Rule publishing method and system based on multi-stage time wheel

Abstract

The invention relates to the technical field of insurance data processing, and provides a rule publishing method and system based on a multi-stage time wheel. The method comprises the following steps: acquiring a first insurance case record; constructing an insurance database according to the first mapping relation in the record; extracting a target case, and obtaining target case features and target case delimits; adding the target case to a predetermined multi-stage time wheel according to the target case limit, and activating a processing instruction of the target case; according to the processing instruction, traversing the target case characteristics in an insurance database, and issuing a target case rule; and performing case regularization processing on the target case according to the target case rule. According to the method and the device, the technical problems of lagging in starting of a traditional insurance case rule, high system load and low case processing efficiency are solved, and the effects of efficiently starting and stopping the case rule in real time, reducing the system load and improving the case processing accuracy and circulation efficiency through a rule release mechanism based on a multi-stage time wheel are realized.

Description

Technical Field

[0001] The present application relates to the technical field of insurance data processing, and in particular to a rule publishing method and system based on a multi-level time wheel. Background Art

[0002] In the field of insurance and financial risk control, facing multiple insurance companies and multiple levels of institutions, the activation and deactivation of rules often rely on the functions of the rule expert operating system. However, when emergency cases (such as floods, mudslides, explosions, etc.) occur, the existing system often cannot activate the corresponding rules in time, resulting in delayed or erroneous rule triggering, which increases the difficulty of case handling. The case characteristics of different insurance institutions are different, and the enabled rules are also different. When multiple insurance companies enable rules at the same time, a large number of scheduled tasks will be generated, resulting in excessive system load and request response delays, which in turn affects the efficiency of case handling. The existing technology uses a delay queue to handle delayed activation of rules, but when the task delay is unevenly distributed or the delay level supported by the delay queue is limited, it will face problems such as thread idling and low efficiency, and it is difficult to meet the needs of large-scale real-time rule scheduling.

[0003] Based on this problem, the present invention proposes a rule publishing method and system based on a multi-level time wheel. By introducing the multi-level time wheel technology, a large number of rule activation tasks can be efficiently scheduled with a small amount of thread resources, which significantly reduces the system load, improves the accuracy of rule triggering and case flow efficiency, realizes real-time rule-based processing of insurance cases, and effectively improves the system concurrency capability and overall performance. Summary of the invention

[0004] The present application provides a rule publishing method and system based on a multi-level time wheel, aiming to solve the technical problems of delayed activation of traditional insurance case rules, high system load and low case processing efficiency.

[0005] In view of the above problems, the present application provides a rule publishing method and system based on a multi-level time wheel.

[0006] The first aspect disclosed in the present application provides a rule publishing method based on a multi-level time wheel, the method comprising: obtaining a first insurance case record, the first insurance case record refers to the record information of any historical insurance case of any insurance platform among the multiple insurance platforms; constructing an insurance database according to a first mapping relationship of the first insurance case in the first insurance case record, wherein the first mapping relationship refers to a mapping relationship between a first case feature and a first case rule; extracting a target case from a list of pending cases, and obtaining a target feature of the target case, the target feature including a target case feature and a target case deadline of the target case; adding the target case to a predetermined multi-level time wheel according to the target case deadline, and activating a processing instruction of the target case through the operation of the predetermined multi-level time wheel; according to the processing instruction, traversing the target case feature in the insurance database, and publishing a target case rule; and performing case rule-based processing on the target case according to the target case rule.

[0007] Another aspect disclosed in the present application provides a rule publishing system based on a multi-level time wheel, the system comprising: a case record acquisition module, the case record acquisition module is used to acquire a first insurance case record, the first insurance case record refers to the record information of any historical insurance case of any insurance platform in the multiple insurance platforms; a database construction module, the database construction module is used to construct an insurance database according to a first mapping relationship of the first insurance case in the first insurance case record, wherein the first mapping relationship refers to the mapping relationship between the first case feature and the first case rule; a case feature acquisition module, the case feature acquisition module is used to extract from the list of pending cases A target case, and obtains the target features of the target case, wherein the target features include the target case features and the target case deadline of the target case; a processing instruction activation module, wherein the processing instruction activation module is used to add the target case to a predetermined multi-level time wheel according to the target case deadline, and activate the processing instructions of the target case through the operation of the predetermined multi-level time wheel; a case rule publishing module, wherein the case rule publishing module is used to traverse the target case features in the insurance database according to the processing instructions, and publish target case rules; a rule-based processing module, wherein the rule-based processing module is used to perform case rule-based processing on the target case according to the target case rules.

[0008] One or more technical solutions provided in this application have at least the following technical effects or advantages: The above-mentioned rule publishing method based on a multi-level time wheel first obtains the record information of historical insurance cases. These record information comes from case data of multiple insurance platforms, which provides a basis for subsequent rule publishing. Subsequently, an insurance database is constructed based on the mapping relationship between case features extracted from case records and rules. Through this mapping, cases can be quickly associated with applicable rules to improve the efficiency of rule matching. When processing a target case, the target case is extracted from the list of pending cases and its features are obtained, including case details and processing deadlines. Then, according to the deadline of the target case, the case is added to the predetermined multi-level time wheel, and the processing instructions of the case are activated through the efficient operation of the time wheel. The introduction of the multi-level time wheel effectively solves the problem of delayed processing of a large number of tasks, uses a small amount of thread resources to achieve accurate scheduling of large batches of tasks, reduces system load, and improves the concurrent processing capability of the system. According to the processing instructions, the features related to the target case in the insurance database are traversed, and then the rules applicable to the case are published. Through this traversal, the accuracy and timeliness of the rules are ensured, and delays or errors in rule publishing are avoided. Afterwards, the case is standardized according to the published target case rules. The entire process uses a multi-level time wheel scheduling strategy to achieve real-time rule publishing and precise triggering, significantly improving the system's processing efficiency and accuracy, while reducing case processing response delays and system burdens, thereby improving the overall efficiency of case circulation.

[0009] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0011] Figure 1 The figure is a flowchart of a rule publishing method based on a multi-level time wheel in an embodiment.

[0012] Figure 2 The following is an architecture diagram of a rule publishing system based on a multi-level time wheel in one embodiment.

[0013] Explanation of the accompanying drawings: case record acquisition module 1, database construction module 2, case feature acquisition module 3, processing instruction activation module 4, case rule publishing module 5, rule-based processing module 6. DETAILED DESCRIPTION

[0014] The embodiments of the present application provide a rule publishing method and system based on a multi-level time wheel to solve the technical problems of delayed activation of traditional insurance case rules, high system load and low case processing efficiency.

[0015] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without making creative work are within the scope of protection of this application.

[0016] It should be noted that the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions. For example, a process, method, system, product or server that includes a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or modules that are not explicitly listed or inherent to these processes, methods, products or devices.

[0017] Embodiment 1, as Figure 1 As shown, the present application provides a rule publishing method based on a multi-level time wheel, the method comprising: A first insurance case record is obtained, where the first insurance case record refers to record information of any historical insurance case of any insurance platform among the multiple insurance platforms.

[0018] The rule publishing method based on multi-level time wheel is applied to a rule publishing system that is connected to multiple insurance platforms. The system can receive insurance case data from multiple different insurance platforms by establishing communication connections with these platforms, and process these data using multi-level time wheel technology.

[0019] In the embodiment of the present application, the system terminal obtains the record information of a historical insurance case from multiple insurance platforms. These records may come from any insurance platform and cover the details of the insurance case, including case characteristics, processing procedures, and applicable rules, etc., providing basic data for subsequent rule scheduling. By obtaining the records of these historical cases, information can be shared between different insurance platforms to ensure the comprehensiveness and accuracy of rule matching, thereby realizing accurate rule publishing and processing for future cases.

[0020] An insurance database is constructed according to a first mapping relationship of the first insurance case in the first insurance case record, wherein the first mapping relationship refers to a mapping relationship between a first case feature and a first case rule.

[0021] In one embodiment, the system terminal constructs an insurance database according to the first mapping relationship of the first insurance case in the first insurance case record. The first mapping relationship refers to the association between the characteristics of the first insurance case (such as the time, location, type of accident, etc.) and the corresponding rules. By using these existing mapping relationships, relevant case characteristics and rules can be stored in the insurance database, so that when handling new cases, the rules can be quickly matched and applied according to the mapping relationships in the database, thereby improving the efficiency and accuracy of case handling.

[0022] A target case is extracted from the list of pending cases, and target features of the target case are obtained, wherein the target features include target case features and target case deadlines of the target case.

[0023] In one embodiment, the system terminal extracts the target case from the list of pending cases and obtains the target features of the case. The target features include the case features of the target case and the deadline of the target case (i.e., the time limit for case processing). By extracting these target features, the case features can be analyzed in real time, and based on these features, the applicable rules can be matched to ensure that the time requirements can be met during the processing process, and the accuracy and timeliness of the regularized case processing can be guaranteed.

[0024] The target case is added to a predetermined multi-level time wheel according to the target case deadline, and the processing instruction of the target case is activated by running the predetermined multi-level time wheel.

[0025] In one embodiment, the system terminal adds the target case to a predetermined multi-level time wheel according to the deadline of the target case. The multi-level time wheel is a ring structure similar to a clock, which is divided into multiple slots, each slot represents a time period, for example, it is divided into 8 slots, each slot represents 1s. The system terminal calculates the appropriate slot position according to the deadline of the target case and puts the case into the corresponding slot. For example, assuming that the target case needs to be executed after 12 seconds, the case will be placed in the slot of the current hour hand position plus 12 seconds through calculation. The multi-level time wheel manages the delayed execution of tasks through its round mechanism. When the lower wheel rotates one circle, the higher wheel rotates one grid, which is similar to the relationship between the second hand and the minute hand. The system terminal determines the execution time of the target case according to the round value (round information) of the target case. If round is 0, the target case will be executed immediately when the hour hand reaches the slot; if round is greater than 0, the target case will be executed after multiple rounds. During the operation of the time wheel, the round value will decrease with each round until it becomes 0, and the processing instructions corresponding to the target case will be activated to execute the processing steps for the target case. Through this multi-level time wheel design, a large number of real-time and delayed tasks can be efficiently scheduled, avoiding the problem of a large number of threads idling, and greatly reducing the system load. At the same time, with the hierarchical design of the multi-level time wheel, even when facing hundreds of thousands or even millions of tasks, it can still ensure high execution efficiency and ensure that the target case is processed in time within its deadline.

[0026] Furthermore, the present application includes: Extract the first-level deadline and the second-level deadline in the deadline of the target case in sequence; add the first-level deadline and the second-level deadline to the predetermined multi-level time wheel in sequence according to the time wheel scheduling mechanism to obtain a scheduling result; and activate processing instructions for each case in the list of pending cases in sequence according to the scheduling result.

[0027] Preferably, the system terminal extracts the first-level deadline and the second-level deadline from the target case deadline of the target case in sequence. The first-level deadline represents the task that needs to be handled with priority, such as the preliminary review of the urgent case; the second-level deadline represents the task that needs to be handled in a longer period of time, such as the comprehensive evaluation or case closure of the case. The two deadlines represent different time priorities. Subsequently, according to the scheduling mechanism of the multi-level time wheel, the first-level deadline and the second-level deadline are added to the corresponding slots of the multi-level time wheel respectively. After calculating the specific delay time of the first-level deadline and the second-level deadline, these tasks will be assigned to the appropriate time slots. For example, if the time wheel is divided into 8 slots, each slot represents 1 second, and the hour hand of the current time wheel points to the second slot. For the task with the first-level deadline, assuming that it is required to be executed after 3 seconds, the execution position of the task will be calculated, that is, the current hour hand is in the second slot, plus the task execution time after 3 seconds, it is concluded that the task needs to be assigned to the fifth slot, and the task is added to the fifth slot of the time wheel at this time, and the task is executed when the hour hand turns to this position. For the second-level deadline, assuming that the task needs to be executed after 12 seconds, it will be calculated according to the multi-level scheduling mechanism of the time wheel. Since the current time wheel has 8 slots, 12 seconds has exceeded one round and multiple rounds of scheduling are required. Through calculation, 12 seconds divided by 8 seconds is one round, and the remainder is 4, so the task needs to be added to the 6th slot in the next round. In order to indicate that the task needs to wait for one round before execution, the round value is set to 1, indicating that the task will be executed after one round. When the clock moves to the 6th slot, the round value is checked. If the round value is 1, it will not be executed immediately, but wait until the round value is reduced to 0 before executing the task. Through this mechanism, real-time and delayed tasks can be effectively managed. As the multi-level time wheel runs, the system terminal generates the order results. The order results are based on the order of the time wheel slots to ensure that the first-level deadline tasks are processed first, and then the second-level deadline tasks are processed in chronological order. The multi-level structure of the time wheel ensures that after the rotation of the first-level time wheel is completed, the tasks of the higher-level time wheel will be reallocated to the lower-level time wheel in a cascade manner, avoiding task backlogs and ensuring system load balancing. Afterwards, according to the order scheduling results, the processing instructions are activated for the cases in the list of cases to be processed in turn. When the time wheel runs to the slot corresponding to the first-level deadline, the processing instructions related to it are activated; when the time wheel further rotates to the slot corresponding to the second-level deadline, the instructions related to it are activated. This order scheduling and dispatching mechanism can ensure that tasks of different priorities are executed on time, reduce system load, and improve processing efficiency.

[0028] Furthermore, this application provides target case deadlines, including: The target multi-level deadline time for the target case deadline is obtained, and the target multi-level deadline time includes grade time, month time, day time, hour time, classification time and second time.

[0029] Optionally, when scheduling, the system terminal will obtain multi-level deadlines for the target case, which are progressive, including grade time, month time, day time, hour time, grade time, and second time. Through this hierarchical structure, tasks of different priorities can be better managed. For example, grade time and month time correspond to long-term tasks, while second and grade time correspond to tasks that require quick response. By processing each level of time, the target case is assigned to the corresponding time wheel slot. In the time wheel mechanism, different time levels are equivalent to multiple rounds of operation. When processing shorter tasks, such as second level, it will be arranged in the closer slot of the time wheel. For longer time tasks at grade and month level, multiple rounds of scheduling will be used to ensure that these tasks will not be executed immediately, but will be processed after multiple rounds of operation until the appropriate time node. Through this refined multi-level time management, a large number of delayed and real-time tasks can be efficiently scheduled while ensuring time accuracy to avoid system overload.

[0030] According to the processing instruction, the target case characteristics are traversed in the insurance database, and the target case rules are issued.

[0031] In one embodiment, after activating the processing instruction, the system terminal traverses the case features of the target case in the insurance database according to the processing instruction. Through the traversal, these features will be analyzed and fitted, and compared with the historical case features in the insurance database to find scenes that match the target case features. This process is not just a simple feature search, but also includes the calculation of the matching degree of the features to ensure that the historical case that best matches the current case situation is found. After the matching is completed, the target case rules that meet the matching degree threshold with the target case will be automatically released. The released target case rules are selected based on the mapping relationship between the case features and the case rules in the insurance database to ensure accurate matching and real-time triggering of the rules, so that the system terminal can timely and accurately process the cases in a regularized manner, reduce the need for human intervention, and improve the efficiency and accuracy of case processing.

[0032] Furthermore, this application provides rules for publishing target cases, including: Obtain any case in the insurance database, wherein the arbitrary case has an arbitrary mapping relationship, wherein the arbitrary mapping relationship refers to a mapping relationship between arbitrary case features and arbitrary case rules; introduce a feature vectorization strategy to vectorize the arbitrary case features and the target case features in turn, and obtain an arbitrary case vector and a target case vector respectively; compare the arbitrary case vector with the target case vector to obtain an arbitrary matching degree; determine whether the arbitrary matching degree reaches a matching degree threshold; if so, use the arbitrary case rule as the target case rule of the target case and publish it.

[0033] Preferably, the system terminal obtains any case from the insurance database, and these cases associate case features with case rules through any mapping relationship. The features of each case, such as the time of accident, the location of accident, the cause of accident, etc., are associated with their corresponding rules through mapping relationships. Subsequently, a feature vectorization strategy is introduced to vectorize the features of any case and the features of the target case. Specifically, each case feature (such as the time of accident, the location of accident, etc.) is converted into a numerical feature vector representation. In this way, any case generates an arbitrary case vector, and the target case generates a target case vector. Subsequently, the arbitrary case vector is compared with the target case vector, and the similarity between the two is calculated by a predetermined matching degree calculation function to obtain an arbitrary matching degree. This matching degree measures the similarity between the features of the two cases. The higher the matching degree, the closer the case features are. Afterwards, it is determined whether the arbitrary matching degree reaches a preset matching degree threshold value. If the matching degree reaches or exceeds the threshold, the system terminal uses the rule of the arbitrary case as the rule of the target case and publishes the determined target case rule. Through the above process, the system terminal converts the case features into vectors and matches the rules according to the similarity of these vectors. This mechanism of feature vectorization and matching degree calculation enables the system to efficiently and accurately find the most suitable rules for the target case and significantly improve the degree of automation in case processing.

[0034] Furthermore, the present application provides a feature vectorization strategy, including: The feature quantization strategy is embedded with feature attributes, and the feature attributes include the time of the accident, the location of the accident and the cause of the accident. The cause of the accident includes the responsible party and the damaged parts with identification of the degree of damage.

[0035] Optionally, the feature vectorization strategy embeds specific feature attributes, which include the time, location and cause of the accident. Among them, the feature attributes of the cause of the accident are further refined into the responsible party and the damaged parts with damage degree identification. The system terminal quantifies the basic attributes of the case such as time and location, and then identifies the responsible party as an important feature of the case responsibility analysis. At the same time, for the damaged parts, this information will be quantified as part of the feature vector by identifying its damage degree. This multi-level and multi-dimensional feature vectorization processing can capture the key features of the case in more detail, providing a more accurate basis for subsequent case matching and rule publishing.

[0036] Further, the present application provides obtaining the arbitrary matching degree, including: A predetermined labeling scheme is obtained according to the feature vector quantization strategy; the arbitrary case features and the target case features are labeled in turn according to the predetermined labeling scheme to obtain the arbitrary case vector and the target case vector respectively; a predetermined matching degree calculation function is obtained according to the feature vector quantization strategy; and a comparison calculation is performed on the arbitrary case vector and the target case vector according to the predetermined matching degree calculation function to obtain the arbitrary matching degree.

[0037] Optionally, the system terminal obtains a predetermined label scheme according to the feature vectorization strategy. The label scheme defines how to label case features and completes the vectorization of case features. For the time of the accident, the time of the accident is composed of elements such as year, month, day, hour, minute, and second. When vectorizing, these time elements will be split into independent values. For example, if the time of the accident is 15:30:10 on October 21, 2011, it will be vectorized as [2011,10,21,15,30,10], so that each time point can be accurately represented. For the location of the accident, vectorization processing will be performed based on the geographical location information. The system terminal can use longitude and latitude or hierarchical location description for encoding. For example, if geocoding is used, the longitude and latitude of the location of the accident will be recorded and directly converted into a vector. For the damaged part, it will be based on the specific situation of the damaged parts and the degree of damage. For example, if the front of the vehicle is slightly damaged, the front of the vehicle will be coded as [1], the slight damage will be coded as [1], and the final vectorization result of the damaged part will be [1,1]. If the rear of the vehicle is severely damaged, the vectorization result will be [2,3], indicating that the rear is severely damaged. For the responsible party, vectorization will be performed according to different types of responsible parties and their responsibility ratios. Assuming that the responsible party is an individual and bears 100% of the responsibility, the individual can be coded as [1], the responsibility ratio is [1.0], and the final responsibility direction is [1,1.0]. If the responsible party is a company and bears 50% of the responsibility, the vectorization result will be [2,0.5], indicating that the company bears part of the responsibility. Through these specific vectorization strategies, the system can convert various features of the case into numerical vectors to form a complete case vector representation. For example, a complete feature vector of a case may be: time of accident [2024, 10, 21, 15, 30, 10], location of accident [39.9042, 116.4074], cause of accident [1, 1], location of injury [1, 1], responsible party [1, 1.0]. Through this series of feature vectorization processing, the system terminal can structure the complex information of the case, which is convenient for subsequent matching calculations. Subsequently, the system terminal labels the arbitrary case features and the target case features in turn according to the predetermined label scheme obtained by matching. Through this process, the features of the arbitrary case and the target case are converted into their respective vector representations, that is, the feature vectors corresponding to the case features of the same case are spliced ​​head to tail according to the order of the case features, and finally the arbitrary case vector and the target case vector are obtained. These vectorized feature data contain the core information of each case, which is convenient for the subsequent matching process. Afterwards, a predetermined matching degree calculation function is obtained according to the feature vectorization strategy. The calculation function is used to compare the similarity between two case vectors and is set based on the cosine similarity algorithm. The matching calculation function can quantitatively evaluate the similarity between any case and the target case.Then, the system terminal uses this matching degree calculation function to compare and calculate the arbitrary case vector and the target case vector. By calculating the difference between the arbitrary case vector and the target case vector, the arbitrary matching degree is obtained, which indicates the degree of similarity between the two cases in terms of features. The higher the matching degree, the more similar the features of the two cases are, and based on this, it can be determined whether the arbitrary case rule is applicable to the target case. Through this process, the system terminal organically combines the labeling, vectorization and matching degree calculation of case features, achieving accurate case matching and rule recommendation.

[0038] The target case is processed in a case-rule-based manner according to the target case rule.

[0039] In one embodiment, after determining the rules applicable to the target case, the system terminal will perform regular processing on the target case according to these rules. This process includes applying the processing logic and case processing steps in the rules to the target case, as well as the regular processing effect of the case. The system terminal will analyze the characteristics of the target case and match these characteristics with the published rules to ensure that each step meets the requirements of the rules. In addition, the system terminal can schedule and process the case according to the time requirements and priorities in the rules. Through the multi-level time wheel scheduling mechanism, it can be ensured that each processing step is executed at the appropriate time node, avoiding processing delays and reducing the system load. Regular processing includes but is not limited to automatically triggering tasks such as compensation calculation and responsibility division to ensure that the case is carried out according to the predetermined process and finally outputs processing results that meet the standards.

[0040] Furthermore, the present application provides regularized processing effects, including: Obtain a regularized processing record for each case in the list of pending cases, wherein the regularized processing record includes a first processing record of a first case and a second processing record of a second case; take the first processing time in the first processing record and the second processing time in the second processing record to obtain an average case processing time; take the first concurrent processing number in the first processing record and the second concurrent processing number in the second processing record to obtain an average concurrent processing number; normalize and weight the average case processing time and the average concurrent processing number to obtain processing fitness, wherein the processing fitness is used to characterize the regularized processing effect of each case in the list of pending cases.

[0041] Preferably, the system terminal obtains the regular processing record of each case from the list of pending cases. The processing record of each case includes its corresponding processing time and concurrent processing number. For example, the record of the first case includes the first processing time and the first concurrent processing number, while the record of the second case includes the second processing time and the second concurrent processing number. Subsequently, the average processing time of the two cases is calculated respectively. The average case processing time is obtained by taking the sum of the processing time of the first case and the second case and dividing it by the number of cases. The average case processing time reflects the flow efficiency of the case in the system and the adaptability of the system under different conditions, that is, the case fitness. A shorter processing time means that the system is more efficient and adaptable in regular processing. At the same time, the average concurrent processing number is also calculated. The average concurrent processing number is obtained by taking the sum of the concurrent processing number of the first case and the second case and dividing it by the number of cases. The average concurrent processing number indicates the load capacity and adaptability of the system when processing multiple tasks, that is, the system fitness. A higher number of concurrent processing means that the system can process more cases at the same time and improve the concurrent load capacity. After that, the average case processing time and the average number of concurrent processing are normalized and weighted. The normalization step is to convert the processing time and the number of concurrent processing into the same magnitude through the maximum and minimum method so that weighted sum can be performed. The weighting process combines the case processing efficiency and the system concurrent load capacity according to the predetermined weight value to obtain a comprehensive processing fitness. Among them, the weight value is determined based on specific business needs. For some emergency case processing scenarios (such as natural disaster emergency processing in insurance claims), processing timeliness is more important, so the weight will be tilted towards the case processing time. In the scenario of multi-task processing or batch case processing, concurrent capacity is more important, and the weight of the number of concurrent processing will be greater. The calculated processing fitness is used to measure the regular processing effect of each case in the list of pending cases. Through this fitness, it is possible to judge whether the current processing strategy meets expectations and make adjustments when necessary to ensure that the system can still maintain efficient case processing capabilities under high load conditions.

[0042] In summary, the embodiments of the present application have at least the following technical effects: The embodiment of the present application obtains insurance case records and constructs an insurance database based on the mapping relationship between case features and rules. After extracting the target case from the list of pending cases, it is added to the multi-level time wheel according to the deadline of the target case, and the processing instructions of the case are activated by the operation of the time wheel. Subsequently, the target case features in the insurance database are traversed, applicable rules are issued, and the target case is processed in a regularized manner. The method further includes extracting the multi-level deadlines (such as grade, month, day, hour, grade and second) of the target case, and scheduling the case according to the scheduling mechanism of the time wheel. In addition, a feature vectorization strategy is introduced to convert the features of any case and the target case into vectors, and then the matching degree is calculated to determine whether the rules of any case are applicable to the target case. Finally, according to the processing records, the average case processing time and the average number of concurrent processing are calculated, and the processing fitness is obtained by normalization weighting to evaluate the regular processing effect of the case. These technical effects jointly solve the technical problems of delayed activation of traditional insurance case rules, high system load and low case processing efficiency. They achieve the effect of real-time and efficient activation and deactivation of case rules through a rule publishing mechanism based on a multi-level time wheel, reducing system load and improving the accuracy and circulation efficiency of case processing.

[0043] Embodiment 2 is based on the same inventive concept as the rule publishing method based on a multi-level time wheel in the above embodiment. Figure 2 As shown, the present application provides a rule publishing system based on a multi-level time wheel, and the system includes: a case record acquisition module 1, the case record acquisition module 1 is used to acquire a first insurance case record, and the first insurance case record refers to the record information of any historical insurance case of any insurance platform in the multiple insurance platforms; a database construction module 2, the database construction module 2 is used to construct an insurance database according to a first mapping relationship of the first insurance case in the first insurance case record, wherein the first mapping relationship refers to the mapping relationship between the first case feature and the first case rule; a case feature acquisition module 3, the case feature acquisition module 3 is used to extract a target case from a list of cases to be processed The target case is obtained by: obtaining the target features of the target case, wherein the target features include the target case features and the target case deadline of the target case; a processing instruction activation module 4, wherein the processing instruction activation module 4 is used to add the target case to a predetermined multi-level time wheel according to the target case deadline, and activate the processing instructions of the target case through the operation of the predetermined multi-level time wheel; a case rule publishing module 5, wherein the case rule publishing module 5 is used to traverse the target case features in the insurance database according to the processing instructions, and publish target case rules; a rule-based processing module 6, wherein the rule-based processing module 6 is used to perform case rule-based processing on the target case according to the target case rules.

[0044] Furthermore, the processing instruction activation module 4 is also used to execute the following method: Extract the first-level deadline and the second-level deadline in the deadline of the target case in sequence; add the first-level deadline and the second-level deadline to the predetermined multi-level time wheel in sequence according to the time wheel scheduling mechanism to obtain a scheduling result; and activate processing instructions for each case in the list of pending cases in sequence according to the scheduling result.

[0045] Furthermore, the processing instruction activation module 4 is also used to execute the following method: The target multi-level deadline time for the target case deadline is obtained, and the target multi-level deadline time includes grade time, month time, day time, hour time, classification time and second time.

[0046] Furthermore, the case rule publishing module 5 is also used to execute the following method: Obtain any case in the insurance database, wherein the arbitrary case has an arbitrary mapping relationship, wherein the arbitrary mapping relationship refers to a mapping relationship between arbitrary case features and arbitrary case rules; introduce a feature vectorization strategy to vectorize the arbitrary case features and the target case features in turn, and obtain an arbitrary case vector and a target case vector respectively; compare the arbitrary case vector with the target case vector to obtain an arbitrary matching degree; determine whether the arbitrary matching degree reaches a matching degree threshold; if so, use the arbitrary case rule as the target case rule of the target case and publish it.

[0047] Furthermore, the case rule publishing module 5 is also used to execute the following method: The feature quantization strategy is embedded with feature attributes, and the feature attributes include the time of the accident, the location of the accident and the cause of the accident. The cause of the accident includes the responsible party and the damaged parts with identification of the degree of damage.

[0048] Furthermore, the case rule publishing module 5 is also used to execute the following method: A predetermined labeling scheme is obtained according to the feature vector quantization strategy; the arbitrary case features and the target case features are labeled in turn according to the predetermined labeling scheme to obtain the arbitrary case vector and the target case vector respectively; a predetermined matching degree calculation function is obtained according to the feature vector quantization strategy; and a comparison calculation is performed on the arbitrary case vector and the target case vector according to the predetermined matching degree calculation function to obtain the arbitrary matching degree.

[0049] Furthermore, the regularization processing module 6 is also used to execute the following method: Obtain a regularized processing record for each case in the list of pending cases, wherein the regularized processing record includes a first processing record of a first case and a second processing record of a second case; take the first processing time in the first processing record and the second processing time in the second processing record to obtain an average case processing time; take the first concurrent processing number in the first processing record and the second concurrent processing number in the second processing record to obtain an average concurrent processing number; normalize and weight the average case processing time and the average concurrent processing number to obtain processing fitness, wherein the processing fitness is used to characterize the regularized processing effect of each case in the list of pending cases.

[0050] It should be noted that the above-mentioned sequence of the embodiments of the present application is only for description and does not represent the advantages and disadvantages of the embodiments. And the above-mentioned specific embodiments of this specification are described. The processes depicted in the accompanying drawings do not necessarily require the specific order and continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0051] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

[0052] This specification and drawings are merely exemplary illustrations of the present application and are deemed to cover any and all modifications, variations, combinations or equivalents within the scope of the present application. Obviously, a person skilled in the art may make various modifications and variations to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalents, the present application intends to include these modifications and variations.

Claims

1. A rule publishing method based on a multi-level time wheel, characterized in that: The rule issuing method based on the multi-level time wheel is applied to a rule issuing system based on the multi-level time wheel, the rule issuing system based on the multi-level time wheel is connected to multiple insurance platforms in communication, and the rule issuing method based on the multi-level time wheel includes: Acquire a first insurance case record, where the first insurance case record refers to record information of any historical insurance case of any insurance platform among the multiple insurance platforms; Building an insurance database according to a first mapping relationship of the first insurance case in the first insurance case record, wherein the first mapping relationship refers to a mapping relationship between a first case feature and a first case rule; Extracting a target case from the list of pending cases, and obtaining target features of the target case, wherein the target features include target case features and target case deadlines of the target case; Adding the target case to a predetermined multi-level time wheel according to the target case deadline, and activating the processing instruction of the target case through the operation of the predetermined multi-level time wheel; According to the processing instruction, traverse the target case characteristics in the insurance database and publish the target case rules; The target case is processed in a case-rule-based manner according to the target case rule.

2. A rule publishing method based on a multi-level time wheel according to claim 1, characterized in that: include: Sequentially extract the first-level deadline and the second-level deadline in the deadline of the target case; According to the time wheel scheduling mechanism, the first-level deadline and the second-level deadline are sequentially added to the predetermined multi-level time wheel to obtain a scheduling result; According to the order arrangement result, the processing instructions are activated for each case in the list of pending cases in turn.

3. A rule publishing method based on a multi-level time wheel according to claim 2, characterized in that: The target multi-level deadline time for the target case deadline is obtained, and the target multi-level deadline time includes grade time, month time, day time, hour time, classification time and second time.

4. The rule publishing method based on a multi-level time wheel according to claim 1, characterized in that: include: Obtain any case in the insurance database, wherein the any case has any mapping relationship, wherein the any mapping relationship refers to a mapping relationship between any case feature and any case rule; A feature vectorization strategy is introduced to sequentially vectorize the arbitrary case features and the target case features to obtain an arbitrary case vector and a target case vector respectively; Comparing the arbitrary case vector with the target case vector to obtain an arbitrary matching degree; Determining whether any matching degree reaches a matching degree threshold; If achieved, the arbitrary case rule will be used as the target case rule of the target case and published.

5. A rule publishing method based on a multi-level time wheel according to claim 4, characterized in that: The feature quantization strategy is embedded with feature attributes, and the feature attributes include the time of the accident, the location of the accident and the cause of the accident. The cause of the accident includes the responsible party and the damaged parts with identification of the degree of damage.

6. A rule publishing method based on a multi-level time wheel according to claim 4, characterized in that: include: Acquire a predetermined labeling scheme according to the feature vectorization strategy; Label the arbitrary case features and the target case features in sequence according to the predetermined labeling scheme to obtain the arbitrary case vector and the target case vector respectively; Acquire a predetermined matching degree calculation function according to the feature vectorization strategy; The arbitrary case vector and the target case vector are compared and calculated according to the predetermined matching degree calculation function to obtain the arbitrary matching degree.

7. A rule publishing method based on a multi-level time wheel according to claim 1, characterized in that: Also includes: Obtaining a regularized processing record of each case in the list of pending cases, wherein the regularized processing record includes a first processing record of a first case and a second processing record of a second case; Taking the first processing time in the first processing record and the second processing time in the second processing record to obtain an average case processing time; Taking the first concurrent processing quantity in the first processing record and the second concurrent processing quantity in the second processing record to obtain the average concurrent processing quantity; The processing fitness is obtained by normalizing and weighting the average case processing time and the average number of concurrent processing, wherein the processing fitness is used to characterize the regular processing effect of each case in the list of cases to be processed.

8. A rule publishing system based on a multi-level time wheel, characterized in that: The steps for implementing the rule publishing method based on a multi-level time wheel as described in any one of claims 1 to 7 include: A case record acquisition module: acquiring a first insurance case record, wherein the first insurance case record refers to record information of any historical insurance case of any insurance platform among the multiple insurance platforms; Database construction module: constructing an insurance database according to a first mapping relationship of the first insurance case in the first insurance case record, wherein the first mapping relationship refers to a mapping relationship between a first case feature and a first case rule; Case feature acquisition module: extracts a target case from the list of pending cases and acquires target features of the target case, wherein the target features include target case features and target case deadlines of the target case; A processing instruction activation module: adding the target case to a predetermined multi-level time wheel according to the target case deadline, and activating the processing instruction of the target case through the operation of the predetermined multi-level time wheel; Case rule publishing module: according to the processing instruction, traverse the target case features in the insurance database and publish the target case rules; Rule-based processing module: performs case-based processing on the target case according to the target case rules.

Images