed medication, and corresponding association weight values. Finally, within each standard service path, there are directed lines representing the standard logical relationship between the prescribed medication and the billing data, and corresponding association weight values. (See here for more information.) Figure 4 The schematic diagram shown is a standard service map according to some embodiments of the present disclosure, such as Figure 4 As shown, the standard service graph 400 includes three standard service paths. The first standard service path can be represented as symptom type A1 → examination item B1 → prescription medication C1 → billing data D1; the second standard service path can be represented as symptom type A2 → examination item B2 → prescription medication C2 → billing data D2; and the third standard service path can be represented as symptom type A3 → examination item B3 → prescription medication C3 → billing data D3. Here, different association weight values ​​characterize the association strength between two corresponding standard service nodes. For example, an association weight of 1 indicates a standard association strength; an association weight of 0.75 indicates a moderate association strength; and an association weight of 2 indicates a relatively strong association.

[0036] In some implementations, when the processing terminal determines the standard service path corresponding to the service path to be evaluated from the standard service graph, it can perform a conversion process on the standard service graph to convert each standard service path into a standard path graph (e.g., represented as an adjacency matrix or adjacency list, etc., a graph structure format, including standard path nodes arranged in rows, standard path nodes arranged in columns, and association weight values ​​between the two standard path nodes corresponding to each standard service edge). Then, based on the association weight values ​​assigned to each standard service edge in the standard service graph, the association weight values ​​corresponding to the two service nodes of each service edge in the service path to be evaluated can be determined, and based on the service path to be evaluated and the association weight values ​​corresponding to the two service nodes of each service edge, the service path to be evaluated can be converted into a path graph to be evaluated. Finally, by calculating the similarity between the path graph to be evaluated and each standard path graph, the standard service path corresponding to the standard path graph with the highest similarity can be used as the standard service path corresponding to the service path to be evaluated.

[0037] In box 306, method 300 can determine the node deviation score of each service node in the service path to be evaluated based on the service path to be evaluated and the standard service path. In some implementations, the processing terminal can use the GraphSAGE algorithm to process the service path to be evaluated to obtain the feature vectors corresponding to each service node in the service path to be evaluated, and can also use the GraphSAGE algorithm to process the standard service path to obtain the feature vectors corresponding to each standard service node in the standard service path. The node deviation score of each service node in the service path is determined by calculating the cosine similarity or Euclidean distance between the feature vectors corresponding to each service node and the feature vectors corresponding to the corresponding standard service nodes. Here, taking the processing of the service path to be evaluated using the GraphSAGE algorithm as an example, the context information of each service node can be extracted from the path graph corresponding to the service path, such as the association weight values ​​of adjacent service nodes and the service edges they belong to, and the service nodes and context information are encoded to obtain the feature vectors corresponding to each service node.

[0038] In one example, when the processing terminal determines the node deviation score of each service node in the path to be evaluated, it can determine the node deviation score of the corresponding service node based on the preset interval of each cosine similarity after obtaining the cosine similarity between the feature vector corresponding to each service node and the feature vector corresponding to the corresponding standard service node. For example, when the cosine similarity is in the preset first interval, the node deviation score of the corresponding service node can be determined to be 0.5, which means that the difference between the service node and the corresponding standard node is large; when the cosine similarity is in the preset second interval, the node deviation score of the corresponding service node can be determined to be 0, which means that the difference between the service node and the corresponding standard node is small. Of course, in some examples of this disclosure, the node deviation score of each service node can also be set to other values, and is not limited to this.

[0039] In box 308, method 300 can determine the service deviation score of the service data to be evaluated based on the node deviation scores of each service node in the service path to be evaluated. In some implementations, the processing terminal can sum the node deviation scores of each service node in the service path to be evaluated and use the sum as the service deviation score of the service data to be evaluated. It is understood that when the service deviation score of the service data to be evaluated is high, it indicates that the reasonableness of the service data to be evaluated is poor, such as the possibility of over-service or over-consumption, and the user can be reminded to inquire with the service provider for details; when the service deviation score of the service data to be evaluated is low, it indicates that the reasonableness of the service data to be evaluated is high, and the service deviation score of the service data to be evaluated can be displayed to the user so that the user can more intuitively understand the reasonableness of the service data to be evaluated.

[0040] In this way, standard service paths corresponding to the service paths to be evaluated based on standard service graphs can be determined. Then, node deviation scores and service deviation scores for each service node in the service path to be evaluated can be determined based on the standard service paths. This not only utilizes knowledge graphs to perform multi-dimensional correlation analysis on the service paths to be evaluated, ensuring the accuracy and objectivity of the results, but also reflects the reasonableness of the service data to be evaluated by obtaining the service deviation scores, thereby significantly improving service transparency and user trust.

[0041] Figure 5 An example process for determining the node deviation score of each service node in a service path to be evaluated, according to some embodiments of the present disclosure, is illustrated. This example process 500 for determining the node deviation score of each service node in a service path to be evaluated can be obtained by a processing terminal based on a graph neural network computing engine well known in the art. Figure 5 As shown in box 502, the example process 500 for determining the node deviation score of each service node in the service path to be evaluated can determine the structural deviation score of each service node in the service path to be evaluated based on the service path to be evaluated and the standard service path. Here, the structural deviation score of the service node can be understood as the deviation score between each service node in the service path to be evaluated and the standard service path. The deviation score between the service node in the service path to be evaluated and the standard service path can be determined based on the matching results between the service node in the service path to be evaluated and the standard service path.

[0042] In some implementations, when determining the structural deviation score of each service node in the service path to be evaluated based on the service path to be evaluated and the standard service path, the processing terminal can determine whether each service node in the service path to be evaluated matches the standard service path. In one example, the processing terminal can extract the context information of each service node from the path graph corresponding to the service path to be evaluated, including, for example, the association weight values ​​of adjacent service nodes and their respective service edges (the association weight values ​​of the service edges where the service node is located can be determined based on the association weight values ​​assigned to each standard service edge in the standard service graph), and encode each service node and its context information to obtain the feature vector corresponding to each service node. Here, the processing terminal can also use the same method to determine the feature vectors corresponding to each standard service node in the standard service path.

[0043] Subsequently, the processing terminal can determine the corresponding standard service nodes in the standard service path based on the locations of each service node in the service path to be evaluated. It then determines whether each service node in the service path to be evaluated matches the standard service path by calculating the cosine similarity (or Euclidean distance) between the feature vectors corresponding to each service node and the feature vectors corresponding to the standard service nodes. Understandably, when the cosine similarity exceeds a preset cosine similarity threshold, it can be determined that the corresponding service node in the service path to be evaluated matches the standard service path; when the cosine similarity does not exceed the preset cosine similarity threshold, it can be determined that the corresponding service node in the service path to be evaluated does not match the standard service path.

[0044] Subsequently, in response to determining that a service node matches a standard service path, the processing terminal may determine a structural deviation score for the corresponding service node as a first score, which may be, for example, 0. Furthermore, in response to determining that a service node does not match a standard service path, the processing terminal may determine a structural deviation score for the corresponding service node as a second score, which may be, for example, 0.5. Of course, in some embodiments of this disclosure, the second score may be set to other values ​​and is not limited thereto.

[0045] In box 504, the example process 500 for determining the node deviation score of each service node in the service path to be evaluated can determine the data deviation score of the corresponding service node based on the node multimodal data corresponding to the service node and preset verification rules. Here, the node multimodal data includes at least two different modalities of modal data, such as at least two modalities of data selected from electronic text data, image data, audio data, video data, and wearable device data, which can be uploaded by the user when the processing terminal obtains the service data to be evaluated. It can be understood that the data deviation score of the service node can be understood as the deviation score between the different modal data corresponding to each service node in the service path to be evaluated, and the deviation score between the different modal data corresponding to the service node in the service path to be evaluated can be determined based on the data contradiction results between the different modal data corresponding to the service node. It should be noted that some service nodes in the service path to be evaluated do not have corresponding multimodal data. For example, taking all service nodes in the service path to be evaluated, including disease type, examination items, prescribed medications, and fee data, when the user uploads electronic text, X-ray images, and physiological indicators, since these three modal data are all related to the diagnosis and treatment stage of the disease, the examination items and prescribed medications in the service path to be evaluated are all related to these three modal data. However, the disease type and fee data are not related to these three modal data. Therefore, the data deviation score corresponding to the disease type and fee data can be set to 0.

[0046] In some implementations, when determining the data deviation score of a service node based on the node multimodal data corresponding to the service node and preset verification rules, the processing terminal can determine whether there is a data contradiction in the node multimodal data based on the node multimodal data corresponding to the service node and preset verification rules. Here, the preset verification rules can be: if the similarity between the modal data in the node multimodal data does not exceed a preset modal similarity threshold, it is determined that there is a data contradiction in the corresponding node multimodal data; if it exceeds the preset modal similarity threshold, it is determined that there is no data contradiction in the corresponding node multimodal data. In one example, the processing terminal can extract feature data from each modal data of the node multimodal data. For example, it can extract feature data such as symptom description and symptom location from electronic text, feature data such as time points and lesion locations with lesion features from X-ray images, and feature data such as detection indicator names and detection indicator time series values ​​from physiological indicators. Next, modal alignment processing can be performed on each modal data. For example, a corresponding path graph can be constructed based on the feature data of each modal data (each node in the path graph can correspond to the feature data of the modal data, and the association weight value assigned to any two nodes can be determined based on the association strength between the corresponding feature data). Then, modal aggregation processing is performed on each node in the path graph (e.g., aggregation processing of adjacent node information within the same modality and aggregation processing of cross-modal node information, and also aggregation processing of the corresponding service node information in the path graph corresponding to the service path to be evaluated, so as to establish an association between the modal data and the path graph corresponding to the service path to be evaluated. This node information includes the association weight value assigned to the corresponding service node and the corresponding service edge). The nodes and the corresponding modal aggregation results are then embedded to obtain the fused feature data of each node. Subsequently, the fused feature data of each node corresponding to each modal data can be encoded to obtain the feature vector of each modal data in the same dimension. Afterwards, by calculating the similarity between the feature vectors corresponding to each modal data, combined with preset verification rules, it can be determined whether there are data contradictions in the multimodal data of the nodes.

[0047] It is understandable that when the similarity between any two feature vectors in the feature vectors corresponding to each modality data does not exceed the preset modality similarity threshold, it is determined that there is a data contradiction in the node multimodal data of the corresponding service node; when the similarity between any two feature vectors exceeds the preset modality similarity threshold, it is determined that there is no data contradiction in the node multimodal data of the corresponding service node.

[0048] Subsequently, in response to determining that there is a data contradiction in the multimodal data of a service node, the processing terminal may determine the data deviation score of the corresponding service node as a third score, which may be, for example, 0.1. Furthermore, in response to determining that there is no data contradiction in the multimodal data of a service node, the processing terminal may determine the data deviation score of the corresponding service node as a fourth score, which may be, for example, 0. Of course, in some embodiments of this disclosure, the third score may be set to other values. For example, when the similarity between any two feature vectors is determined to be much less than a preset modal similarity threshold, i.e., the difference between the corresponding two modal data is large, the third score may be determined to be 0.2, and this is not a limitation.

[0049] In some embodiments, when the processing terminal determines that there is a data contradiction in the multimodal data of a node and determines that the data deviation score of the corresponding service node is the third score, it can also determine one or more contradictory modal data combinations from the modal data of that node's multimodal data. It is understood that when the similarity between any two feature vectors in the feature vectors corresponding to each modal data does not exceed a preset modal similarity threshold, the two modal data corresponding to those two feature vectors can be determined as a contradictory modal data combination.

[0050] Subsequently, the processing terminal can determine a third score based on the number of contradictory modal data combinations in the node's multimodal data, and use this third score as the data deviation score for the corresponding service node. In one example, each contradictory modal data combination corresponds to a preset deviation coefficient, and the processing terminal can determine the third score by multiplying the number of contradictory modal data combinations by the preset deviation coefficient.

[0051] Please see Figure 6 The illustration shows an example process for determining a service node's data deviation score as a third value according to some embodiments of this disclosure, such as... Figure 6As shown, taking X-ray examination as an example of a service node in the service path to be evaluated, the example process 600 for determining the data deviation score of the service node as the third score can be based on three modal data corresponding to X-ray examination: medical record text data, image data, and physiological indicator data. The similarity between medical record text data and image data, the similarity between image data and physiological indicator data, and the similarity between physiological indicator data and medical record text data can be determined respectively. Specifically, if the similarity between medical record text data and image data is below a preset threshold, indicating a data inconsistency, then this medical record text data and image data can be identified as a contradictory modality data combination, and the preset deviation coefficient for this contradictory modality data combination can be 0.1. Similarly, if the similarity between image data and physiological indicator data is below a preset threshold, indicating no data inconsistency, then the similarity between physiological indicator data and medical record text data is also below a preset threshold, indicating a data inconsistency, and thus this physiological indicator data and medical record text data can be identified as a contradictory modality data combination, and the preset deviation coefficient for this contradictory modality data combination can be 0.1. Then, the number of contradictory modality data combinations is multiplied by the preset deviation coefficient to obtain 0.2, and 0.2 is determined as the data deviation score for the service node.

[0052] In box 506, the example process 500 for determining the node deviation score of each service node in the service path to be evaluated can determine the node deviation score of the corresponding service node based on the structural deviation score and the data deviation score of each service node. Here, the processing terminal can perform weighted summation of the structural deviation score and the data deviation score of each service node based on preset weight values ​​to obtain the node deviation score of the corresponding service node.

[0053] This approach enables the simultaneous consideration of multimodal data from service nodes within the evaluated service path when using knowledge graphs for multidimensional correlation analysis. By employing multimodal data alignment, the joint representation problem of multimodal data is addressed, improving the accuracy of multimodal data analysis and enhancing the reliability of node deviation scores within the evaluated service data. Furthermore, a lightweight framework based on a graph neural network computing engine allows even low-computing-power processing terminals to evaluate the reasonableness of service data, thus meeting the service evaluation needs of processing terminals with varying performance levels.

[0054] In some implementations, when determining the node deviation score of a service node based on its structural deviation score and data deviation score, the processing terminal can also determine the node deviation score of the corresponding service node based on its structural deviation score, data deviation score, and a preset data deviation correction coefficient. Here, the preset data deviation correction coefficient can be understood as a preset coefficient used to correct the data deviation score, and its range can be set between 15% and 30%. In one example, the method for determining the node deviation score of a service node can be found in the expression shown below: Node deviation score = Structure deviation score + Data deviation score × Preset data deviation correction coefficient In this way, the impact of multimodal data on the node deviation score of service nodes can be effectively weighed, further making the node deviation score of service nodes in the service data to be evaluated more reliable.

[0055] Figure 7 An example process for determining a node deviation score of a service node according to some embodiments of the present disclosure is shown. This example process 700 for determining the node deviation score of a service node can be obtained by a processing terminal based on a graph neural network computing engine well known in the art. Figure 7 As shown in box 702, the example process 700 for determining the node deviation score of a service node can determine the feature vector corresponding to each modality in the node multimodal data based on the node multimodal data corresponding to the service node. In some implementations, the processing terminal can extract feature data from each modality of the node multimodal data. For example, it can extract feature data such as symptom description and symptom location from electronic text, feature data such as time points and lesion locations with lesion features from X-ray images, and feature data such as detection indicator names and time series values ​​of detection indicators from physiological indicators. Next, modal alignment processing can be performed on each modal data. For example, a corresponding path graph can be constructed based on the feature data of each modal data (each node in the path graph can correspond to the feature data of the modal data, and the association weight value assigned to any two nodes can be determined based on the association strength between the corresponding feature data). Then, modal aggregation processing is performed on each node in the path graph (for example, including the aggregation processing of adjacent node information in the same modality and the aggregation processing of cross-modal node information, and of course, it can also include the aggregation processing of the corresponding service node information in the path graph corresponding to the service path to be evaluated, so as to establish an association between the modal data and the path graph corresponding to the service path to be evaluated. This node information includes the association weight value assigned to the corresponding service node and the corresponding service edge). The nodes and the corresponding modal aggregation results are then embedded to obtain the fused feature data of each node. Then, the fused feature data of the nodes corresponding to each modal data can be encoded to obtain the feature vector of each modal data in the same dimension.

[0056] In box 704, the example process 700 for determining the node deviation score of a service node can determine the adjusted modal weights corresponding to each modal data based on the feature vectors corresponding to each modal data in the node's multimodal data and the feature vectors of the corresponding service nodes. Here, the method for determining the feature vectors of each service node in the service path to be evaluated can be found above, and will not be elaborated further. It is understood that the adjusted modal weights corresponding to each modal data range from 0 to 1, and the sum of the adjusted modal weights corresponding to each modal data of each service node is 1. For example, taking X-ray examination and antibiotic prescription as service nodes, both X-ray examination and antibiotic prescription have three modalities of data: medical record text data, image data, and physiological indicator data. Among them, the modal weight corresponding to the medical record text data of X-ray examination can be 0.4, the modal weight corresponding to the image data can be 0.3, and the modal weight corresponding to the physiological indicator data can be 0.3; the modal weight corresponding to the medical record text data of antibiotic prescription can be 0.5, the modal weight corresponding to the image data can be 0.2, and the modal weight corresponding to the physiological indicator data can be 0.3.

[0057] In some implementations, when determining the adjusted modal weights corresponding to each modal data, the processing terminal can calculate the similarity between the feature vectors of each service node and the feature vectors corresponding to each modal data, based on the feature vectors of each service node and the feature vectors corresponding to each modal data. Based on all similarities, the adjusted modal weights corresponding to each modal data are then determined. In one example, after calculating the similarity between the feature vectors of each service node and the feature vectors corresponding to each modal data, the processing terminal can sum the similarities corresponding to all modal data and determine the ratio between the similarity of each modal data and the summation result as the adjusted modal weight corresponding to the corresponding modal data.

[0058] In some embodiments of this disclosure, when determining the modal weights corresponding to each modal data, the processing terminal can further adjust the preset modal weights based on the differences in the similarities between the feature vectors of each service node and the feature vectors corresponding to each modal data after calculating the similarity between them. Here, taking a medical service scenario including three modal data types—electronic text data, image data, and physiological indicator data—as an example, when there are significant differences between the similarity of the image data and the similarity of the electronic medical record text data and the similarity of the physiological indicator data, the preset modal weight of the image data can be reduced (e.g., reduced by 0.2) in the preset modal weights corresponding to the medical service scenario, while the preset modal weights of the electronic text data and the physiological indicator data can be increased (e.g., increased by 0.1 respectively).

[0059] In box 706, the example process 700 for determining the node deviation score of a service node can determine the target data deviation correction coefficient based on the adjusted modal weights corresponding to each modal data and the preset modal weights. Here, the preset modal weights can correspond to the service scenarios of the service data to be evaluated; that is, different service scenarios have corresponding preset modal weights. For example, in the medical service scenario, which includes three modal data types—electronic text data, image data, and physiological indicator data—since image data has certain reference value in the diagnostic stage, the preset modal weights for these three modal data types can be set to 0.4, 0.3, and 0.3, respectively. That is, in the medical service scenario, the preset modal weight for electronic text data is 0.4, the preset modal weight for image data is 0.3, and the preset modal weight for physiological indicator data is 0.3.

[0060] In one implementation, when determining the target data deviation correction coefficient, the processing terminal can determine whether the adjusted modal weights and preset modal weights corresponding to each modal data are consistent, based on the adjusted modal weights and preset modal weights corresponding to each modal data. The target data deviation correction coefficient is then determined based on the consistency results of the adjusted modal weights and preset modal weights corresponding to each modal data. In one example, the correction coefficient corresponding to the number of inconsistent consistency results for each modal data of the service node can be determined within a range of 15%-30%, and this correction coefficient is used as the target data deviation correction coefficient. Here, the number of inconsistent consistency results corresponds to the correction coefficient; for example, when the number of inconsistent consistency results is 0, the corresponding correction coefficient is 15%; when the number of inconsistent consistency results is 1, the corresponding correction coefficient is 20%; when the number of inconsistent consistency results is 2, the corresponding correction coefficient is 25%; and when the number of inconsistent consistency results exceeds 2, the corresponding correction coefficient is 30%.

[0061] In box 708, the example process 700 for determining the node deviation score of a service node can determine the node deviation score of the corresponding service node based on the structural deviation score, data deviation score, and target data deviation correction coefficient of each service node. In one implementation, the method for determining the node deviation score of a service node can be referred to the expression shown below: Node deviation score = Structure deviation score + Data deviation score × Target data deviation correction coefficient In this way, the modal weights of each modal data can be dynamically adjusted to improve the reliability of the deviation correction coefficients corresponding to the multimodal data, thereby ensuring the accuracy and rationality of the node deviation scores of the service nodes in the service data to be evaluated.

[0062] In some implementations, when the processing terminal acquires the service data to be evaluated, it can also acquire service scenario information that indicates the service scenario of the service data to be evaluated. The service scenario indicated by the service scenario information can be any one of medical service scenario, automobile repair service scenario, housekeeping service scenario, medical aesthetic service scenario, and education and training service scenario. The service scenario information can be obtained by user input or by the processing terminal based on the electronic text data input by the user.

[0063] Different service scenarios have corresponding standard service graphs. When constructing the standard service graph for a service scenario, the processing terminal can base it on... Figure 1 Example environment 100 shows a first server 102 acquiring a sample service dataset corresponding to a service scenario. This sample service dataset includes multiple service data samples, each of which includes one or more modal data. In one example, taking a medical service scenario as an example, the processing terminal can acquire multiple diagnostic reports from various hospitals, and can also collect at least one modal data from each diagnostic report, including electronic text data, image data, audio data, video data, drug batch numbers, or wearable device data (such as pet smart collars or smart bracelets).

[0064] Subsequently, the processing terminal can standardize the sample service dataset to obtain corresponding standard service data, and extract standard service nodes corresponding to the service scenario from the standard service data. Here, standardizing the sample service dataset can involve converting unstructured data (such as text data or image data), semi-structured data (such as JSON format data or XML format data), and structured data (such as tabular data) in the sample service dataset into data of a unified format (such as feature vectors or graph structures). Of course, it can also include data standardization processing methods well-known in the field, such as semantic alignment processing, data cleaning processing, data denoising processing, and missing value completion processing, which will not be elaborated on here.

[0065] It is understandable that standard service nodes corresponding to a service scenario include one or more of the following: service recipient, service item type, material consumption, and consumption indicators. In one example, taking a medical service scenario, the standard service data is a standardized diagnostic report. The disease type corresponding to the service recipient, the examination items corresponding to the service item type, the prescribed medications corresponding to the material consumption, and the fee data corresponding to the consumption indicators extracted from this diagnostic report can all be used as standard service nodes.

[0066] Subsequently, the processing terminal can also be based on Figure 1Example environment 100 shows a second server 103 that obtains a preset rule base corresponding to the service scenario, and constructs a standard service graph based on each standard service node in the standard service data and the preset rule base. Here, the preset rule base can be understood as an industry standard rule base for the service scenario. For example, when the service scenario is a medical service scenario, the preset rule base may include industry standard rule bases for medical service scenarios such as disease classification library, observation indicator identifier naming and encoding library, drug encoding library, surgical operation classification library, medical and health care general procedure encoding library, and unified medical language system.

[0067] In one example, the processing terminal can construct a standard service graph based on the standard service nodes in the standard service data and a preset rule base using a dual-channel update mechanism. For instance, the standard service nodes and the preset rule base in the standard service data can be stored separately. Here, all standard service data can be integrated into a dynamic practice library. Then, the preset rule base is standardized to extract the standard service nodes corresponding to the service scenario and the standard logical relationships between the standard service nodes from the standardized preset rule base. By integrating the standard service nodes corresponding to the service scenario and the standard logical relationships between the standard service nodes, a standard service graph is constructed. The standard service graph includes multiple standard service paths. Each standard service path includes four standard service nodes (corresponding to service objects, service item types, material consumption, and consumption indicators, respectively) and service edges between every two adjacent standard service nodes. Each service edge can be a directed line representing the standard logical relationship between the corresponding two standard service nodes, and each service edge has a correlation weight value corresponding to the correlation strength reflected by the standard logical relationship.

[0068] Subsequently, an attention mechanism can be used to fuse the standard service nodes in the standard service graph and the corresponding standard service nodes in the dynamic practice library to update the standard service nodes in the standard service graph. For example, attention weights can be calculated based on the context information of each standard service node in the standard service graph and the context information of the corresponding standard service nodes in the dynamic practice library. The results of the two weight calculations can then be combined to perform a weighted calculation on the standard service nodes in the standard service graph and the corresponding standard service nodes in the dynamic practice library, and the standard service nodes in the standard service graph can be updated to reflect this weighted calculation result. Furthermore, new standard service nodes that differ from those in the preset rule base can be selected from the dynamic practice library. The standard service graph can then be updated based on these new standard service nodes and the logical relationships between them and the new standard service nodes in the dynamic practice library. This can include adding one or more standard service paths and adjusting the association weight values ​​assigned to existing standard service edges, and is not limited to these methods.

[0069] In addition, the processing terminal can also update the standard service graph through online learning. In one example, after obtaining new service data samples, the processing terminal can extract one or more standard service nodes and the standard logical relationships between them, and update the standard service graph based on these standard service nodes and their relationships. This could include adding one or more standard service paths and adjusting the association weights assigned to existing standard service edges, but is not limited to these steps.

[0070] In this way, more effective and targeted standard service maps can be constructed based on the service scenarios of the service data to be evaluated, thereby ensuring the accuracy and rationality of the node deviation scores of service nodes in the service data to be evaluated.

[0071] In some implementations, when the processing terminal determines the standard service path corresponding to the service path to be evaluated from the standard service graph based on the service data to be evaluated, it can also determine the scene service graph corresponding to the service scenario information of the service data to be evaluated from a graph library including one or more scene service graphs as the standard service graph. Here, the graph library includes one or more service scenarios and the scene service graphs corresponding to each service scenario. Each scene service graph can be constructed from the sample service dataset and the preset rule base corresponding to the corresponding service scenario.

[0072] Please see Figure 8 The illustration shows a schematic diagram of a spectrum library according to some embodiments of the present disclosure. For example... Figure 8 As shown, the service scenario types in the graph library 800 include medical service scenarios, auto repair service scenarios, housekeeping service scenarios, and education and training service scenarios. Medical service scenarios have corresponding medical service graphs, auto repair service scenarios have corresponding auto repair service graphs, housekeeping service scenarios have corresponding housekeeping service graphs, and education and training service scenarios have corresponding education and training service graphs. Specifically, the standard service node types for each standard service path in the medical service graph can be represented as disease type, examination items, drug type, and fee data, respectively; the standard service node types for each standard service path in the auto repair service graph can be represented as fault code type, examination items, replacement part type, and fee data, respectively; the standard service node types for each standard service path in the housekeeping service graph can be represented as house area, cleaning items, consumable usage, and fee data, respectively; and the standard service node types for each standard service path in the education and training service graph can be represented as student level, course content, and tuition fee, respectively.

[0073] In this way, standard service graphs corresponding to various service scenarios can be quickly switched based on the graph library, enabling the processing terminal to evaluate the rationality of service data corresponding to various service scenarios. This not only reduces the deployment cost of service rationality evaluation for different service scenarios, but also effectively ensures the accuracy and reliability of the rationality evaluation results of service data corresponding to each service scenario.

[0074] Figure 9 A flowchart of yet another method for evaluating a service according to some embodiments of the present disclosure is shown. This method 900 can be processed by a processing terminal based on a graph neural network computing engine well known in the art. Figure 9 As shown in box 902, the method 900 for evaluating the service can determine the service deviation score of the service data to be evaluated based on the node deviation score of each service node in the service path to be evaluated. The method for determining the service deviation score of the service data to be evaluated is described above and will not be elaborated upon here.

[0075] In box 904, the method 900 for evaluating services can determine historical service data corresponding to the service scenario of the service data to be evaluated based on the service scenario information of the service data to be evaluated. Here, historical service data corresponding to the service scenario of the service data to be evaluated can be understood as one or more normal service data where the service scenario is consistent with the service scenario of the service data to be evaluated, and one or more service nodes are consistent with one or more service nodes of the service data to be evaluated. In some implementations, the processing terminal can filter out multiple service data samples containing one or more service nodes in the service path to be evaluated from the sample service dataset corresponding to the service scenario of the service data to be evaluated, and determine one or more normal service data as historical service data from the multiple service data samples through manual experience or preset rules. In one example, multiple normal service data containing service item types in the service path to be evaluated from the sample service dataset can be filtered out as historical service data. For example, taking a medical service scenario as an example, the service item type in the historical service data can be a blood routine examination, and the consumption data in the historical service data can be 200 yuan.

[0076] In box 906, the method 900 for evaluating services can determine a basic threshold for deviation from the service scenario of the service data to be evaluated based on historical service data corresponding to the service scenario of the service data to be evaluated. In some implementations, the processing terminal can determine the corresponding historical service path based on the historical service data corresponding to the service scenario of the service data to be evaluated, then calculate the node deviation score of each service node in the historical service path, and sort all the node deviation scores of the service nodes in ascending order. It is understood that the methods for determining the corresponding historical service path based on historical service data and calculating the node deviation score of each service node in the historical service path can be found above, and will not be elaborated further here.

[0077] Subsequently, the processing terminal can determine the node deviation score at a preset position from the node deviation scores of all service nodes after sorting and processing, as the basic deviation threshold corresponding to the service scenario of the service data to be evaluated. Here, the preset position can be understood as a proportional coefficient set based on human experience. For example, if the preset position is 20% and the number of node deviation scores of all service nodes is 100, the node deviation score at the 20th position in the node deviation scores of all service nodes after sorting and processing can be determined as the basic deviation threshold corresponding to the service scenario of the service data to be evaluated.

[0078] Of course, some embodiments of this disclosure involve, after determining the deviation score of a node at a preset position, multiplying the deviation score of the node at the preset position by a preset premium coefficient, and using the product as the basic deviation threshold corresponding to the service scenario of the service data to be evaluated. This allows the preset premium coefficient to ensure that the basic deviation threshold meets the reasonableness of the premium for the service scenario. Here, different service scenarios have corresponding preset premium coefficients. For example, the preset premium coefficient for a medical service scenario can be 1, the preset premium coefficient for a car repair service scenario can be 1, and the preset premium coefficient for a cosmetic surgery service scenario can be 1.5, but is not limited to these.

[0079] In box 908, the method 900 for evaluating a service can determine a warning result for the service data to be evaluated based on a basic deviation threshold corresponding to the service scenario of the service data to be evaluated and a service deviation score of the service data to be evaluated. In one implementation, the processing terminal can compare the service deviation score of the service data to be evaluated with a threshold range corresponding to the basic deviation threshold to determine a warning level for the service data to be evaluated. This warning level can be any one of no risk, low risk, medium risk, and high risk. Furthermore, when the warning level of the service data to be evaluated is determined to be any one of low risk, medium risk, and high risk, corresponding warning details can be determined based on the feature values ​​of the service node, and the warning level of the service data to be evaluated and the warning details of the service node can be integrated into a warning result for the service data to be evaluated. Here, the threshold range corresponding to the basic deviation threshold can be understood as at least two threshold ranges determined based on the basic deviation threshold and the threshold setting range, and different threshold ranges correspond to different warning levels. For example, taking a basic deviation threshold of 0.2 and a threshold setting range of 0-1, three threshold intervals can be determined: 0-0.2, 0.2-0.4, 0.4-0.6, and 0.6-1. The warning level corresponding to the threshold interval of 0-0.2 (excluding 0 but including 0.2) can be no risk; the warning level corresponding to the threshold interval of 0.2-0.4 (excluding 0.2 but including 0.4) can be low risk; the warning level corresponding to the threshold interval of 0.4-0.6 (excluding 0.4 but including 0.6) can be medium risk; and the warning level corresponding to the threshold interval of 0.6-1 (excluding 0.6 but including 1) can be high risk.

[0080] In one example, when the processing terminal determines the corresponding warning details based on the feature values ​​of a service node, it can parse the corresponding feature data from the service data to be evaluated based on multiple feature dimensions, and transform the corresponding feature data based on the set range of each feature dimension to obtain the initial feature value of the service node. Here, the type of service node can be either the service item type or the material consumption quantity, and the initial feature value of the service node includes the initial feature values ​​corresponding to each feature dimension. For example, in a medical service scenario, taking X-ray examination as the service node, the corresponding set multiple feature dimensions include "whether it is executed" (set range is 0 and 1), "parameter compliance" (set range is 0-1), and "result reporting status" (set range is 0 and 1). When the initial feature value of X-ray examination is (0.0, 0.32, 0.0), it can be represented that X-ray examination was not executed, the parameter compliance of X-ray examination is partially compliant, and the result of X-ray examination was not reported.

[0081] Subsequently, the processing terminal can encode the service node and its corresponding context information into an embedding vector based on the path graph corresponding to the service path to be evaluated. Then, based on the aforementioned multiple feature dimensions and their respective ranges, the embedding vector is reconstructed to obtain the reconstructed feature value of the service node. This reconstructed feature value includes the reconstructed feature values ​​corresponding to each feature dimension. For example, in a medical service scenario, taking X-ray examination as the service node, the corresponding multiple feature dimensions include "whether it is executed" (range 0 and 1), "parameter compliance" (range 0-1), and "result reporting status" (range 0 and 1). For instance, when the reconstructed feature value of the X-ray examination is (0.91, 0.89, 0.76), it can be represented as: the X-ray examination should be executed, the X-ray examination parameter compliance is highly compliant, and the X-ray examination result has been reported.

[0082] Subsequently, the processing terminal can determine the warning details for the corresponding node based on the difference between the initial and reconstructed feature values ​​of the service node. In one example, the processing terminal can calculate the feature value difference for each feature dimension based on the initial and reconstructed feature values ​​of the service node, and determine the warning details based on the feature dimension corresponding to the largest feature value difference. For example, taking an X-ray inspection as the service node, with multiple feature dimensions including "execution status," "parameter compliance," and "result reporting status," when the initial feature value of the X-ray inspection is (0.0, 0.32, 0.0) and the reconstructed feature value is (0.91, 0.89, 0.76), the feature value difference for the feature dimension "execution status" is the largest, thus determining that the warning details for the X-ray inspection could be "necessary inspection not performed."

[0083] This approach enables multi-threshold early warning processing for service deviation scores in the service data to be evaluated, resulting in clearer and more intuitive evaluation results. For example, it can identify unnecessary parts replacement in the car repair service scenario, detect inflated working hours or excessive consumption of cleaning agents in the housekeeping service scenario, and identify the promotion of expensive courses beyond the scope of capabilities in the education and training service scenario. Furthermore, it can provide universally applicable standards for early warning results based on the basic deviation threshold, ensuring that most of the service data to be evaluated is not misjudged as abnormal.

[0084] In some implementations, when determining the warning result for the service data to be evaluated based on the basic deviation threshold corresponding to the service scenario and the service deviation score of the service data to be evaluated, the processing terminal can also determine the adjustment parameter of the basic deviation threshold based on the regional information in the user information. Here, the regional information in the user information can be used to indicate the user's location, and this regional information can be input by the user or automatically obtained by the processing terminal based on a location application. It is understood that the adjustment parameter of the basic deviation threshold can be a regional consumption index that corresponds to the regional information. For example, taking the medical service scenario as an example, if the cost of medical services in a first-tier city is greater than that in a second- or third-tier city, then the regional consumption index corresponding to a first-tier city is greater than that corresponding to a second- or third-tier city. In one example, the regional consumption index corresponding to a second- or third-tier city can be 1, the regional consumption index corresponding to a first-tier city can be 1.2, and the regional consumption index corresponding to a remote city can be 0.8.

[0085] In some embodiments of this disclosure, the adjustment parameters for the basic threshold of deviation also include a qualification level corresponding to the service provider. This service provider is used to indicate the name of the service company providing the service data to be evaluated, and can be obtained by user input or by the processing terminal based on electronic text data input by the user. It is understood that, taking a medical service scenario as an example, if the qualification of the service provider as a tertiary hospital is greater than that of the service provider as a secondary hospital, then the qualification level corresponding to the service provider as a tertiary hospital is greater than the qualification level corresponding to the service provider as a secondary hospital. In one example, the qualification level corresponding to the service provider as a tertiary hospital can be 1, the qualification level corresponding to the service provider as a secondary hospital can be 0.9, and the qualification level corresponding to the service provider as a secondary hospital can be 0.8.

[0086] Subsequently, the processing terminal can determine the warning result for the service data to be evaluated based on the adjustment parameters of the basic deviation threshold, the basic deviation threshold corresponding to the service scenario of the service data to be evaluated, and the service deviation score of the service data to be evaluated. In one example, the processing terminal can adjust the basic deviation threshold corresponding to the service scenario of the service data to be evaluated based on the adjustment parameters of the basic deviation threshold to obtain the deviation adjustment threshold, and determine the warning result for the service data to be evaluated based on this deviation adjustment threshold and the service deviation score of the service data to be evaluated. Here, taking the adjustment parameters of the basic deviation threshold including the regional consumption index and qualification level as an example, the method for determining the deviation adjustment threshold can be referred to the expression shown below: Deviation adjustment threshold = Basic deviation threshold × (1 + α × Regional consumption index + β × Service qualification) In the above expression, α and β can be understood as preset adjustment constants used to balance the impact of regional consumption index and service qualifications.

[0087] It is understandable that the processing terminal adjusts the threshold based on the deviation and the service deviation score of the service data to be evaluated, and determines the warning result of the service data to be evaluated in the manner described above, so it will not be elaborated on here.

[0088] In this way, the basic deviation threshold can be dynamically adjusted based on the adjustment parameters of the basic deviation threshold, so that the adjusted basic deviation threshold is compatible with the actual needs of different service scenarios (such as regional differences and reasonable premiums). This not only avoids the data of the service to be evaluated from being misjudged as abnormal, but also ensures the accuracy and reliability of the early warning results of the data of the service to be evaluated.

[0089] In some implementations, when determining the warning result of the service data to be evaluated based on the basic deviation threshold corresponding to the service scenario and the service deviation score of the service data to be evaluated, the processing terminal can also determine the deviating service nodes in the evaluation path of the service data to be evaluated based on the warning result. It is understood that the warning result of the service data to be evaluated includes the warning level of the service data to be evaluated and the warning details of the service nodes. The processing terminal can determine the corresponding service nodes in the evaluation path of the service data to be evaluated as deviating service nodes based on the warning details of the service nodes.

[0090] Subsequently, the processing terminal can mark the off-service nodes in the evaluation path of the service data to be evaluated and display the marked evaluation path. Here, the marking can be based on preset colors, such as yellow. In one example, the processing terminal can use a visualization library tool to process the evaluation path of the service data to be evaluated, generate a visualization map of the evaluation path, and mark the off-service nodes in the visualization map with yellow.

[0091] In some embodiments of this disclosure, when marking deviating service nodes in the evaluation path of the service data to be evaluated, the processing terminal can, after generating a visual map of the evaluation path, determine the warning level of each service node based on the node deviation score of each service node and the aforementioned threshold range, and mark the corresponding service nodes according to the marking method corresponding to different warning levels. For example, when the warning level of a service node is medium risk, the corresponding service node can be marked in yellow; when the warning level of a service node is high risk, the corresponding service node can be marked in red.

[0092] Of course, after generating a visual map of the path to be evaluated, the processing terminal can also annotate the corresponding service nodes based on the warning details of the service nodes. For example, when the warning details for an X-ray examination indicate that the necessary examination was not performed, a red standard warning can be applied to the X-ray examination; when the warning details for prescribing antibiotics indicate that the use of imported drugs has not been confirmed, a yellow warning can be applied to the prescribed antibiotics.

[0093] Understandably, when the processing terminal displays the evaluation path of the service data to be evaluated after the labeling process, it can label the visualization map of the evaluation path, integrate the labeled visualization map, the basic deviation threshold, the deviation score of the service data to be evaluated, and the warning results into a PDF / HTML report, and display the PDF / HTML report on the display interface of the third-party application used for service evaluation or the display interface of the mini-program used for service evaluation.

[0094] Figure 10 A schematic diagram of an interface showing the evaluation path for displaying service data to be evaluated, according to some embodiments of this disclosure, is shown. For example... Figure 10 As shown, the interface 1000 displaying the evaluation path of the service data to be evaluated shows a deviation score of 0.54, a basic deviation threshold of 0.2, a regional consumption index of 1.2, a qualification level of 0.9, a warning level of medium risk, and a labeled visualization. The labeled visualization includes a pet medical service path representing pet vomiting → X-ray examination → antibiotic prescription → cost 800. The X-ray examination is labeled with a rectangle and the warning detail is "necessary examination not performed." The antibiotic prescription is labeled with a rectangle and the warning detail is "imported drugs used without confirmation."

[0095] In this way, the early warning results of the service data to be evaluated can be visualized and presented to users. This not only makes it easier for users to understand the data of the service to be evaluated, but also provides users with interpretable decision-making basis, further improving service transparency and user trust.

[0096] Figure 11 Example block diagrams of a system for evaluating services according to some embodiments of this disclosure are shown. The various embodiments in this specification are described in a progressive manner, with reference to each other for similar or identical parts. Each embodiment focuses on describing the differences from other embodiments. In particular, the apparatus embodiments are relatively simple in description because they are substantially similar to the method embodiments; relevant details can be found in the descriptions of the method embodiments. Figure 11As shown, the system 1100 for evaluating services includes a data acquisition module 1102, configured to acquire service data to be evaluated. The service data to be evaluated includes corresponding user information and a service path to be evaluated. The service path to be evaluated includes one or more service nodes and one or more service edges, each with a corresponding pair of service nodes. The system 1100 also includes a path determination module 1104, configured to determine a standard service path corresponding to the service path to be evaluated from a standard service graph based on the service data to be evaluated. The standard service path includes one or more standard service nodes and one or more standard service edges, each with a corresponding pair of standard service nodes. The system 1100 also includes a score calculation module 1106, configured to determine the node deviation score of each service node in the service path to be evaluated based on the service path to be evaluated and the standard service path. Furthermore, the system 1100 also includes a score generation module 1108, configured to determine the service deviation score of the service data to be evaluated based on the node deviation scores of each service node in the service path to be evaluated.

[0097] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this specification are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in or transmitted through a computer-readable storage medium. The computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., coaxial cable, fiber optic, Digital Subscriber Line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any available medium accessible to a computer or a data storage device such as a server or data center that integrates one or more available media. The available media may be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., Digital Versatile Discs (DVDs)), or semiconductor media (e.g., Solid State Disks (SSDs)).

[0098] Figure 12 Block diagrams of electronic devices that can implement various embodiments of the present disclosure are shown. For example... Figure 12As shown, the electronic device 1200 includes a processor 1201, which can perform various appropriate actions and processes according to computer program instructions loaded into random access memory (RAM) 1203 based on computer program instructions stored in read-only memory (ROM) 1202. The RAM 1203 may also store various programs and data required for the operation of the electronic device 1200. The processor 1201, ROM 1202, and RAM 1203 are interconnected via a bus 1204. An input / output (I / O) interface 1205 is also connected to the bus 1204.

[0099] The various processes and procedures described above, such as method 300, can be executed by processor 1201. For example, in some embodiments, method 300 may be implemented as a software program tangibly contained in a machine-readable medium. In some embodiments, part or all of the software program may be loaded into and / or installed on electronic device 1200 via ROM 1202. When the software program is loaded into RAM 1203 and executed by processor 1201, one or more actions of method 300 described above may be performed.

[0100] The functions described above in this document can be performed at least in part by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: field programmable gate arrays (FPGAs), application-specific integrated circuits (ASICs), application-specific standard products (ASSPs), systems-on-a-chip (SoCs), payload programmable logic devices (CPLDs), and so on.

[0101] The program code used to implement the methods of this disclosure may be written in any combination of one or more programming languages. This program code may be provided to a processor or controller of a general-purpose computer, special-purpose computer, or other programmable data processing apparatus, such that when executed by the processor or controller, the program code causes the functions / operations specified in the flowcharts and / or block diagrams to be implemented. The program code may be executed entirely on a machine, partially on a machine, as a standalone software package partially on a machine and partially on a remote machine, or entirely on a remote machine or server.

[0102] This disclosure can be a method, apparatus, system, and / or program product. The program product may include a machine-readable storage medium on which machine-readable program instructions for performing various aspects of this disclosure are loaded. The machine-readable program instructions described herein can be downloaded from the machine-readable storage medium to various computing / processing devices, or downloaded via a network, such as the Internet, a local area network, a wide area network, and / or a wireless network, to an external computer or external storage device. The network may include copper transmission cables, fiber optic transmissions, wireless transmissions, routers, firewalls, switches, gateway computers, and / or edge servers. A network adapter card or network interface in each computing / processing device receives the machine-readable program instructions from the network and forwards them to the machine-readable storage medium in the respective computing / processing device.

[0103] Machine program instructions used to perform the operations of this disclosure may be assembly instructions, instruction set architecture (ISA) instructions, machine instructions, machine-dependent instructions, microcode, firmware instructions, state setting data, or source code or object code written in any combination of one or more programming languages, including object-oriented programming languages ​​such as Smalltalk, C++, etc., and conventional procedural programming languages ​​such as the "C" language or similar programming languages. Machine-readable program instructions may be executed entirely on a user's computer, partially on a user's computer, as a standalone software package, partially on a user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving a remote computer, the remote computer may be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or may be connected to an external computer (e.g., via the Internet using an Internet service provider). In some embodiments, electronic circuitry, such as programmable logic circuitry, field-programmable gate arrays (FPGAs), or programmable logic arrays (PLAs), is personalized by utilizing state information from the machine-readable program instructions to implement various aspects of this disclosure.

[0104] In the context of this disclosure, a machine-readable medium can be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device. A machine-readable medium can be a machine-readable signal medium or a machine-readable storage medium. Machine-readable media can be, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing. More specific examples of machine-readable storage media include electrical connections based on one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing. Furthermore, although operations are depicted in a specific order, this should be understood as requiring that such operations be performed in the specific order shown or in sequential order, or requiring that all illustrated operations be performed to achieve the desired result. In certain environments, multitasking and parallel processing may be advantageous. Similarly, while several specific implementation details are included in the foregoing discussion, these should not be construed as limiting the scope of this disclosure. Certain features described in the context of individual embodiments may also be implemented in combination in a single implementation. Conversely, various features described in the context of a single implementation may also be implemented individually or in any suitable sub-combination in multiple implementations.

[0105] Although the subject matter has been described using language specific to structural features and / or methodological logic, it should be understood that the subject matter defined in the appended claims is not necessarily limited to the specific features or actions described above. Rather, the specific features and actions described above are merely illustrative examples of implementing the claims.

Claims

1. Methods for evaluating services, including: Obtain service data to be evaluated, the service data to be evaluated having corresponding user information, the service data to be evaluated including the service path to be evaluated, the service path to be evaluated including one or more service nodes and one or more service edges, the service edge having corresponding service node pairs; Based on the service data to be evaluated, a standard service path corresponding to the service path to be evaluated is determined from the standard service graph. The standard service path includes one or more standard service nodes and one or more standard service edges. The standard service edge has a corresponding pair of standard service nodes. Based on the service path to be evaluated and the standard service path, determine the node deviation score of each service node in the service path to be evaluated; as well as Based on the node deviation score of each service node in the service path to be evaluated, the service deviation score of the service data to be evaluated is determined.

2. The method according to claim 1, wherein the service data to be evaluated further includes node multimodal data corresponding to the service node, and the node multimodal data includes modal data of at least two different modalities; and The step of determining the node deviation score for each service node in the service path to be evaluated, based on the service path to be evaluated and the standard service path, includes: The structural deviation score of each service node in the service path to be evaluated is determined based on the service path to be evaluated and the standard service path. Based on the node multimodal data corresponding to the service node and the preset verification rules, the data deviation score of the corresponding service node is determined; as well as Based on the structural deviation score and data deviation score of each service node, the node deviation score of the corresponding service node is determined.

3. The method according to claim 2, wherein determining the node deviation score of the corresponding service node based on the structural deviation score and data deviation score of each service node includes: Based on the structural deviation score, data deviation score, and preset data deviation correction coefficient of each service node, the node deviation score of the corresponding service node is determined.

4. The method according to claim 2, wherein determining the node deviation score of the corresponding service node based on the structural deviation score and data deviation score of each service node includes: Based on the node multimodal data corresponding to the service node, determine the feature vector corresponding to each modality in the node multimodal data; Based on the feature vectors corresponding to each modal data in the node multimodal data and the feature vectors of the corresponding service nodes, the adjustment modal weights corresponding to each modal data are determined; Based on the adjusted modal weights and preset modal weights corresponding to each modal data, the target data deviation correction coefficient is determined; as well as Based on the structural deviation score, data deviation score, and target data deviation correction coefficient of each service node, the node deviation score of the corresponding service node is determined.

5. The method according to claim 2, wherein determining the structural deviation score of each service node in the service path to be evaluated based on the service path to be evaluated and the standard service path includes: Based on the service path to be evaluated and the standard service path, determine whether each service node in the service path to be evaluated matches the standard service path; In response to determining that the service node matches the standard service path, the structural deviation score of the corresponding service node is determined as a first score; as well as In response to determining that the service node does not match the standard service path, the structural deviation score of the corresponding service node is determined as the second score.

6. The method according to claim 2, wherein determining the data deviation score of the corresponding service node based on the node multimodal data corresponding to the service node and preset verification rules includes: Based on the node multimodal data corresponding to the service node and the preset verification rules, it is determined whether there are data contradictions in the node multimodal data; In response to the determination that there is a data contradiction in the multimodal data of the node, the data deviation score of the corresponding service node is determined to be the third score; as well as In response to the determination that there is no data contradiction in the multimodal data of the node, the data deviation score of the corresponding service node is determined to be the fourth score.

7. The method according to claim 6, wherein the step of determining the data deviation score of the corresponding service node as a third score in response to determining that there is a data contradiction in the multimodal data of the node includes: In response to determining that there is a data contradiction in the node multimodal data, one or more contradictory modal data combinations are determined from the modal data of the node multimodal data; as well as The third score is determined based on the number of contradictory modal data combinations in the multimodal data of the node, and the third score is used as the data deviation score of the corresponding service node.

8. The method according to any one of claims 1-7, wherein the service data to be evaluated further includes service scenario information for indicating the service scenario of the service data to be evaluated, the service scenario having a corresponding standard service graph, and the method for constructing the standard service graph of the service scenario includes: Obtain a sample service dataset corresponding to the service scenario, wherein the sample service dataset includes multiple service data samples, and the service data samples include one or more modal data; The sample service dataset is standardized to obtain corresponding standard service data, and multiple standard service nodes corresponding to the service scenario are extracted from the standard service data. as well as Based on the standard service nodes in the standard service data and the preset rule base, a standard service graph is constructed; the preset rule base has multiple standard service edges, and each standard service edge has a corresponding pair of standard service nodes.

9. The method according to claim 8, wherein the service scenario has corresponding historical service data, and the method further includes: Based on the service scenario information of the service data to be evaluated, determine the historical service data corresponding to the service scenario of the service data to be evaluated; Based on historical service data corresponding to the service scenarios of the service data to be evaluated, a basic threshold for deviation from the service scenarios of the service data to be evaluated is determined. as well as Based on the basic threshold of deviation corresponding to the service scenario of the service data to be evaluated and the service deviation score of the service data to be evaluated, the early warning result of the service data to be evaluated is determined.

10. The method according to claim 9, wherein the user information includes regional information indicating the area where the user is located, and the step of determining the warning result of the service data to be evaluated based on the basic threshold of deviation corresponding to the service scenario of the service data to be evaluated and the service deviation score of the service data to be evaluated further includes: Based on the regional information in the user information, the adjustment parameters of the basic threshold for deviation are determined; as well as Based on the adjustment parameters of the deviation base threshold, the deviation base threshold corresponding to the service scenario of the service data to be evaluated, and the service deviation score of the service data to be evaluated, the early warning result of the service data to be evaluated is determined.

11. The method according to claim 9, wherein determining the early warning result of the service data to be evaluated based on the basic deviation threshold corresponding to the service scenario of the service data to be evaluated and the service deviation score of the service data to be evaluated further includes: Based on the early warning results of the service data to be evaluated, the off-service nodes in the evaluation path of the service data to be evaluated are determined; as well as The off-service nodes in the evaluation path of the service data to be evaluated are marked, and the evaluation path of the service data to be evaluated after the marking process is displayed.

12. The method according to claim 8, wherein determining the standard service path corresponding to the service path to be evaluated from the standard service map based on the service data to be evaluated further includes: Based on the service scenario information of the service data to be evaluated, a scenario service graph corresponding to the service scenario information of the service data to be evaluated is determined from a graph library including one or more scenario service graphs as a standard service graph. Each scenario service graph in the graph library has a corresponding service scenario.

13. Systems for evaluating information, including: The data acquisition module is configured to acquire service data to be evaluated. The service data to be evaluated has corresponding user information. The service data to be evaluated includes a service path to be evaluated. The service path to be evaluated includes one or more service nodes and one or more service edges. The service edge has a corresponding pair of service nodes. The path determination module is configured to determine, based on the service data to be evaluated, a standard service path corresponding to the service path to be evaluated from the standard service graph. The standard service path includes one or more standard service nodes and one or more standard service edges, and the standard service edge has a corresponding pair of standard service nodes. The scoring calculation module is configured to determine the node deviation score of each service node in the service path to be evaluated based on the service path to be evaluated and the standard service path. as well as The scoring generation module is configured to determine the service deviation score of the service data to be evaluated based on the node deviation score of each service node in the service path to be evaluated.

14. A computer-readable storage medium having a computer program stored thereon, the computer-readable storage medium storing instructions that, when executed on a computer or processor, cause the computer or processor to perform the steps of the method as claimed in any one of claims 1-12.

15. An electronic device comprising: One or more processors, and A memory associated with the one or more processors, the memory being used to store program instructions that, when read and executed by the one or more processors, perform the steps of the method according to any one of claims 1-12.

Images