Personnel work influence evaluation method, system and device and storage medium

Abstract

The invention provides a personnel work influence evaluation method and device, equipment and a storage medium, and the method comprises the steps: extracting an influence event containing a personnel main body, a behavior verb and a business object from a work order text; determining a reference weight of a semantic influence factor according to the behavior verb; determining an enhancement factor of a business context according to the business object; obtaining an extraction confidence coefficient of the influence event; and according to the reference weight, the enhancement factor and the extraction confidence, calculating a score of the influence event so as to quantify the work influence of the personnel subject. By adopting the method, the objectivity and the pertinence of the evaluation basis are ensured, and the accuracy and the fairness of the evaluation result are realized.

Description

Technical Field

[0001] This application relates to the field of computer technology, and more specifically, to a method, apparatus, device, and storage medium for evaluating the impact of personnel work. Background Technology

[0002] In the field of InsurTech, operations and maintenance (O&M) personnel are a key force in ensuring the stable operation of core systems, preventing business risks, and supporting cross-departmental collaboration. Their work directly impacts the continuity of insurance business, fund security, and regulatory compliance. With the increasing digitalization of business operations, O&M platforms have accumulated massive amounts of work order response texts containing root cause analysis, risk warnings, and technical suggestions. The implicit technical contributions contained within this unstructured data have become an important potential basis for evaluating the value of O&M personnel's work.

[0003] In the current insurance industry, human evaluation is an important existing technology for assessing the work impact of operations and maintenance personnel. Specifically, it is a 360-degree human evaluation model, in which relevant parties such as superiors, colleagues, and cross-departmental partners combine their daily work impressions with subjective scoring, written evaluations, and other methods to comprehensively evaluate the work performance of operations and maintenance personnel, and finally form an impact evaluation result, which is widely used in talent assessment, promotion review and other scenarios.

[0004] The existing technology has significant drawbacks: the evaluation results are highly dependent on the evaluator's subjective judgment and are easily affected by non-business factors such as job exposure, interpersonal relationships, and personal biases. It cannot mine the implicit technical contributions contained in the work order text based on objective data, resulting in a serious disconnect between the evaluation results and the actual work impact of the operation and maintenance personnel, making it difficult to achieve fair and accurate quantitative assessment. Summary of the Invention

[0005] In view of this, the purpose of this application is to provide a method, device, equipment and storage medium for evaluating the impact of personnel work, which ensures the objectivity and relevance of the evaluation basis and achieves the accuracy and fairness of the evaluation results.

[0006] In a first aspect, embodiments of this application provide a method for evaluating the work influence of personnel, the method comprising: Extract impact events from work order texts that include personnel subjects, action verbs, and business objects; The baseline weights of the semantic influence factors are determined based on the action verbs. The enhancement factor of the business context is determined based on the business object; Obtain the extraction confidence level of the impactful event; The score of the influential event is calculated based on the baseline weight, the enhancement factor, and the extraction confidence level to quantify the work influence of the subject.

[0007] Optionally, the step of extracting impact events containing personnel subjects, action verbs, and business objects from the work order text includes: The work order text is output to the trained event triplet extraction language model to obtain the influence event.

[0008] Optionally, the semantic influence factor includes at least one of the following types: risk intervention power, knowledge output power, innovative suggestion power, and collaborative driving power; The determination of the baseline weights for the semantic influence factor based on the action verb includes: Match the action verbs with a predefined dictionary of action verbs; The type of the semantic influence factor is determined based on the matching results, and a corresponding baseline weight is assigned. or, The action verbs are classified using an intent classification model; The type of the semantic influence factor is determined based on the classification results, and a corresponding baseline weight is assigned.

[0009] Optionally, determining the enhancement factor of the business context based on the business object includes: Obtain the work unit data associated with the business object; Determine whether the data of the work unit meets the preset high-risk scenario conditions; If satisfied, the business context enhancement factor is set to the first weight value; If the condition is not met, the business context enhancement factor is set to a second weight value, wherein the first weight is greater than the second weight.

[0010] Optionally, the preset high-risk scenario conditions include at least one of the following: The refund amount associated with the work order is greater than a preset threshold; The work order involves regulatory reporting; The fault level of the work order is a core fault level; The system to which the work order belongs is the core business system.

[0011] Optionally, the step of calculating the score of the influential event based on the baseline weight, the enhancement factor, and the extraction confidence level to quantify the work influence of the personnel subject includes: Calculate the product of the baseline weight, the enhancement factor, and the extraction confidence level, and use the product as the score of the influential event.

[0012] Optionally, the method further includes: In response to the validity of the warning corresponding to the influential event, the global weight of the semantic influence factor of the corresponding type is updated based on the verification result.

[0013] Secondly, embodiments of this application provide a device for evaluating the impact of personnel work, the device comprising: The Impact Event Extraction module is used to extract impact events containing personnel subjects, action verbs, and business objects from work order texts; The baseline weight determination module is used to determine the baseline weight of the semantic influence factor based on the action verb; An enhancement factor determination module is used to determine the enhancement factor of the business context based on the business object; The confidence level extraction module is used to obtain the extraction confidence level of the influential event. The impact event scoring module is used to calculate the score of the impact event based on the baseline weight, the enhancement factor and the extraction confidence level, so as to quantify the work influence of the subject.

[0014] Optionally, the step of extracting impact events containing personnel subjects, action verbs, and business objects from the work order text includes: The work order text is output to the trained event triplet extraction language model to obtain the influence event.

[0015] Optionally, the semantic influence factor includes at least one of the following types: risk intervention power, knowledge output power, innovative suggestion power, and collaborative driving power; The determination of the baseline weights for the semantic influence factor based on the action verb includes: Match the action verbs with a predefined dictionary of action verbs; The type of the semantic influence factor is determined based on the matching results, and a corresponding baseline weight is assigned. or, The action verbs are classified using an intent classification model; The type of the semantic influence factor is determined based on the classification results, and a corresponding baseline weight is assigned.

[0016] Optionally, determining the enhancement factor of the business context based on the business object includes: Obtain the work unit data associated with the business object; Determine whether the data of the work unit meets the preset high-risk scenario conditions; If satisfied, the business context enhancement factor is set to the first weight value; If the condition is not met, the business context enhancement factor is set to a second weight value, wherein the first weight is greater than the second weight.

[0017] Optionally, the preset high-risk scenario conditions include at least one of the following: The refund amount associated with the work order is greater than a preset threshold; The work order involves regulatory reporting; The fault level of the work order is a core fault level; The system to which the work order belongs is the core business system.

[0018] Optionally, the step of calculating the score of the influential event based on the baseline weight, the enhancement factor, and the extraction confidence level to quantify the work influence of the personnel subject includes: Calculate the product of the baseline weight, the enhancement factor, and the extraction confidence level, and use the product as the score of the influential event.

[0019] Optionally, the device further includes: The global weight update module is used to update the global weight of the semantic influence factor of the corresponding type based on the verification result when the early warning corresponding to the influence event is verified to be valid.

[0020] Thirdly, embodiments of this application provide a computer device, including: a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the computer device is running, the processor communicates with the memory via the bus. When the machine-readable instructions are executed by the processor, the steps of the personnel work influence evaluation method described in any of the optional embodiments of the first aspect are performed.

[0021] Fourthly, embodiments of this application provide a computer-readable storage medium storing a computer program that, when executed by a processor, performs the steps of the personnel work influence evaluation method described in any of the optional embodiments of the first aspect.

[0022] The technical solution provided in this application includes, but is not limited to, the following beneficial effects: Extracting impactful events containing personnel, action verbs, and business objects from work order texts allows for the precise extraction of core information reflecting work value from unstructured work order texts. This clarifies the responsible party, specific actions, and related business scenarios that bind the impact, providing clear and traceable evaluation criteria for subsequent quantitative assessments and avoiding the problem of evaluations lacking specific behavioral support.

[0023] The baseline weight of semantic influence factors is determined based on the behavioral verbs, which enables differentiated value definition of different types of work behaviors of operation and maintenance personnel, distinguishes the influence levels of different behaviors such as risk intervention and knowledge output, and makes the evaluation results reflect the technical content and value differences of the behavior itself, avoiding the one-sidedness of quantifying all work behaviors equally.

[0024] By identifying enhancement factors for the business context based on the business object, work behavior can be correlated with the risk level of the business scenario. This allows effective interventions in high-risk business scenarios to obtain evaluations that are more in line with actual business value, and directly links the scoring results to core needs such as business continuity and financial security, thereby improving the practicality and relevance of the evaluation.

[0025] Obtaining the extraction confidence level of the influential events enables credibility verification of the extracted evaluation criteria, filters out ambiguous, inaccurate, or invalid text information, ensures that the basic data used to calculate the score is true and reliable, avoids evaluation result deviations caused by distorted criteria, and improves the accuracy of the evaluation.

[0026] The score of the impact event is calculated based on the benchmark weight, the enhancement factor, and the extraction confidence level. This integrates three key dimensions: behavioral value, business risk, and information credibility, to achieve accurate quantification from specific work behavior to actual impact. The evaluation results objectively reflect the real work contributions of operation and maintenance personnel, solving the problem that traditional evaluation methods are unable to quantify implicit technical value.

[0027] In summary, the above steps are interconnected, forming a complete evaluation logic from information extraction, value definition, scenario adaptation, credibility verification to comprehensive quantification. This ensures the objectivity and relevance of the evaluation basis, as well as the accuracy and fairness of the evaluation results. It effectively breaks through the limitations of traditional evaluation methods and provides reliable support for the scientific evaluation of the work influence of operation and maintenance personnel.

[0028] To make the above-mentioned objectives, features and advantages of this application more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0029] To more clearly illustrate the technical solutions of the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. It should be understood that the following drawings only show some embodiments of this application and should not be regarded as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.

[0030] Figure 1 A flowchart of a method for evaluating the work influence of personnel provided in Embodiment 1 of this application is shown; Figure 2 A flowchart of a benchmark weight determination method provided in Embodiment 1 of this application is shown; Figure 3 A flowchart of the second benchmark weight determination method provided in Embodiment 1 of this application is shown; Figure 4 A flowchart of an enhancement factor determination method provided in Embodiment 1 of this application is shown; Figure 5 This illustration shows a structural schematic diagram of a personnel work influence evaluation device provided in Embodiment 2 of this application; Figure 6 A schematic diagram of the structure of a computer device provided in Embodiment 3 of this application is shown. Detailed Implementation

[0031] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. The components of the embodiments of this application described and shown in the accompanying drawings can generally be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiments of this application provided in the accompanying drawings is not intended to limit the scope of the claimed application, but merely represents selected embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without inventive effort are within the scope of protection of this application.

[0032] Example 1 This application applies not only to insurance operations and maintenance, but also to the evaluation of knowledge-based positions in finance, telecommunications, and other industries that rely on work order systems.

[0033] To facilitate understanding of this application, the following is combined with... Figure 1 The flowchart illustrating the personnel work influence evaluation method provided in Embodiment 1 of this application will be described in detail for Embodiment 1 of this application.

[0034] See Figure 1 As shown, Figure 1 A flowchart of a method for evaluating the work influence of personnel provided in Embodiment 1 of this application is shown, wherein the method includes steps S101 to S105: S101: Extract impact events from work order texts that include personnel subjects, action verbs, and business objects.

[0035] Specifically, work order text is unstructured text data, specifically referring to work order response text on the operations and maintenance platform, including natural language content such as root cause analysis, processing steps, temporary solutions, and recovery verification. Before extraction, it is necessary to access relevant data from the company's unified operations and maintenance platform, complete the desensitization, deduplication, and template statement cleaning (such as "processed" and "please confirm"), and associate it with the system to which it belongs (such as the refund engine), fault level (P0–P4), refund amount, and whether it involves regulatory reporting, etc.

[0036] The extraction process employs a BERT (Bidirectional Encoder Representations from Transformers, a bidirectional pre-trained language model based on the Transformer architecture) model, fine-tuned for the Insurtech field, combined with an Insurtech dictionary to improve accuracy. The final result is a structured impact event in the form of (subject, action verb, business object), such as (Zhang Gong, discovery, refund engine not verified reinstatement status). This triple serves as the smallest unit of scoring, enabling the binding of responsible parties, the judgment of behavioral value, and the association with business risk scenarios.

[0037] S102: Determine the baseline weight of the semantic influence factor based on the action verb.

[0038] Specifically, the semantic influence factor is a semantic feature extracted from work order responses that reflects the professional depth and risk control capabilities of operations and maintenance personnel. It is divided into four categories, each with a clear extraction logic and benchmark weight: risk intervention capability (extracting statements that proactively discover / warn / block business risks, benchmark weight 0.4), knowledge output capability (extracting behaviors such as explaining complex rules and writing operation guides, benchmark weight 0.3), innovation suggestion capability (extracting content that proposes process optimization and system improvement suggestions, benchmark weight 0.2), and collaboration driving capability (extracting behaviors such as cross-departmental collaboration and progress promotion, benchmark weight 0.1).

[0039] Benchmark weights ( The determination of the action verbs adopts a dual path: the system first matches the action verbs with a predefined action verb dictionary (such as "discovery → risk intervention power" and "assistance → collaborative driving power"). The core of this dictionary is to establish a data mapping relationship. If the dictionary is not matched, a lightweight BERT intent classifier that is independent of the triple extraction model is called. The most likely factor type is determined by the intent probability output by the classifier, and then the corresponding baseline weight is assigned. The intent classifier only focuses on the application level and does not need to consider the training process and the actual training results.

[0040] S103: Determine the enhancement factor of the business context based on the business object.

[0041] Specifically, the Business Context Enhancement Factor (Wcontext) is a weighting amplification coefficient dynamically assigned based on the importance, funding scale, or fault level of the business system associated with the event. Its determination depends on the unit data associated with the business object. The determination of core systems follows a custom rule, that is, systems at or above the Level 3 Insurance Protection Level are core systems (such as refund, policy, and actuarial systems). There is no explicit list but there are unified judgment standards.

[0042] The enhancement factor is determined as follows: if the data of the work unit meets any high-risk scenario conditions (refund amount > 100,000 yuan, involving regulatory reporting, P0 / P1 level fault, or belonging to a core business system), then the value is 1.5; otherwise, the value is 1.0. High-risk scenarios can be expanded to include customer privacy, compliance risks, and other scenarios based on high-frequency words in the order data.

[0043] S104: Obtain the extraction confidence level of the impact event.

[0044] Specifically, this confidence level ( The confidence score is the NLP model's evaluation score of the authenticity of the extracted influential events. The value range is [0,1]. It is obtained by the domain-fine-tuned BERT model, which outputs the predicted probability of key components (verb + object) when extracting triples, and then calibrates it by temperature scaling. Temperature scaling is an existing deep learning technique that solves the problem of "overconfidence" in model prediction by optimizing the temperature parameter T, and ensures that the confidence level is more in line with the actual situation.

[0045] The confidence level is related to the clarity of the technical description in the work order response. For example, a technical description that includes clear risk points and processing logic can achieve a confidence level of around 0.93, while a vague description will have a lower confidence level.

[0046] S105: Calculate the score of the influential event based on the baseline weight, the enhancement factor, and the extraction confidence level to quantify the work influence of the personnel subject.

[0047] Specifically, the scoring uses a three-factor product formula. The employee's influence score I(t) within the time window t is calculated using the following formula:

[0048] The unique definition of each character in the formula is as follows: : Represents the cumulative work influence score of the subject within the time window t, which is the core result for quantifying implicit technical contributions; n: represents the total number of effective impact events extracted from the work order responses of this person within the time window t; e: Represents the sequence number (from 1 to n) of a single impact event, corresponding to each (subject, action verb, business object) triple; : Represents the baseline weight of the semantic influence factor corresponding to the e-th influential event, with a value of any one of 0.4, 0.3, 0.2, or 0.1, determined by the behavioral verb matching dictionary or model classification; : Represents the business context enhancement factor corresponding to the e-th impact event. It is 1.5 for high-risk scenarios and 1.0 for ordinary scenarios. It is determined by the work unit data associated with the business object. : Represents the extraction confidence level of the e-th influential event, with a value range of [0,1], obtained after calibration by the BERT model prediction probability.

[0049] This formula incorporates behavioral values. Business risks Confidence of identification This three-dimensional approach, previously unseen in operations or HR systems, enables end-to-end quantification from "what was written" to "how much loss was avoided." A complete calculation example is as follows: A work order response reads, "Urgent: It was discovered that the refund engine skipped the cooling-off period verification in the reinstated policy scenario, with a single overpayment risk exceeding 150,000 yuan." The extracted triples are (Engineer Liu, Discovery, Refund engine skipped cooling-off period verification), where... =0.4 ("Discovery" matches risk intervention capacity) =1.5 (Refunds exceeding 100,000 yuan are considered high-risk scenarios). =0.93, single event score =0.4×1.5×0.93=0.558. If Liu Gong has two other valid events within the time window t, with scores of 0.32 and 0.28 respectively, then I(t)=0.558+0.32+0.28=1.158.

[0050] The scoring results can be used to statistically analyze the top 10% of high-influence personnel and expert reviewers, solving the problems that traditional results-oriented KPI assessments cannot distinguish differences in technical value and that 360-degree human evaluation lacks objective data support.

[0051] In an optional implementation, the extraction of impact events containing personnel subjects, action verbs, and business objects from the work order text includes: Specifically, this implementation plan is the core operation in the semantic influence factor extraction process, which aims to transform unstructured chemical single texts with low signal-to-noise ratio into quantifiable structured event streams, providing a unique input unit for subsequent scoring.

[0052] The work order text is output to the trained event triplet extraction language model to obtain the influence event.

[0053] Specifically, the trained language model is a BERT model, which has been fine-tuned specifically for the field of insurance technology. It is also combined with an insurance technology dictionary to assist in extraction, ensuring the complete association and accurate identification of the three elements: subject, action verb, and business object, and avoiding extraction bias caused by the special nature of industry terminology.

[0054] In an optional implementation, the semantic influence factor includes at least one of the following types: risk intervention power, knowledge output power, innovation suggestion power, and collaborative driving power.

[0055] Specifically, the typical statement corresponding to risk intervention capability is "the refund logic here may lead to overpayment," the core of which is to identify and proactively prevent business risks; knowledge output capability focuses on knowledge accumulation behaviors such as explaining complex rules, writing operation guides, and answering professional questions; innovation suggestion capability targets content with improvement such as process optimization and system improvement; and collaboration driving capability focuses on behaviors that promote work progress such as cross-departmental collaboration, task allocation, and progress promotion. These four factors together cover the core value output scenarios of operation and maintenance personnel in the process of system maintenance.

[0056] See Figure 2 As shown, Figure 2 The flowchart of a benchmark weight determination method provided in Embodiment 1 of this application is shown, wherein determining the benchmark weight of the semantic influence factor based on the action verb includes steps S201-S202: Specifically, this flowchart corresponds to the "rule matching path" determined by the baseline weights. It is suitable for scenarios where action verbs can hit a predefined dictionary, and it is a way to quickly and accurately assign baseline weights. The core approach, together with the "model classification path," forms a complementary mechanism to ensure the comprehensiveness of the determination of benchmark weights.

[0057] S201: Match the action verb with a predefined dictionary of action verbs.

[0058] Specifically, the predefined behavioral verb dictionary is only a conceptual design and has no explicit fixed list. Its core is to establish a mapping relationship between behavioral verbs and semantic influence factor types to facilitate classification and matching in large models. In addition to "discovery → risk intervention power", it also includes typical mapping examples such as "assistance → collaborative driving power" that are suitable for insurance operation and maintenance scenarios.

[0059] S202: Determine the type of the semantic influence factor based on the matching results and assign a corresponding baseline weight.

[0060] Specifically, if an action verb successfully matches an entry in the dictionary, it will be directly associated with the corresponding semantic influence factor type, and a preset baseline weight will be assigned to that type. For example, matching "propose" (corresponding to innovative suggestion power) will assign 0.2, and matching "explain" (corresponding to knowledge output power) will assign 0.3. All benchmark weights are initially fixed values, and can be dynamically adjusted according to business importance in the future.

[0061] See Figure 3 As shown, Figure 3 The flowchart of the second benchmark weight determination method provided in Embodiment 1 of this application is shown, wherein the step of determining the benchmark weight of the semantic influence factor based on the action verb includes steps S301-S302: Specifically, this flowchart corresponds to the "model classification path" where the baseline weights are determined. It is applicable to scenarios where action verbs do not hit the predefined action verb dictionary. A lightweight BERT intent classifier is used to provide a fallback judgment, avoiding baseline weights that are affected by incomplete dictionary coverage. () missing.

[0062] S301: Classify the action verbs using an intent classification model.

[0063] Specifically, the intent classification model is a lightweight BERT intent classifier, which is an independent model from the triple extraction model. Its core function is to focus on the intent recognition of behavioral verbs in insurance operation and maintenance scenarios. It can quickly output the corresponding factor type probability without a complicated training process.

[0064] S302: Determine the type of the semantic influence factor based on the classification results and assign a corresponding baseline weight.

[0065] Specifically, the classifier outputs the probability of the action verb corresponding to one of four semantic influence factors. The system selects the factor type with the highest probability as the final classification result and assigns a baseline weight to that type. This ensures that even obscure action verbs not covered by the dictionary can be accurately matched to the corresponding baseline weight.

[0066] The four types of semantic influence factors defined in this application and their benchmark weights are shown in Table 1 below:

[0067] Table 1 In an optional implementation, see Figure 4 As shown, Figure 4 The flowchart of an enhancement factor determination method provided in Embodiment 1 of this application is shown, wherein the step of determining the enhancement factor of the business context based on the business object includes steps S401 to S404: Specifically, the flowchart fully presents the business context enhancement factor ( The dynamic determination logic is based on judging the risk level by the work unit data associated with the business object. It amplifies the scoring proportion of personnel influence in high-value business scenarios through differentiated weights, reflecting the evaluation logic that "the higher the business risk, the greater the value of effective intervention".

[0068] S401: Obtain the work unit data associated with the business object.

[0069] Specifically, unit data is the core basis for determining enhancement factors. In addition to the system to which it belongs, the fault level, the refund amount, and whether it involves regulatory reporting, relevant data such as customer privacy and compliance risks can be expanded according to business development needs to ensure the comprehensiveness of risk assessment.

[0070] S402: Determine whether the work unit data meets the preset high-risk scenario conditions.

[0071] Specifically, the pre-set high-risk scenario conditions are formulated based on the core needs of insurance business, covering four key dimensions: fund security (refund amount > 100,000 yuan), regulatory compliance (involving regulatory reporting), system stability (P0 / P1 level failure), and core business (the system belongs to a core system of level 3 or above). Meeting any one of these conditions will result in a high-risk scenario.

[0072] S403: If satisfied, the business context enhancement factor is set to the first weight value.

[0073] Specifically, the first weight value is fixed at 1.5. This value is set in combination with the risk level and impact assessment requirements of insurance business. It aims to highlight the value of the technical intervention and risk warning behaviors of operation and maintenance personnel in high-risk scenarios, so that the score is more in line with the actual contribution of the business.

[0074] S404: If not satisfied, the business context enhancement factor is set to a second weight value, wherein the first weight is greater than the second weight.

[0075] Specifically, the second weight value is fixed at 1.0, which is applicable to ordinary business scenarios. By differentiating it from the first weight value, the influence score can be reasonably distinguished under different risk levels, avoiding misjudgment of value caused by a "one-size-fits-all" scoring logic.

[0076] In an optional implementation, the preset high-risk scenario conditions include at least one of the following: Specifically, the high-risk scenario conditions are the core rules of this method, which are customized for the characteristics of the insurance industry. They can be dynamically expanded based on high-frequency risk-related terms in order data. For example, scenarios such as "risk of customer privacy leakage" and "violation of compliance and regulatory requirements" can be added later to ensure the adaptability of the evaluation model.

[0077] The refund amount associated with the work order is greater than a preset threshold.

[0078] Specifically, the preset threshold is clearly set at 100,000 yuan. That is, when the refund amount associated with a work order exceeds 100,000 yuan, it is directly judged as a high-risk scenario. This threshold is set in combination with the importance of the security of large sums of money in insurance business and can be adjusted according to the company's business scale and risk appetite.

[0079] The work order in question involves regulatory reporting.

[0080] Specifically, work orders involving regulatory reporting are directly related to industry compliance requirements. If a failure or risk occurs, it may lead to serious consequences such as regulatory penalties. Therefore, they are included in high-risk scenarios, highlighting the key value of operations and maintenance personnel in ensuring compliance.

[0081] The fault level of the work order is the core fault level.

[0082] Specifically, the core fault levels are clearly defined as P0 and P1 level faults. These types of faults have a significant impact on business continuity and system stability, requiring rapid response and effective handling by operations and maintenance personnel. The corresponding technical interventions are of higher value and are therefore included in high-risk scenarios.

[0083] The system to which the work order belongs is the core business system.

[0084] Specifically, the criteria for determining core business systems are custom-defined systems at or above the Level 3 Insurance Protection Level. These mainly include systems that directly affect the core processes of insurance business, such as refunds, policies, and actuarial work. There is no explicit list of systems, but there are unified rules for determining the insurance protection level.

[0085] In an optional implementation, calculating the score of the influential event based on the baseline weight, the enhancement factor, and the extraction confidence level to quantify the work influence of the individual subject includes: Specifically, this calculation method is the core of the dynamic influence scoring model of this application, realizing "behavioral value" through the product of three factors. ) + Business Risks ( ) + Recognition credibility ( The three-dimensional weighted aggregation of "" solves the problem of the one-sidedness of traditional evaluation methods, making the scoring more objective and convincing.

[0086] Calculate the product of the baseline weight, the enhancement factor, and the extraction confidence level, and use the product as the score of the influential event.

[0087] Specifically, the product calculation logic is concise and covers key evaluation dimensions, including confidence level ( After temperature scaling calibration, the value is taken as [0,1] to ensure that the model's predicted probability is more reliable. Through this calculation method, templated responses (such as "resolved") will have a score close to 0 due to the lack of complete triplet elements or effective behavioral descriptions, while the scores of real technical interventions, risk warnings and other behaviors will be significantly improved, achieving "good performance and good score".

[0088] In an optional implementation, the method further includes: in response to the early warning corresponding to the influence event being verified as valid, updating the global weight of the semantic influence factor of the corresponding type based on the verification result.

[0089] Specifically, this mechanism is the core patent point of this application, namely the defect feedback-driven weight self-optimization mechanism. It feeds back the model weights with actual business results, forming a closed loop of "early warning-verification-reinforcement", which enables the evaluation model to have dynamic evolution capabilities and improve monthly stability without manual parameter tuning.

[0090] Specifically, the weight update uses the following formula:

[0091] The unique definition of each character in the formula is as follows: : Represents the global baseline weight of a certain type of semantic influence factor after the update. After the update, the total weight of the four types of factors should still be maintained at 1.0 (e.g., after the weight of risk intervention is increased, the weights of other factors are adjusted proportionally). : Indicates the global baseline weight of this type of semantic influence factor before the update (the initial value is the corresponding value among 0.4, 0.3, 0.2, and 0.1); : Represents the smoothing coefficient, with a fixed value of 0.9, used to control the magnitude of weight updates and avoid excessive weight fluctuations due to a single verification result; : Represents the impact gain coefficient brought about by this effective early warning, with a value range of [0,1]. It is calculated as "potential loss avoided this time ÷ maximum single refund amount in the past 30 days".

[0092] The quantification of potential losses focuses on maintenance work orders, using the difference between the erroneous refund amount and the correct refund amount as the core calculation basis. The largest single refund amount in the past 30 days is used as the reference denominator, and the ratio is mapped to the [0,1] interval. For example, if the largest single refund amount in the past 30 days is 500,000 yuan, and the potential loss avoided by this warning is 300,000 yuan, then... =30÷50=0.6, if the weight of a certain factor before the update is... After the update =0.4+(1-0.9)×0.6=0.46.

[0093] The effective determination of defects is divided into two scenarios: quantifiable scenarios (such as avoiding clear financial losses) can be automatically determined as effective after the work order is completed, while some scenarios that cannot be directly quantified require manual review and confirmation; since each warning is recorded in the form of a triple, the system can accurately locate the type of semantic influence factor that needs to be adjusted, ensuring the targeted nature of the weight update.

[0094] The system processing method of this method adopts periodic batch processing (such as daily batch), which can be flexibly adjusted according to the company's assessment frequency. During the processing, data needs to be cleaned and the model trained before the results are exported. The system needs to be integrated with the work order system, event processing system, email system, etc., and supports data retrieval and daily / weekly report output. It is suitable for talent assessment, expert identification and knowledge management optimization scenarios.

[0095] Example 2 See Figure 5 As shown, Figure 5 A schematic diagram of a personnel work influence evaluation device provided in Embodiment 2 of this application is shown, wherein the device includes: The Impact Event Extraction Module 501 is used to extract impact events containing personnel subjects, action verbs, and business objects from work order texts. The baseline weight determination module 502 is used to determine the baseline weight of the semantic influence factor based on the action verb. Enhancement factor determination module 503 is used to determine the enhancement factor of the business context based on the business object; The confidence level extraction module 504 is used to obtain the extraction confidence level of the influential event. The impact event scoring module 505 is used to calculate the score of the impact event based on the baseline weight, the enhancement factor and the extraction confidence level, so as to quantify the work influence of the personnel subject.

[0096] In an optional implementation, the extraction of impact events containing personnel subjects, action verbs, and business objects from the work order text includes: The work order text is output to the trained event triplet extraction language model to obtain the influence event.

[0097] In an optional implementation, the semantic influence factor includes at least one of the following types: risk intervention power, knowledge output power, innovative suggestion power, and collaborative driving power; The determination of the baseline weights for the semantic influence factor based on the action verb includes: Match the action verbs with a predefined dictionary of action verbs; The type of the semantic influence factor is determined based on the matching results, and a corresponding baseline weight is assigned. or, The action verbs are classified using an intent classification model; The type of the semantic influence factor is determined based on the classification results, and a corresponding baseline weight is assigned.

[0098] In an optional implementation, determining the enhancement factor of the business context based on the business object includes: Obtain the work unit data associated with the business object; Determine whether the data of the work unit meets the preset high-risk scenario conditions; If satisfied, the business context enhancement factor is set to the first weight value; If the condition is not met, the business context enhancement factor is set to a second weight value, wherein the first weight is greater than the second weight.

[0099] In an optional implementation, the preset high-risk scenario conditions include at least one of the following: The refund amount associated with the work order is greater than a preset threshold; The work order involves regulatory reporting; The fault level of the work order is a core fault level; The system to which the work order belongs is the core business system.

[0100] In an optional implementation, calculating the score of the influential event based on the baseline weight, the enhancement factor, and the extraction confidence level to quantify the work influence of the individual subject includes: Calculate the product of the baseline weight, the enhancement factor, and the extraction confidence level, and use the product as the score of the influential event.

[0101] In an optional implementation, the device further includes: The global weight update module is used to update the global weight of the semantic influence factor of the corresponding type based on the verification result when the early warning corresponding to the influence event is verified to be valid. Example 3 Based on the same application concept, see [link / reference] Figure 6 As shown, Figure 6 This illustration shows a structural schematic diagram of a computer device provided in Embodiment 3 of this application, wherein, as shown... Figure 6 As shown, the computer device 600 provided in Embodiment 3 of this application includes: The computer device 600 includes a processor 601, a memory 602, and a bus 603. The memory 602 stores machine-readable instructions that can be executed by the processor 601. When the computer device 600 is running, the processor 601 communicates with the memory 602 via the bus 603. When the machine-readable instructions are executed by the processor 601, they perform the steps of the personnel work influence evaluation method shown in Embodiment 1 above.

[0102] Example 4 Based on the same concept, embodiments of this application also provide a computer-readable storage medium storing a computer program, which, when executed by a processor, performs the steps of the personnel work impact evaluation method described in any of the above embodiments.

[0103] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the system and apparatus described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0104] The computer program product for evaluating the impact of personnel work provided in this application includes a computer-readable storage medium storing program code. The instructions included in the program code can be used to execute the methods described in the preceding method embodiments. For specific implementation details, please refer to the method embodiments, which will not be repeated here.

[0105] The personnel work impact evaluation device provided in this application embodiment can be specific hardware on a device or software or firmware installed on the device. The implementation principle and technical effects of the device provided in this application embodiment are the same as those in the foregoing method embodiments. For the sake of brevity, any parts not mentioned in the device embodiment can be referred to the corresponding content in the foregoing method embodiments. Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working processes of the systems, devices, and units described above can all be referred to the corresponding processes in the above method embodiments, and will not be repeated here.

[0106] In the embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. The apparatus embodiments described above are merely illustrative. For example, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. Furthermore, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Additionally, the displayed or discussed mutual couplings, direct couplings, or communication connections may be through some communication interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms.

[0107] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0108] In addition, the functional units in the embodiments provided in this application can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0109] If the aforementioned functions are implemented as software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or a portion of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of this application. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0110] It should be noted that similar labels and letters in the following figures indicate similar items. Therefore, once an item is defined in one figure, it does not need to be further defined and explained in subsequent figures. In addition, the terms "first", "second", "third", etc. are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0111] Finally, it should be noted that the above-described embodiments are merely specific implementations of this application, used to illustrate the technical solutions of this application, and not to limit them. The protection scope of this application is not limited thereto. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that any person skilled in the art can still modify or easily conceive of changes to the technical solutions described in the foregoing embodiments, or make equivalent substitutions for some of the technical features, within the scope of the technology disclosed in this application; and these modifications, changes, or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application. All should be covered within the protection scope of this application. Therefore, the protection scope of this application should be determined by the protection scope of the claims.

Claims

1. A method for evaluating the impact of personnel work, characterized in that, The method includes: Extract impact events from work order texts that include personnel subjects, action verbs, and business objects; The baseline weights of the semantic influence factors are determined based on the action verbs. The enhancement factor of the business context is determined based on the business object; Obtain the extraction confidence level of the impactful event; The score of the influential event is calculated based on the baseline weight, the enhancement factor, and the extraction confidence level to quantify the work influence of the subject.

2. The method according to claim 1, characterized in that, The extraction of impact events from work order texts, which include personnel subjects, action verbs, and business objects, includes: The work order text is output to the trained event triplet extraction language model to obtain the influence event.

3. The method according to claim 1, characterized in that, The semantic influence factors include at least one of the following types: risk intervention capability, knowledge output capability, innovation suggestion capability, and collaborative driving capability; The determination of the baseline weights for the semantic influence factor based on the action verb includes: Match the action verbs with a predefined dictionary of action verbs; The type of the semantic influence factor is determined based on the matching results, and a corresponding baseline weight is assigned. or, The action verbs are classified using an intent classification model; The type of the semantic influence factor is determined based on the classification results, and a corresponding baseline weight is assigned.

4. The method according to claim 1, characterized in that, The enhancement factor for determining the business context based on the business object includes: Obtain the work unit data associated with the business object; Determine whether the data of the work unit meets the preset high-risk scenario conditions; If satisfied, the business context enhancement factor is set to the first weight value; If the condition is not met, the business context enhancement factor is set to a second weight value, wherein the first weight is greater than the second weight.

5. The method according to claim 4, characterized in that, The preset high-risk scenario conditions include at least one of the following: The refund amount associated with the work order is greater than a preset threshold; The work order involves regulatory reporting; The fault level of the work order is a core fault level; The system to which the work order belongs is the core business system.

6. The method according to claim 1, characterized in that, The step of calculating the score of the influential event based on the baseline weight, the enhancement factor, and the extraction confidence level to quantify the work influence of the individual subject includes: Calculate the product of the baseline weight, the enhancement factor, and the extraction confidence level, and use the product as the score of the influential event.

7. The method according to claim 1, characterized in that, The method further includes: In response to the validity of the warning corresponding to the influential event, the global weight of the semantic influence factor of the corresponding type is updated based on the verification result.

8. A device for evaluating the impact of personnel work, characterized in that, The device includes: The Impact Event Extraction module is used to extract impact events containing personnel subjects, action verbs, and business objects from work order texts; The baseline weight determination module is used to determine the baseline weight of the semantic influence factor based on the action verb; An enhancement factor determination module is used to determine the enhancement factor of the business context based on the business object; The confidence level extraction module is used to obtain the extraction confidence level of the influential event. The impact event scoring module is used to calculate the score of the impact event based on the baseline weight, the enhancement factor and the extraction confidence level, so as to quantify the work influence of the subject.

9. A computer device, characterized in that, include: The computer device includes a processor, a memory, and a bus. The memory stores machine-readable instructions executable by the processor. When the computer device is running, the processor communicates with the memory via the bus. When the machine-readable instructions are executed by the processor, they perform the steps of the personnel work impact assessment method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, performs the steps of the personnel work impact evaluation method as described in any one of claims 1 to 7.

Images