Service sheet auditing and scoring method and equipment based on engineer score and return visit result
By conducting multi-dimensional analysis of the contract text, combining engineer ratings and follow-up results, and dynamically adjusting weights, the subjectivity and bias issues of service order rating in existing technologies have been resolved, achieving comprehensive and accurate service order review and rating.
Patent Information
- Application Number
- CN202511218924.3
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-08-28
- Publication Date
- 2025-11-14
AI Technical Summary
In existing technologies, relying solely on engineer ratings or follow-up results to rate service orders is subjective and one-sided, and cannot fully and truthfully reflect the actual service quality of the service order.
By performing multi-dimensional analysis on the pre-set contract text, service orders and follow-up results are obtained, multiple service item standards and scoring standards are constructed, and indicators are extracted by combining engineer scores and follow-up results. The weights are dynamically adjusted using neural network fusion processing to achieve a comprehensive score.
It achieves comprehensive and accurate service order review and scoring, which can reflect the true quality of services, adapt to changes in actual conditions, reduce subjective bias, and improve the accuracy and adaptability of scoring.
Smart Images

Figure CN120952257A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of service project management technology, specifically to a service order review and scoring method and equipment based on engineer ratings and follow-up results. Background Technology
[0002] With the rapid development of information technology, service order management systems have been widely applied in numerous industries due to their efficiency and convenience. Service order review, as a core and crucial link in the entire service order management process, directly and significantly impacts the stable operation of the entire system and is closely linked to customer satisfaction. In related technologies, relying solely on engineer ratings is inherently subjective, as engineers are inevitably influenced by personal feelings and experience. Similarly, relying solely on follow-up feedback often focuses on superficial customer feedback, lacking a deep understanding of the underlying service process and technical details. Both methods yield somewhat one-sided scores that fail to comprehensively and accurately reflect the actual service quality of the service order. Summary of the Invention
[0003] This application provides a service order review and scoring method and equipment based on engineer ratings and follow-up results, which can comprehensively and accurately calculate the score of service orders.
[0004] The technical solution of this application embodiment is as follows: In a first aspect, embodiments of this application provide a service order review and scoring method based on engineer ratings and follow-up results, the method comprising: A multi-dimensional analysis of the preset contract text is performed to obtain multiple service item standards, each of which includes an indicator standard and a corresponding scoring standard. Obtain service orders and follow-up results, extract indicators from the service orders to obtain multiple first scoring indicators, each first scoring indicator corresponding to an indicator standard, obtain first scoring values from engineers who score the first scoring indicators according to the scoring standards, and extract indicators from the follow-up results to obtain multiple second scoring indicators and second scoring values corresponding to each second scoring indicator, each second scoring indicator corresponding to an indicator standard, and the second scoring values are set according to the scoring standards. The first scoring indicators and the second scoring indicators are different. Each of the first scoring indicators and each of the second scoring indicators constitutes indicator data, and each of the first scoring values and each of the second scoring values constitutes the scoring data corresponding to the indicator data. Both the indicator data and the scoring data are represented by sets. The target allocation weights corresponding to the indicator data are obtained by weighting the indicator data and the scoring data according to the indicator data and the scoring data. Obtain the triggering conditions, and dynamically adjust the weight assigned to the target based on the triggering conditions to obtain the dynamically adjusted weight; The audit score result is obtained by performing neural network fusion processing on the dynamically adjusted weights, the indicator data, and the scoring data.
[0005] In the above technical solution, the pre-set contract text is first analyzed from multiple dimensions to obtain multiple service item standards. The contract text stipulates the service quality boundaries. Through multi-dimensional analysis, the requirements of each service item are specified to obtain indicator standards and scoring standards, providing a reference for subsequent scoring. Then, service orders and follow-up results are obtained. Indicators are extracted from the service orders to obtain multiple first scoring indicators, each corresponding to an indicator standard. The first score value is obtained by the engineer based on the scoring standard for the first scoring indicator. Indicators are extracted from the follow-up results to obtain multiple second scoring indicators and their corresponding second score values, each corresponding to an indicator standard. The second score value is set according to the scoring standard. The first and second scoring indicators are different. Two evaluation methods are obtained: engineer scores and follow-up results, to facilitate a more comprehensive service analysis and avoid the bias of subjective opinions. The process involves: constructing indicator data from each primary and secondary scoring indicator; then constructing corresponding scoring data from each primary and secondary scoring value to support subsequent calculations based on the indicator and scoring data; assigning weights to the indicator data based on the indicator and scoring data to obtain the target weights for each indicator; reflecting the impact of different scoring indicators on service order review scores to accurately reflect the true quality of the service; obtaining trigger conditions and dynamically adjusting the target weights based on these conditions to obtain dynamically adjusted weights; and finally, performing neural network fusion processing on the dynamically adjusted weights, indicator data, and scoring data to obtain the review score result, thus achieving a comprehensive and accurate service review score by integrating indicators from different aspects.
[0006] In conjunction with some embodiments of the first aspect, in some embodiments, the step of dynamically adjusting the target allocation weight based on the triggering condition to obtain the dynamically adjusted weight includes: When the triggering condition is one of the following: customer satisfaction fluctuation condition, business strategy change condition, and data distribution offset condition, the triggering condition is matched with a preset trigger type table to obtain matching adjustment rules. The trigger type table includes trigger entries corresponding to the triggering condition and matching adjustment rules corresponding to the trigger entries. The target is assigned a weight dynamically using the matching adjustment rule to obtain the dynamically adjusted weight.
[0007] In conjunction with some embodiments of the first aspect, in some embodiments, the step of matching the triggering condition with a preset triggering type table to obtain matching adjustment rules includes: Under the condition that the customer satisfaction fluctuation condition matches the trigger entry in the trigger type table, the standard deviation of the satisfaction index in the historical indicator data is calculated to obtain the monthly satisfaction standard deviation. If the monthly satisfaction standard deviation is greater than the preset satisfaction threshold, the target allocation weight is increased by the preset fluctuation threshold as the matching adjustment rule. Under the condition that the business strategy change condition matches the trigger entry in the trigger type table, the preset input adjustment coefficient is multiplied by the target allocation weight as the matching adjustment rule; Under the condition that the data distribution offset condition matches the trigger entry in the trigger type table, the KL divergence of the historical indicator data and the indicator data is calculated to obtain the degree of distribution change. If the degree of distribution change is greater than the preset distribution threshold, the step of weighting the indicator data according to the indicator data and the score data to obtain the target allocation weight corresponding to the indicator data is executed to recalculate the target allocation weight and use the recalculated target allocation weight as the matching adjustment rule.
[0008] In conjunction with some embodiments of the first aspect, in some embodiments, the step of dynamically adjusting the target allocation weight based on the triggering condition to obtain the dynamically adjusted weight includes: When the triggering condition is a preset time point update, obtain the historical indicator data corresponding to the historical service order; Entropy calculation is performed on the historical indicator data to obtain the entropy change value. If the entropy change value is greater than the preset change threshold, the target assigned weight is discarded, and the indicator data is weighted using the preset entropy weight method to obtain the dynamic adjustment weight. If the entropy change value is less than or equal to a preset change threshold, the target allocation weight is used as the dynamic adjustment weight.
[0009] In conjunction with some embodiments of the first aspect, in some embodiments, the step of assigning weights to the indicator data based on the indicator data and the scoring data to obtain the target allocation weights corresponding to the indicator data includes: The scoring data is normalized to obtain normalized scores; The information entropy of the indicator data and the normalized score is calculated using a preset entropy weight method to obtain the entropy weight corresponding to the indicator data. The association weights corresponding to the indicator data are obtained by using the preset CRITIC algorithm to perform correlation calculations on the indicator data and the normalized score. The product of the entropy weight and the preset first adjustment coefficient and the product of the correlation weight and the preset second adjustment coefficient are added together to obtain the linear adjustment weight, wherein the sum of the weights corresponding to all index data in the linear adjustment weight is 1. The entropy weight, the first adjustment coefficient, the correlation weight, and the second adjustment coefficient are used to fit the linear adjustment weight to obtain a fitting result. The weight corresponding to the smallest fitting result is used as the initial allocation weight corresponding to the index data. The initial allocation weights are corrected to obtain the target allocation weights.
[0010] In conjunction with some embodiments of the first aspect, in some embodiments, the step of correcting the initial allocation weights to obtain the target allocation weights includes: For each pair of indicators in the aforementioned indicator data, an importance comparison is performed to construct a judgment matrix; The consistency ratio is calculated by performing a consistency calculation on the judgment matrix. If the consistency ratio is less than a preset consistency threshold, the weight of the judgment matrix is calculated to obtain a compliance weight. The initial allocation weights are balanced using the compliance weights to obtain the target allocation weights.
[0011] In conjunction with some embodiments of the first aspect, in some embodiments, after obtaining the service order and follow-up results, extracting indicators from the service order to obtain multiple first scoring indicators, each first scoring indicator corresponding to one indicator standard, obtaining the first scoring value of the engineer scoring the first scoring indicator according to the scoring standard, and extracting indicators from the follow-up results to obtain multiple second scoring indicators and the second scoring value corresponding to each second scoring indicator, the method further includes: The first scoring indicator is deduplicated to obtain the first deduplicated indicator, and the second scoring indicator is deduplicated to obtain the second deduplicated indicator. The first score value corresponding to the first deduplication indicator is filled with missing values according to the degree of contribution to obtain the first filled score, wherein the degree of contribution is the upper limit of the scoring standard in the service item standard. The second score value corresponding to the second deduplication indicator is semantically filled to obtain the second filled score. Each of the first deduplication indicators and each of the second deduplication indicators constitutes indicator data, and each of the first filling scores and each of the second filling scores constitutes the score data corresponding to the indicator data.
[0012] Secondly, embodiments of this application provide a service order review and scoring device based on engineer ratings and follow-up results, the device comprising: The first data processing module is used to perform multi-dimensional analysis on the preset contract text to obtain the standards for each service item. Each service item standard includes an indicator standard and a scoring standard corresponding to the indicator standard. The data acquisition module is used to acquire service orders and follow-up results, extract indicators from the service orders to obtain multiple first scoring indicators, each first scoring indicator corresponding to an indicator standard, acquire first scoring values of engineers scoring the first scoring indicators according to the scoring standards, and extract indicators from the follow-up results to obtain multiple second scoring indicators and second scoring values corresponding to each second scoring indicator, each second scoring indicator corresponding to an indicator standard, the second scoring values being set according to the scoring standards, and the first scoring indicators and the second scoring indicators being different. The second data processing module is used to construct indicator data from each of the first scoring indicators and each of the second scoring indicators, and to construct the scoring data corresponding to the indicator data from each of the first scoring values and each of the second scoring values. Both the indicator data and the scoring data are represented by sets. The weight allocation module is used to allocate weights to the indicator data based on the indicator data and the score data, so as to obtain the target allocation weights corresponding to the indicator data. The weight adjustment module is used to obtain triggering conditions and dynamically adjust the weights assigned to the target based on the triggering conditions to obtain dynamically adjusted weights. The scoring calculation module is used to perform neural network fusion processing based on the dynamically adjusted weights, the indicator data, and the scoring data to obtain the review scoring result.
[0013] Thirdly, embodiments of this application provide an electronic device including a processor, a memory, a user interface, a communication bus, and a network interface. The processor, the memory, the user interface, and the network interface are respectively connected to the communication bus. The memory is used to store instructions. The user interface and the network interface are used to communicate with other devices. The processor is used to execute the instructions stored in the memory to cause the electronic device to perform the method described in any one of the first aspects.
[0014] Fourthly, embodiments of this application provide a computer-readable storage medium storing instructions that, when executed, perform the method described in any one of the methods provided in the first aspect above.
[0015] In summary, one or more technical solutions provided in the embodiments of this application have at least the following technical effects or advantages: 1. The process begins with a multi-dimensional analysis of the pre-defined contract text, resulting in multiple service item standards. The contract text defines service quality boundaries, and this multi-dimensional analysis specifies the requirements for each service item, yielding indicator standards and scoring criteria to provide a reference for subsequent scoring. Next, service requests and follow-up results are obtained. Indicators are extracted from the service requests, resulting in multiple primary scoring indicators, each corresponding to an indicator standard. Engineers assign first-level scores to these indicators based on the scoring standards. Similarly, follow-up results are used to extract secondary scoring indicators, each corresponding to an indicator standard. The second-level scores are determined based on the scoring standards. Since the primary and secondary scoring indicators differ, both engineer scores and follow-up results are used for evaluation, allowing for a more comprehensive service analysis and avoiding the influence of subjective opinions. The system employs a multi-faceted approach: it constructs indicator data from various primary and secondary scoring indicators, and then further constructs corresponding scoring data from these indicators, providing support for subsequent calculations. Weights are allocated to the indicator data based on the indicator and scoring data to obtain target weights. This weight allocation reflects the impact of different scoring indicators on service order review scores, ensuring a more accurate reflection of service quality. Triggering conditions are identified, and the target weights are dynamically adjusted based on these conditions, resulting in dynamically adjusted weights. These adjustments are then used to adjust the weight ratios according to actual conditions, making the overall score more closely reflect actual service quality. Finally, a neural network fusion process is performed on the dynamically adjusted weights, indicator data, and scoring data to obtain the review score result. This integration of different indicators achieves a comprehensive and accurate service review score. Therefore, it effectively solves the problem that single-point scoring in related technologies is too one-sided and cannot fully and accurately reflect the actual service quality of a service order.
[0016] 2. By adjusting the weights based on one of the following triggers—customer satisfaction fluctuation conditions, business strategy change conditions, and data distribution deviation conditions—not only can the bias caused by subjective weight allocation be reduced, but the system can also adapt to changes in actual conditions, making the overall score more in line with the actual service quality and ensuring the accuracy of the score.
[0017] 3. Dynamic adjustments are made based on time points, with adjustments performed once a month at that specific time. This allows for adaptation to changes in actual circumstances, making the overall score more closely reflect the actual service quality and ensuring the accuracy of the score.
[0018] 4. By allocating weights at different levels, a reasonable distribution of weights is achieved, which improves the adaptability and flexibility of the algorithm and meets the diverse business needs of different industries and enterprises.
[0019] 5. By integrating engineer ratings and user feedback and taking a comprehensive approach from multiple perspectives, we have achieved an accurate evaluation of the service order review results. Attached Figure Description
[0020] Figure 1 This is a flowchart illustrating a service order review and scoring method based on engineer ratings and follow-up results, provided in one embodiment of this application. Figure 2 This is a schematic diagram of a service order review and scoring method based on engineer rating and follow-up results, provided in one embodiment of this application. Figure 3 This is a schematic diagram of a service order review and scoring method based on engineer ratings and follow-up results, provided in another embodiment of this application. Figure 4 yes Figure 1 A flowchart illustrating a sub-step of step S500; Figure 5 This is a schematic diagram of the structure of a service order review and scoring device based on engineer ratings and follow-up results provided in one embodiment of this application; Figure 6 This is a schematic diagram of the structure of an electronic device provided in one embodiment of this application. Detailed Implementation
[0021] To enable those skilled in the art to better understand the technical solutions in this specification, the technical solutions in the embodiments of this specification 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.
[0022] In the description of the embodiments of this application, the words "for example" or "for instance" are used to indicate examples, illustrations, or explanations. Any embodiment or design that is described as "for example" or "for instance" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design options. Rather, the use of the words "for example" or "for instance" is intended to present the relevant concepts in a specific manner.
[0023] In the description of the embodiments of this application, the term "multiple" means two or more. For example, multiple systems means two or more systems, and multiple screen terminals means two or more screen terminals. Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the indicated technical features. Thus, a feature defined with "first" or "second" may explicitly or implicitly include one or more of that feature. The terms "comprising," "including," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.
[0024] This application provides a service order review and scoring method, apparatus, electronic device, and readable storage medium based on engineer ratings and follow-up results. The method first performs multi-dimensional analysis on a preset contract text to obtain multiple service item standards. The contract text defines service quality boundaries. Through multi-dimensional analysis, the requirements of each service item are specified, resulting in indicator standards and scoring standards, providing a reference for subsequent scoring. Then, the service order and follow-up results are obtained. Indicators are extracted from the service order to obtain multiple first scoring indicators, each corresponding to an indicator standard. First score values are obtained based on the engineer's rating of the first scoring indicators according to the scoring standards. Indicators are also extracted from the follow-up results to obtain multiple second scoring indicators and corresponding second score values, each corresponding to an indicator standard. The second score values are set according to the scoring standards. The first and second scoring indicators differ. The method then obtains both the engineer's rating and the follow-up results. This assessment method employs several approaches to facilitate comprehensive service analysis and mitigate the influence of subjective opinions. It constructs indicator data from primary and secondary scoring indicators, and further refines this data with corresponding score data, providing support for subsequent calculations. Weights are allocated to the indicator data based on the score data, resulting in target weights that reflect the impact of different scoring indicators on service order review scores and thus reveal the true quality of the service. Triggering conditions are identified, and the target weights are dynamically adjusted accordingly, resulting in dynamically adjusted weights. These adjustments are then used to adjust weight ratios based on actual conditions, ensuring the overall score more closely reflects the actual service quality. Finally, a neural network fusion process is used to integrate the dynamically adjusted weights, indicator data, and score data to obtain the final review score, achieving a comprehensive and accurate service review score by combining indicators from different aspects.
[0025] It should be noted that this service order review and scoring method, based on engineer ratings and follow-up results, can be used for fault repair, software upgrades, and software development in the software and internet sector; or for the inspection and optimization of servers or network equipment in system operation and maintenance. It can also be applied to services in cloud services, finance, and healthcare, demonstrating its wide applicability and providing a comprehensive service review score that reflects the true quality of the service.
[0026] The technical solutions provided in the embodiments of this application will be further described below with reference to the accompanying drawings.
[0027] Reference Figure 1 , Figure 1 This is a flowchart illustrating a service order review and scoring method based on engineer ratings and follow-up results provided in this application embodiment. The service order review and scoring method based on engineer ratings and follow-up results is applied to a service order review and scoring device based on engineer ratings and follow-up results. The method is executed by a processor in an electronic device or a readable storage medium, and includes steps S100, S200, S300, S400, S500, and S600.
[0028] Step S100: Perform multi-dimensional analysis on the preset contract text to obtain multiple service item standards. Each service item standard includes indicator standards and corresponding scoring standards.
[0029] In one embodiment, the preset contract text is a service contract signed with the customer. The service contract includes multiple service items, which can include specific service operations, service results, service time, service feedback, user needs, service progress, service personnel, user information, fees, etc. Different contracts are signed for different services, and the service items in the contracts may vary slightly, but this does not affect the multi-dimensional analysis of different contract texts to extract service item standards. Multi-dimensional analysis can be conducted from the perspectives of attribute layer, relationship network layer, and risk feature layer. The attribute layer extracts service items through timeliness, quality, and cost dimensions. For example, the timeliness dimension includes response cycle, processing time, and delay rate; the quality dimension includes solution accuracy, number of rework, and compliance checks; and the cost dimension includes budget execution rate, change costs, and penalties for breach of contract. The relationship network layer includes historical cooperation records, service progress, and acceptance payments; and the risk feature layer includes dispute resolution items. A multimodal model is used to perform multi-dimensional analysis of the contract text. This multimodal model includes multiple extraction models. For example, a convolutional neural network model or machine learning algorithm is used to analyze the attribute layer, extracting the corresponding indicator standards and scoring standards for each service item. A knowledge graph is used to analyze the relationship network layer, constructing a relationship graph based on historical cooperation records, service progress, and acceptance payments to obtain the correlation between the indicator standards corresponding to the service items and their degree of influence. A recurrent neural network model or its variants are used to analyze the risk feature layer, obtaining risk indicators and corresponding response plans. By analyzing the contract text and performing service item analysis at different levels, multiple service item standards are obtained. These service item standards include indicator standards and their corresponding scoring standards. The indicator standards and their corresponding scoring standards form multiple service item standards, which are then used to constrain engineer scoring and follow-up scores, ensuring the rationality of the scoring indicators and scores. The scoring criteria corresponding to the indicator standards provide a reasonable processing range, that is, the scoring criteria are expressed as a numerical range. The subsequent scoring process of engineers is constrained by the above processing range. For example, it is given that processing should be carried out within 3 working days. If it exceeds 3 working days, it will exceed the agreed deadline, and the score of the service order will need to be reduced.
[0030] It should be noted that the contract text lists various service items, including attributes such as service time and service quality. These attributes correspond to indicator standards. The contract text may or may not specify service standards for these indicators. If service standards are specified, the scoring range corresponding to these service standards constitutes the scoring criteria. For example, the start and end times of the service period are specified, providing a processing time range, which serves as a reference for subsequent scoring values.
[0031] In the absence of agreed-upon service standards, historical service data is used as a reference. Taking processing time as an example, the minimum and maximum processing times in historical service data are found. The minimum is used as the lower limit, and the maximum as the upper limit, forming a processing time range, which serves as the scoring standard. This range provides a reference for subsequent scoring values. Multiple service item standards can be derived from contract texts or historical experience, and these standards provide a reference for subsequent scoring.
[0032] Step S200: Obtain service order and follow-up results; extract indicators from service order to obtain multiple first scoring indicators, each corresponding to an indicator standard; obtain the first scoring value of the first scoring indicator given by the engineer according to the scoring standard; and extract indicators from follow-up results to obtain multiple second scoring indicators and the second scoring value corresponding to each second scoring indicator, each corresponding to an indicator standard; the second scoring value is set according to the scoring standard, and the first and second scoring indicators are different.
[0033] In one embodiment, when serving a user, service personnel record relevant service information and user information to form a service order. The service order includes attributes of relevant information and attributes of user information. Attributes of relevant information include service items, service progress, and service time, while attributes of user information include user name, gender, and user needs. These attributes are the corresponding scoring indicators. Natural language processing (NLP) methods are used to extract indicators from the service order, resulting in multiple first scoring indicators. The NLP method uses indicator standards from the service item standards as prompts to extract indicators from the service order. The extracted first scoring indicators are stored. Engineers score each first scoring indicator in the service order to obtain the corresponding first score value, providing a data foundation for subsequent comprehensive review. It should be noted that engineers refer to the scoring standards in the service item standards when scoring, searching for the first scoring indicator against the indicator standards in each service item standard to obtain the corresponding scoring standard, and then scoring based on that standard.
[0034] like Figure 2 As shown, the system upgrade service item service form includes attributes such as user name, address, service personnel, service type, processing time, system upgrade version, and number of rework attempts. The attributes in the service form are extracted to form the first scoring index. Engineers score the values entered for each of the first scoring indexes to obtain the first score value.
[0035] Based on the stored service orders, user feedback can be collected through questionnaires or online methods. User evaluations of the service orders are collected as follow-up results, which are then read via a pre-defined network interface to prepare for subsequent scoring. When using questionnaires, options for completion or evaluation are designed. A second scoring indicator is obtained based on the questions or evaluation options. The second score value corresponding to the second scoring indicator is obtained based on the user's completion or selection, where the user's completion or selection is set according to the scoring criteria in the service item standards. For example, the evaluation item is a user satisfaction score, with 5 scores or 5 options for the user to choose from, the scores and options being set with reference to the scoring criteria range. The score or option selected by the user is used as the second score value. User completion can be opinions submitted by the user; prompts are set according to the scoring criteria to facilitate the user's completion. Opinions are the indicator, and the completed content is the score value. Similarly, collecting user feedback online is similar to the questionnaire survey described above and will not be elaborated here. When users fill in text or words as input, the corresponding evaluation items and content are extracted using natural language processing (NLP). These extracted evaluation items are then used as a second scoring indicator. The evaluation content is then mapped to a score to obtain the second score value. Specifically, the NLP method uses the indicator standards in the service item standards as prompts to extract keywords from each evaluation item, thus obtaining the second scoring indicator. Alternatively, an OCR (Optical Character Recognition) algorithm can be used to match the indicator standards in the service item standards with the evaluation items to obtain the second scoring indicator. The score mapping for the evaluation content can be done manually by filling in the scores according to the set scoring standards, resulting in the second score value corresponding to the second scoring indicator. The corresponding score is entered by referring to the service evaluation history.
[0036] In one embodiment, after obtaining service orders and follow-up results, extracting indicators from the service orders to obtain multiple first rating indicators, each corresponding to an indicator standard, obtaining the first rating value of the engineers' scores on the first rating indicators according to the rating standards, and extracting indicators from the follow-up results to obtain multiple second rating indicators and the second rating values corresponding to each second rating indicator, the service order review and scoring method based on engineer scores and follow-up results further includes: when extracting evaluation indicators from the service orders, there may be duplicates. A preset text similarity algorithm is used to deduplicate the first rating indicators to obtain first deduplicated indicators. The text similarity algorithm can be TF-IDF or cosine similarity. TF-IDF is used to calculate the similarity between each first rating indicator. If the similarity is greater than 80%, it indicates that the two are duplicates, and one of them is deleted. When extracting user feedback, there may also be cases of identical keywords. Deduplication of the second rating indicators is performed to obtain second deduplicated indicators. This deduplication process also uses a text similarity algorithm, and the specific process is similar to the deduplication of the first rating indicators described above, and will not be elaborated here.
[0037] Then, if the first score value corresponding to the first deduplication indicator is empty, it can be filled with a default value. The default value is determined according to the degree of contribution, which is the upper limit of the scoring standard in the service item standard. For example, if the service time stipulated in the contract text is 3-5 working days, and the scoring standard included in the service item standard is in the range of 3-5, then the upper limit of the scoring standard is 5, the degree of contribution is 5, and the default value to be filled is 5. Alternatively, the first score value can be filled with the average of historical data to obtain the first filled score. Filling in the missing first score value is beneficial to the accuracy of subsequent calculations.
[0038] When the second rating value is text, semantic imputation is performed on the second rating value corresponding to the second deduplication indicator. The imputed text is then mapped to a score to obtain the second imputed score. Specifically, a recurrent neural network model or its variant is used to perform semantic analysis and prediction on the second rating value to obtain the imputed text. The recurrent neural network model is a pre-trained model capable of accurate prediction using contextual logic. Historical data is used to adjust the parameters of the trained model, giving the network good generalization ability in text prediction and improving the accuracy of imputation. Mapping the score of the imputed text is similar to mapping the second rating value based on the filled feedback record, and will not be elaborated here. It should be noted that when the user opinion indicator is present, if the user does not fill in the corresponding content, it is assumed that the service is good, and the corresponding score is the maximum value of the rating standard. By imputing missing values, the overall user feedback can be obtained, enabling a comprehensive review of the service order.
[0039] After the above data processing, each first deduplication indicator and each second deduplication indicator constitutes indicator data. This indicator data is represented as a set. Adding each first and second deduplication indicator to the set facilitates subsequent calculation of the pairwise relationships between indicators within the set and also facilitates the comprehensive calculation of the first and second deduplication indicators. Each first filled score and each second filled score constitutes the corresponding score data for the indicator data. This score data is also represented as a set, with the data in the score data corresponding to the indicator data for convenient subsequent calculations.
[0040] Step S300: The first scoring indicator and the second scoring indicator are combined to form indicator data, and the first score value and the second score value are combined to form the score data corresponding to the indicator data.
[0041] In one embodiment, each first rating indicator and each second rating indicator constitutes indicator data. This indicator data is represented as a set. Adding each first rating indicator and each second rating indicator to the set facilitates subsequent determination of the correlation between pairs of indicators within the set and also facilitates comprehensive calculation of the first and second rating indicators. Each first rating value and each second rating value constitutes the corresponding rating data for the indicator data. This rating data is also represented as a set, with the data in the rating data corresponding to the indicator data for convenient subsequent calculations. By integrating engineer ratings and user feedback, and considering multiple aspects comprehensively, an accurate evaluation of the service order review results is achieved. Step S400: Based on the indicator data and the scoring data, assign weights to the indicator data to obtain the target allocation weights corresponding to the indicator data.
[0042] Specifically, the indicator data and score data are weighted according to the indicator data and score data to obtain the target allocation weights corresponding to the indicator data. This includes, but is not limited to, normalizing the score data since the first and second score values are obtained from different score sources. Using the max-min algorithm to process the score data can eliminate the computational problems caused by inconsistent data, resulting in normalized scores.
[0043] Then, the information entropy of the indicator data and normalized scores is calculated using a pre-defined entropy weighting method to obtain the entropy weights corresponding to the indicator data. Specifically: First, the proportion of each indicator in the indicator data is calculated. This is done by calculating the sum of the normalized scores corresponding to all indicators, and then dividing the normalized score of each indicator by the sum to obtain the proportion of each indicator. Next, the proportion is multiplied by its logarithm (base e), and the product of the proportions of all indicators and the logarithm of the proportion is added together. This sum is then multiplied by an adjustment coefficient to obtain the information entropy of each indicator. The adjustment coefficient is calculated by taking the logarithm of the sample size and then its reciprocal, ensuring the information entropy is within the range of 0-1. The information entropy of each indicator is then used to assign weights to the indicator data: 1 is subtracted from the information entropy to obtain the median value. A smaller information entropy indicates higher data dispersion and provides more effective information; a larger median value indicates a higher weight. Finally, the median value of each indicator is compared to the sum of the median values of all indicators to obtain the weight of each indicator. By calculating the information entropy to obtain the entropy weights of each indicator in the indicator data, we can reduce the unreasonable problems caused by subjective weight allocation and facilitate accurate comprehensive evaluation in the future.
[0044] The pre-defined CRITIC algorithm is also used to perform correlation calculations between indicator data and normalized scores to obtain the correlation weights corresponding to the indicator data. Specifically: First, the standard deviation of each normalized score is calculated. This standard deviation reflects the dispersion of the data. The larger the standard deviation, the greater the fluctuation of the normalized score in the indicator data, and the more significant the differences between different samples on this indicator, providing more discriminative information and assigning a higher weight. Then, the correlation between one indicator and other indicators in the indicator data is calculated. The covariance between the indicator and other indicators, the standard deviation of the indicator, and the standard deviations of other indicators are calculated. The standard deviation of the indicator is multiplied by the standard deviations of other indicators to obtain the product value. The covariance is divided by the product value to obtain the correlation corresponding to the indicator. The correlation is subtracted from 1 to obtain the conflict rate, which converts the correlation into an independence measure. When the correlation value is 1, the conflict rate is 0. The conflict rates of each indicator are added together to obtain the conflict rate and the correlation weight. The correlation weight is obtained by multiplying the standard deviation of each indicator by the conflict rate and the correlation weight. This associated warrant reflects the independence between effective information and indicators. Indicators with high independence and more effective information will be assigned higher weights to avoid the influence of other factors and to avoid the evaluation results being biased towards subjective opinions.
[0045] Reinitialize the values of the first and second adjustment coefficients, ensuring that the sum of the first and second adjustment coefficients is 1. For example, both the first and second adjustment coefficients are set to 0.5. Multiply the entropy weight by the first adjustment coefficient and the correlation weight by the second adjustment coefficient, then sum the results to obtain an initial fitted linear adjustment weight. The sum of the weights corresponding to all indicator data in this linear adjustment weight should be 1. If the sum of the weights corresponding to all indicator data in the linear adjustment weight is not 1, discard the values of the first and second adjustment coefficients and reinitialize them to ensure that the linear adjustment weight meets the conditions.
[0046] When the values of the first and second adjustment coefficients ensure that the linear adjustment weight meets the conditions, a minimum fit is performed using the linear adjustment weight as a benchmark, employing entropy weight, the first adjustment coefficient, correlation weight, and the second adjustment coefficient. Specifically, the values of the first and second adjustment coefficients are modified, and the product of the entropy weight and the first adjustment coefficient, and the product of the correlation weight and the second adjustment coefficient are summed. The difference between this sum and the linear adjustment weight is the fitted result. The minimum value is found in the fitted results, and the corresponding first and second adjustment coefficients are obtained. The initial allocation weight is obtained by multiplying the entropy weight by the first adjustment coefficient and adding the correlation weight by the second adjustment coefficient. The initial allocation weight obtained through the above process avoids the review results being biased towards subjective factors, increases the weight of objective factor evaluation, and is conducive to reflecting the actual service quality of the service order.
[0047] In one embodiment, the initial allocation weights are modified to obtain the target allocation weights, including but not limited to: comparing the importance of every two indicators in the indicator data to construct a judgment matrix; performing consistency calculation on the judgment matrix to obtain a consistency ratio; if the consistency ratio is less than a preset consistency threshold, performing weight calculation on the judgment matrix to obtain a compliance weight; and using the compliance weights to perform a balancing calculation on the initial allocation weights to obtain the target allocation weights.
[0048] Specifically, the importance of each pair of indicators in the data is compared, and the degree of importance is represented numerically. For example, 1 indicates equal importance, 3 indicates slightly important, 5 indicates significantly important, and 7 indicates strongly important. The reciprocal is used to represent the inverse comparison. The importance is determined through expert experience and can also be determined based on expert analysis of historical data. Through these comparisons, a judgment matrix is constructed, with the indicators as rows and columns. The diagonal lines of this matrix represent the indicator being compared to itself, with a value set to 1. The other positions are filled with values according to their importance, providing a basis for subsequent calculations.
[0049] Based on the obtained judgment matrix, calculate the largest eigenvalue and matrix order. Then, find the random consistency index based on the matrix order. Subtract the matrix order from the largest eigenvalue, and divide the result by the matrix order minus 1 to obtain the consistency index. Divide the consistency index by the random consistency index to obtain the consistency ratio. If the consistency ratio is less than a preset consistency threshold, calculate the eigenvector corresponding to the largest eigenvalue in the judgment matrix to obtain the compliance weight. Alternatively, the compliance weight can be obtained by averaging the data for each row or column. Consistency verification ensures the compliance weight is effective and conflict-free. If the consistency ratio is greater than or equal to the preset consistency threshold, manually adjust the index and recalculate until the compliance weight is obtained.
[0050] The initial allocation weights are then balanced using compliance weights. Specifically: First, a balancing factor is obtained. This factor is set by professionals based on historical data analysis and their own experience, balancing the influence of subjective and objective factors. The balancing factor is multiplied by the initial allocation weight, and the result of subtracting the balancing factor from 1 is multiplied by the compliance weight. The products of these two multiplications are then added together to obtain the target allocation weight. This balancing factor balances subjective and objective factors, preventing the review results from being biased towards subjective factors and ensuring that the review results accurately reflect the quality of the service. By allocating weights at different levels, a reasonable distribution of weights is achieved, improving the algorithm's adaptability and flexibility, and meeting the diverse business needs of different industries and enterprises.
[0051] Step S500: Obtain the triggering condition, and dynamically adjust the weight assigned to the target based on the triggering condition to obtain the dynamically adjusted weight.
[0052] In one embodiment, the target is assigned a weight dynamically based on a triggering condition to obtain a dynamically adjusted weight, including but not limited to the following steps: Step S510: If the triggering condition is one of the following: customer satisfaction fluctuation condition, business strategy change condition, or data distribution offset condition, the triggering condition is matched with a preset triggering type table to obtain matching adjustment rules. The triggering type table includes trigger entries corresponding to the triggering conditions and matching adjustment rules corresponding to the trigger entries.
[0053] Specifically, the customer satisfaction fluctuation condition is triggered by changes in satisfaction indicators within both current and historical data. A change in these indicators generates a customer satisfaction fluctuation condition, which then triggers subsequent adjustments. The business strategy change condition arises when different indicators in the current and historical data have different emphases. For example, in historical data, processing time had a relatively small impact on ratings. However, in the current data, due to business development or user requirements, services must be completed within a specified timeframe, increasing the focus on processing time. This generates a business strategy change condition, which triggers subsequent adjustments. The data distribution offset condition involves calculating the distribution of current and historical data and using KL divergence to assess whether a significant offset has occurred. This data distribution offset condition triggers subsequent adjustments. Historical data refers to data from the previous month of the current month or the previous quarter of the current quarter.
[0054] like Figure 3 As shown, the trigger conditions are matched against a preset trigger type table to obtain matching adjustment rules, including but not limited to the following steps: Step S511: Under the condition that the customer satisfaction fluctuation condition matches the trigger entry in the trigger type table, calculate the standard deviation of the satisfaction index in the historical indicator data to obtain the monthly satisfaction standard deviation.
[0055] Step S5111: Determine whether the monthly satisfaction standard deviation is greater than the preset satisfaction threshold.
[0056] Step S5112: If the monthly satisfaction standard deviation is greater than the preset satisfaction threshold, the target allocation weight is increased by the preset fluctuation threshold as the matching adjustment rule.
[0057] In some possible embodiments of this application, the trigger type table includes trigger entries corresponding to trigger conditions and matching adjustment rules corresponding to the trigger entries. Trigger entries include customer satisfaction types, business strategy change condition types, data distribution offset condition types, etc. Each trigger entry corresponds to one of the aforementioned trigger conditions. When the customer satisfaction fluctuation condition successfully matches the customer satisfaction type in the trigger type table, it indicates a change in customer satisfaction. The satisfaction index from historical indicator data is retrieved, and its standard deviation is calculated to obtain the monthly satisfaction standard deviation. This standard deviation provides the basis for subsequent weight adjustments. If the monthly satisfaction standard deviation is greater than a preset satisfaction threshold, the target allocation weight is added to the preset fluctuation threshold, and this calculation process is used as the matching adjustment rule. The preset fluctuation threshold is taken between 5% and 10%, and can be determined by random selection.
[0058] Step S5113: If the monthly satisfaction standard deviation is less than or equal to a preset satisfaction threshold, the target allocation weight is reduced by a preset fluctuation threshold as the matching adjustment rule. The target allocation weight is subtracted from the preset fluctuation threshold; this calculation process is used as the matching adjustment rule. By finding the matching adjustment strategy corresponding to the customer satisfaction fluctuation condition, it is beneficial to adjust the weight according to the matching adjustment strategy when satisfaction is triggered, in order to adapt to changes in actual conditions.
[0059] Step S512: If the business strategy change conditions and the trigger entries in the trigger type table are successfully matched, the preset input adjustment coefficient and the target allocation weight are multiplied as the matching adjustment rule.
[0060] In some possible embodiments of this application, when the business strategy change conditions successfully match the customer satisfaction type in the trigger type table, it indicates a change in business focus. A preset input adjustment coefficient is multiplied by the target allocation weight as a matching adjustment rule, providing a basis for subsequent weight adjustments. The preset input adjustment coefficient is a manually adjusted value determined by experts based on business development; this coefficient ranges from 0.8 to 1.2 and is calculated using randomly selected values.
[0061] Step S513: Under the condition that the data distribution offset condition matches the trigger entry in the trigger type table, calculate the KL divergence of the historical indicator data and the indicator data to obtain the degree of distribution change.
[0062] Step S5131: Determine whether the degree of distribution change is greater than a preset distribution threshold; Step S5132: When the degree of distribution change is greater than the preset distribution threshold, perform the step of weighting the indicator data according to the indicator data and the score data to obtain the target allocation weight corresponding to the indicator data, so as to recalculate the target allocation weight and use the recalculated target allocation weight as the matching adjustment rule.
[0063] In some possible embodiments of this application, if the data distribution offset condition successfully matches the data distribution offset condition type in the trigger type table, it indicates that the indicator data distribution may have shifted. The KL divergence between the distribution of historical indicator data and the distribution of the indicator data is calculated to obtain the degree of distribution change. The KL divergence accurately reflects the degree of data change, preparing for subsequent determination of the weight adjustment strategy. If the degree of distribution change exceeds a preset distribution threshold, it indicates a significant indicator data shift, requiring recalculation of weights. The step of weight allocation based on indicator data and scoring data is executed to obtain the target allocation weight corresponding to the indicator data, thus reallocating the weights. This weight reallocation is similar to the process in step S400 above and will not be elaborated here. The recalculated target allocation weight is used as the matching adjustment rule, providing a basis for subsequent weight adjustments.
[0064] Step S5133: If the degree of distribution change is less than or equal to the preset distribution threshold, keep the target allocation weight unchanged and use keeping the target allocation weight unchanged as the matching adjustment rule.
[0065] It should be noted that the text similarity matching method is used to match one of the customer satisfaction fluctuation conditions, business strategy change conditions, and data distribution offset conditions with the preset trigger type table. The text similarity algorithm can be TF-IDF or cosine similarity, which will not be elaborated here.
[0066] Step S520: Dynamically adjust the weights assigned to the target using the matching adjustment rules to obtain the dynamically adjusted weights.
[0067] In some possible embodiments of this application, based on the matching adjustment rules obtained in step S510 for different situations, the target allocation weight is dynamically adjusted using the corresponding matching adjustment rules under the corresponding triggering conditions to obtain the dynamically adjusted weight. The weight adjustment is carried out by triggering one of the customer satisfaction fluctuation conditions, business strategy change conditions, and data distribution offset conditions. This not only reduces the deviation caused by subjective weight allocation, but also adapts to changes in actual situation, making the comprehensive score more in line with the actual service quality and ensuring the accuracy of the score.
[0068] like Figure 4 As shown, the target weights are dynamically adjusted based on triggering conditions to obtain dynamically adjusted weights, including but not limited to the following steps: Step S530: If the trigger condition is a preset time point update, obtain the historical indicator data corresponding to the historical service order.
[0069] In some possible embodiments of this application, the preset time point is the first day of the month or the last day of the month, etc. When the trigger condition is the preset time point update, the update is triggered at that time point of the month to obtain the historical indicator data corresponding to the historical service order. The historical indicator data is the data recorded and stored when reviewing the monthly data. It is obtained by reading the stored historical indicator data so that entropy change calculation can be performed on the historical indicator data in the future.
[0070] Step S540: Calculate the entropy of historical indicator data to obtain the entropy change value; Step S550: Determine whether the entropy change value is greater than the preset change threshold; Step S560: If the entropy change value is greater than the preset change threshold, the target assigned weight is discarded, and the index data is weighted using the preset entropy weight method to obtain the dynamically adjusted weight.
[0071] In some possible embodiments of this application, the information entropy of historical indicator data is calculated using the entropy weight method. The calculation process is similar to step S400 and will not be repeated here. Then, the information entropy is subtracted from 1 to obtain the entropy change value, which reflects the redundancy of the information entropy. When the entropy change value is greater than a preset threshold, it indicates that the effective information provided by the data has increased, and the target allocation weights calculated above can no longer accurately reflect the true situation of the service. Therefore, the target allocation weights are discarded, and the entropy weight method is used again to allocate weights to the indicator data, resulting in dynamically adjusted weights. In the above case, using only the entropy weight method for weight allocation not only allows for rapid weight allocation but also avoids bias towards subjective factors in the review process. Dynamic adjustments are triggered by time points, with adjustments performed once a month at that specific time point. This adapts to changes in actual conditions, making the comprehensive score more closely reflect the actual service quality and ensuring the accuracy of the score.
[0072] Step S570: If the entropy change value is less than or equal to the preset change threshold, the target allocation weight is used as the dynamic adjustment weight.
[0073] In some possible embodiments of this application, when the entropy change value is less than or equal to a preset change threshold, it indicates that the effective information provided by the data has not changed significantly, and the target allocation weight is used as the dynamic adjustment weight. Dynamic adjustment is triggered by a time point, and an adjustment is performed once at that time point each month. This adapts to changes in actual conditions, making the comprehensive score more closely reflect the actual service quality and ensuring the accuracy of the score.
[0074] Step S600: Based on the dynamically adjusted weights, indicator data, and scoring data, perform neural network fusion processing to obtain the audit scoring result.
[0075] In one embodiment, a multimodal neural network model is used to perform neural network fusion processing on dynamically adjusted weights, indicator data, and scoring data to obtain a review score result. This review score result integrates indicators from different aspects, achieving a comprehensive and accurate service review score. The multimodal neural network model includes support vector machines, random forest algorithms, XGBoost models, and attention mechanisms. Support vector machines, random forest algorithms, and XGBoost models are used to extract features from the indicator data and the corresponding scoring data. The extracted features are then concatenated to obtain concatenated features. The dynamically adjusted weights are used as attention weights in the attention mechanism. Attention fusion is performed using the attention weights and concatenated features to obtain the review score result, enabling a relatively comprehensive and accurate score calculation for service orders.
[0076] like Figure 5As shown, this application embodiment provides a service order review and scoring device 100 based on engineer ratings and follow-up results. This device 100 performs multi-dimensional analysis on a preset contract text using a first data processing module 110 to obtain multiple service item standards. The contract text defines service quality boundaries. Through multi-dimensional analysis, the requirements of each service item are specified, resulting in indicator standards and scoring standards, providing a reference for subsequent scoring. The device uses a data acquisition module 120 to acquire service orders and follow-up results, extracting indicators from the service orders to obtain multiple first scoring indicators, each corresponding to an indicator standard. It then obtains the first score value from the engineer's evaluation of the first scoring indicators according to the scoring standards, and extracts indicators from the follow-up results to obtain multiple second scoring indicators and their corresponding second score values, each corresponding to an indicator standard. The second score value is set according to the scoring standards. Since the first and second scoring indicators differ, the device obtains both engineer ratings and follow-up results for a more comprehensive service analysis and to avoid subjective bias. The system addresses the one-sided impact of opinions. The second data processing module 130 constructs indicator data from the various first and second scoring indicators, and then constructs the corresponding scoring data from the various first and second scoring values. Both indicator data and scoring data are represented as sets, providing support for subsequent calculations based on the indicator data and scoring data. The weight allocation module 140 allocates weights to the indicator data and scoring data to obtain the target allocation weights corresponding to the indicator data. This weight allocation reflects the impact of different scoring indicators on the service order review score, thus reflecting the true quality of the service. The weight adjustment module 150 obtains trigger conditions and dynamically adjusts the target allocation weights based on these conditions, resulting in dynamically adjusted weights. These dynamic weight adjustments are triggered by the trigger conditions, allowing the weight ratios to be changed according to actual conditions, making the comprehensive score more closely reflect the actual service quality. Finally, the scoring calculation module 160 performs neural network fusion processing on the dynamically adjusted weights, indicator data, and scoring data to obtain the review score result. This integrates indicators from different aspects, achieving a comprehensive and accurate service review score.
[0077] It should be noted that the first data processing module 110 is connected to the data acquisition module 120, the data acquisition module 120 is connected to the second data processing module 130, the second data processing module 130 is connected to the weight allocation module 140, the weight allocation module 140 is connected to the weight adjustment module 150, and the weight adjustment module 150 is connected to the score calculation module 160. The aforementioned service order review and scoring method based on engineer ratings and follow-up results is applied to the service order review and scoring device 100 based on engineer ratings and follow-up results. The device 100 performs multi-dimensional analysis on a pre-set contract text to obtain multiple service item standards. The contract text defines the service quality boundaries. Through multi-dimensional analysis, the requirements of each service item are specified, resulting in indicator standards and scoring standards, providing a reference for subsequent scoring. Then, the device obtains the service order and follow-up results, extracts indicators from the service order to obtain multiple first scoring indicators, each corresponding to an indicator standard. It obtains the first score value given by the engineer based on the scoring standards for each first scoring indicator, and extracts indicators from the follow-up results to obtain multiple second scoring indicators and their corresponding second score values, each corresponding to an indicator standard. The second score value is set according to the scoring standards, and the first and second scoring indicators differ. The device then obtains the engineer ratings and follow-up results. Two evaluation methods were used to facilitate a more comprehensive service analysis and avoid the bias of subjective opinions. Each primary and secondary scoring indicator was combined to form indicator data, and then each primary and secondary score value was combined to form the corresponding score data, providing support for subsequent calculations based on the indicator and score data. Weights were assigned to the indicator data based on the indicator and score data to obtain the target allocation weights for each indicator. This weight allocation reflects the impact of different scoring indicators on the service order review score, thus reflecting the true quality of the service. Trigger conditions were obtained, and the target allocation weights were dynamically adjusted based on these conditions to obtain dynamically adjusted weights. These dynamic weight adjustments were triggered by the trigger conditions to change the weight ratios according to actual conditions, making the comprehensive score more closely reflect the actual service quality. Finally, a neural network fusion process was used to process the dynamically adjusted weights, indicator data, and score data to obtain the review score result. This fusion of different indicators achieves a comprehensive and accurate service review score.
[0078] It should also be noted that the apparatus provided in the above embodiments is only illustrated by the division of the above functional modules. In actual applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the device can be divided into different functional modules to complete all or part of the functions described above. In addition, the apparatus and method embodiments provided in the above embodiments belong to the same concept, and their specific implementation process can be found in the method embodiments, which will not be repeated here.
[0079] This application also discloses an electronic device. (See reference...) Figure 6 , Figure 6 This is a schematic diagram of the structure of an electronic device according to an embodiment of this application. The electronic device 500 may include: at least one processor 501, at least one network interface 504, a user interface 503, a memory 505, and at least one communication bus 502.
[0080] The communication bus 502 is used to enable communication between these components.
[0081] The user interface 503 may include a display screen and a camera. Optionally, the user interface 503 may also include a standard wired interface and a wireless interface.
[0082] The network interface 504 may optionally include a standard wired interface or a wireless interface (such as a Wi-Fi interface).
[0083] The processor 501 may include one or more processing cores. The processor 501 connects to various parts of the server using various interfaces and lines, and performs various server functions and processes data by running or executing instructions, programs, code sets, or instruction sets stored in memory 505, and by calling data stored in memory 505. Optionally, the processor 501 may be implemented using at least one hardware form of Digital Signal Processing (DSP), Field-Programmable Gate Array (FPGA), or Programmable Logic Array. The processor 501 may integrate one or a combination of several of the following: Central Processing Unit (CPU), Graphics Processing Unit (GPU), and Modem. The CPU primarily handles the operating system, user interface, and applications; the GPU is responsible for rendering and drawing the content required for display; and the modem handles wireless communication. It is understood that the modem may also be implemented as a separate chip without being integrated into the processor 501.
[0084] The memory 505 may include random access memory (RAM) or read-only memory. Optionally, the memory 505 may include a non-transitory computer-readable storage medium. The memory 505 may be used to store instructions, programs, code, code sets, or instruction sets. The memory 505 may include a program storage area and a data storage area, wherein the program storage area may store instructions for implementing an operating system, instructions for at least one function (such as touch function, sound playback function, image playback function, etc.), instructions for implementing the above-described method embodiments, etc.; the data storage area may store data involved in the above-described method embodiments, etc. Optionally, the memory 505 may also be at least one storage device located remotely from the aforementioned processor 501. (Refer to...) Figure 6 The memory 505, which serves as a computer storage medium, may include an operating system, a network communication module, a user interface module, and an application program based on a service order review and scoring method using engineer ratings and follow-up results.
[0085] exist Figure 6 In the illustrated electronic device 500, the user interface 503 is mainly used to provide an input interface for the user and to acquire user input data; while the processor 501 can be used to call an application stored in the memory 505 that is a service order review and scoring method based on engineer ratings and follow-up results. When executed by one or more processors 501, the electronic device 500 performs one or more methods as described in the above embodiments. It should be noted that, for the foregoing method embodiments, for the sake of simplicity, they are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, because according to this application, some steps can be performed in other orders or simultaneously. Secondly, those skilled in the art should also understand that the embodiments described in the specification are all preferred embodiments, and the actions and modules involved are not necessarily essential to this application.
[0086] In the above embodiments, the descriptions of each embodiment have different focuses. For parts not described in detail in a certain embodiment, please refer to the relevant descriptions in other embodiments.
[0087] In the various embodiments provided in this application, it should be understood that the disclosed apparatus can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative; for instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be through some service interface; the indirect coupling or communication connection between apparatuses or units may be electrical or other forms.
[0088] 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.
[0089] Furthermore, the functional units in the various embodiments of 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. The integrated unit can be implemented in hardware or as a software functional unit.
[0090] If the integrated unit is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage device (CMD). Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, or all or part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a memory 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 of the various embodiments of this application. The aforementioned memory includes various media capable of storing program code, such as USB flash drives, portable hard drives, magnetic disks, or optical disks.
[0091] The above are merely exemplary embodiments of this disclosure and should not be construed as limiting the scope of this disclosure. Any equivalent changes and modifications made in accordance with the teachings of this disclosure shall still fall within the scope of this disclosure. Other embodiments of this disclosure will readily conceive of those skilled in the art upon consideration of the specification and the disclosure of practical truths.
[0092] This application is intended to cover any variations, uses, or adaptations of this disclosure that follow the general principles of this disclosure and include common knowledge or customary techniques in the art not described in this disclosure. The specification and embodiments are to be considered exemplary only, and the scope and spirit of this disclosure are defined by the claims.
Claims
1. A service order review and scoring method based on engineer ratings and follow-up results, characterized in that, The method includes: A multi-dimensional analysis of the preset contract text is performed to obtain multiple service item standards, each of which includes an indicator standard and a corresponding scoring standard. Obtain service orders and follow-up results, extract indicators from the service orders to obtain multiple first scoring indicators, each first scoring indicator corresponding to an indicator standard, obtain first scoring values from engineers who score the first scoring indicators according to the scoring standards, and extract indicators from the follow-up results to obtain multiple second scoring indicators and second scoring values corresponding to each second scoring indicator, each second scoring indicator corresponding to an indicator standard, and the second scoring values are set according to the scoring standards. The first scoring indicators and the second scoring indicators are different. Each of the first scoring indicators and each of the second scoring indicators constitutes indicator data, and each of the first scoring values and each of the second scoring values constitutes the scoring data corresponding to the indicator data. Both the indicator data and the scoring data are represented by sets. The target allocation weights corresponding to the indicator data are obtained by weighting the indicator data and the scoring data according to the indicator data and the scoring data. Obtain the triggering conditions, and dynamically adjust the weight assigned to the target based on the triggering conditions to obtain the dynamically adjusted weight; The audit score result is obtained by performing neural network fusion processing on the dynamically adjusted weights, the indicator data, and the scoring data.
2. The method according to claim 1, characterized in that, The step of dynamically adjusting the target allocation weight based on the triggering condition to obtain the dynamically adjusted weight includes: When the triggering condition is one of the following: customer satisfaction fluctuation condition, business strategy change condition, and data distribution offset condition, the triggering condition is matched with a preset trigger type table to obtain matching adjustment rules. The trigger type table includes trigger entries corresponding to the triggering condition and matching adjustment rules corresponding to the trigger entries. The target is assigned a weight dynamically using the matching adjustment rule to obtain the dynamically adjusted weight.
3. The method according to claim 2, characterized in that, The step of matching the triggering conditions with a preset triggering type table to obtain matching adjustment rules includes: Under the condition that the customer satisfaction fluctuation condition matches the trigger entry in the trigger type table, the standard deviation of the satisfaction index in the historical indicator data is calculated to obtain the monthly satisfaction standard deviation. If the monthly satisfaction standard deviation is greater than the preset satisfaction threshold, the target allocation weight is increased by the preset fluctuation threshold as the matching adjustment rule. Under the condition that the business strategy change condition matches the trigger entry in the trigger type table, the preset input adjustment coefficient is multiplied by the target allocation weight as the matching adjustment rule; Under the condition that the data distribution offset condition matches the trigger entry in the trigger type table, the KL divergence of the historical indicator data and the indicator data is calculated to obtain the degree of distribution change. If the degree of distribution change is greater than the preset distribution threshold, the step of weighting the indicator data according to the indicator data and the score data to obtain the target allocation weight corresponding to the indicator data is executed to recalculate the target allocation weight and use the recalculated target allocation weight as the matching adjustment rule.
4. The method according to claim 1, characterized in that, The step of dynamically adjusting the target allocation weight based on the triggering condition to obtain the dynamically adjusted weight includes: When the triggering condition is a preset time point update, obtain the historical indicator data corresponding to the historical service order; Entropy calculation is performed on the historical indicator data to obtain the entropy change value. If the entropy change value is greater than the preset change threshold, the target assigned weight is discarded, and the indicator data is weighted using the preset entropy weight method to obtain the dynamic adjustment weight. If the entropy change value is less than or equal to a preset change threshold, the target allocation weight is used as the dynamic adjustment weight.
5. The method according to claim 1, characterized in that, The step of assigning weights to the indicator data based on the indicator data and the scoring data to obtain the target allocation weights corresponding to the indicator data includes: The scoring data is normalized to obtain normalized scores; The information entropy of the indicator data and the normalized score is calculated using a preset entropy weight method to obtain the entropy weight corresponding to the indicator data. The association weights corresponding to the indicator data are obtained by using the preset CRITIC algorithm to perform correlation calculations on the indicator data and the normalized score. The product of the entropy weight and the preset first adjustment coefficient and the product of the correlation weight and the preset second adjustment coefficient are added together to obtain the linear adjustment weight, wherein the sum of the weights corresponding to all index data in the linear adjustment weight is 1. The entropy weight, the first adjustment coefficient, the correlation weight, and the second adjustment coefficient are used to fit the linear adjustment weight to obtain a fitting result. The weight corresponding to the smallest fitting result is used as the initial allocation weight corresponding to the index data. The initial allocation weights are corrected to obtain the target allocation weights.
6. The method according to claim 5, characterized in that, The step of correcting the initial allocation weights to obtain the target allocation weights includes: For each pair of indicators in the aforementioned indicator data, an importance comparison is performed to construct a judgment matrix; The consistency ratio is calculated by performing a consistency calculation on the judgment matrix. If the consistency ratio is less than a preset consistency threshold, the weight of the judgment matrix is calculated to obtain a compliance weight. The initial allocation weights are balanced using the compliance weights to obtain the target allocation weights.
7. The method according to claim 1, characterized in that, After obtaining service orders and follow-up results, extracting indicators from the service orders to obtain multiple first scoring indicators, each first scoring indicator corresponding to an indicator standard, obtaining first scoring values for the first scoring indicators based on the scoring standards, and extracting indicators from the follow-up results to obtain multiple second scoring indicators and second scoring values corresponding to each second scoring indicator, the method further includes: The first scoring indicator is deduplicated to obtain the first deduplicated indicator, and the second scoring indicator is deduplicated to obtain the second deduplicated indicator. The first score value corresponding to the first deduplication indicator is filled with missing values according to the degree of contribution to obtain the first filled score, wherein the degree of contribution is the upper limit of the scoring standard in the service item standard. The second score value corresponding to the second deduplication indicator is semantically filled to obtain the second filled score. Each of the first deduplication indicators and each of the second deduplication indicators constitutes indicator data, and each of the first filling scores and each of the second filling scores constitutes the score data corresponding to the indicator data.
8. A service order review and scoring device based on engineer ratings and follow-up results, characterized in that, The device includes: The first data processing module is used to perform multi-dimensional analysis on the preset contract text to obtain multiple service item standards, each of which includes an indicator standard and a scoring standard corresponding to the indicator standard. The data acquisition module is used to acquire service orders and follow-up results, extract indicators from the service orders to obtain multiple first scoring indicators, each first scoring indicator corresponding to an indicator standard, acquire first scoring values of engineers scoring the first scoring indicators according to the scoring standards, and extract indicators from the follow-up results to obtain multiple second scoring indicators and second scoring values corresponding to each second scoring indicator, each second scoring indicator corresponding to an indicator standard, the second scoring values being set according to the scoring standards, and the first scoring indicators and the second scoring indicators being different. The second data processing module is used to construct indicator data from each of the first scoring indicators and each of the second scoring indicators, and to construct the scoring data corresponding to the indicator data from each of the first scoring values and each of the second scoring values. Both the indicator data and the scoring data are represented by sets. The weight allocation module is used to allocate weights to the indicator data based on the indicator data and the score data, so as to obtain the target allocation weights corresponding to the indicator data. The weight adjustment module is used to obtain triggering conditions and dynamically adjust the weights assigned to the target based on the triggering conditions to obtain dynamically adjusted weights. The scoring calculation module is used to perform neural network fusion processing based on the dynamically adjusted weights, the indicator data, and the scoring data to obtain the review scoring result.
9. An electronic device, characterized in that, The device includes a processor, a memory, a user interface, a communication bus, and a network interface. The processor, the memory, the user interface, and the network interface are respectively connected to the communication bus. The memory is used to store instructions. The user interface and the network interface are used to communicate with other devices. The processor is used to execute the instructions stored in the memory to cause the electronic device to perform the method as described in any one of claims 1-7.
10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions that, when executed, perform the method as described in any one of claims 1-7.
Citation Information
Patent Citations
Family doctor cancer treatment service assessment method, device and equipment, and storage medium
CN109636160A
Network access quality comprehensive evaluation method and device and electronic equipment
CN116582443A
Teaching quality comprehensive evaluation system based on big data analysis
CN120069651A
Data-driven flood grading method and system
CN120408271A
Method and device for evaluating effectiveness of transformer fire extinguishing system
US20230153737A1