Decoration supervision service quality optimization method and system based on eight-dimensional feature evaluation

Abstract

The invention belongs to the technical field of engineering management, and provides a decoration supervision service quality optimization method and system based on eight-dimensional feature evaluation, and the method comprises the steps: collecting multi-source heterogeneous data related to supervision service through a distributed data processing platform, carrying out the integrated preprocessing and deep fusion of the multi-source heterogeneous data, and carrying out the optimization of the quality of decoration supervision service. Outputting an eight-dimensional feature vector based on a depth feature engineering framework, generating an evaluation report through causal reasoning, generating an improvement suggestion according to an evaluation result, performing real-time updating, pushing the evaluation result and the improvement suggestion to a supervisor through a mobile terminal, tracking an improvement progress, and forming closed-loop management. According to the invention, the supervision personnel are evaluated from eight dimensions including supervision efficiency, service quality and satisfaction, and supervision human efficiency, core influence factors of each dimension are accurately found out, evaluation and improvement directions are output, the supervision personnel are supervised and promoted to learn and promote through the mobile terminal, a professional high-efficiency service team is established, and the efficiency of the supervision personnel is improved. And new development opportunities and revolution power are brought to the decoration industry.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of engineering management, and particularly relates to a decoration supervision service quality optimization method and system based on eight-dimensional feature evaluation. BACKGROUND

[0002] At present, the decoration industry, as an important field closely related to people's daily life, has long been facing many severe challenges. The service relies heavily on manual work, so that every link in the decoration process highly depends on the skills and experience of construction personnel, which directly leads to uneven service quality. Moreover, the market penetration rate is low, although the home decoration market is huge, the integration and application of Internet technology in it are limited, and most consumers still rely on traditional decoration methods and channels, lacking convenient and efficient information acquisition and transaction platforms.

[0003] More prominent is that the industry is extremely scattered, and the market concentration is extremely low, even the highest market value enterprise occupies a very small share in the industry, which not only leads to disordered competition within the industry but also exists in the situation of information asymmetry, so that it is difficult to accurately obtain key information such as decoration quality judgment, price rationality evaluation, and construction progress control. This series of problems affects the healthy development of the decoration industry and needs to be improved.

[0004] In the aspect of decoration supervision evaluation, the existing technology usually adopts a simple and extensive evaluation method. Most of the time, it only focuses on 3-5 explicit indicators, such as simply focusing on the number of times the supervisor is present, whether the project is completed on time, and other superficial factors. This evaluation method lacks comprehensiveness and precision, and cannot deeply analyze the work profile of the supervisor. At the same time, in data processing and analysis, it relies on traditional experience judgment, which is difficult to quickly process complex data and accurately mine the core factors affecting the quality of supervision services. In the management of supervisors, there is often a lack of a closed-loop management mode from evaluation to improvement, even if problems are found, it is difficult to timely and effectively supervise the supervisors to learn and improve, so as to ensure the continuous improvement of the professional and service ability of the supervision team and the steady improvement of the service quality.

[0005] Specifically, the existing decoration supervision evaluation method has the following defects: The evaluation system is one-sided and single: the existing evaluation method only focuses on a few explicit indicators, and cannot comprehensively cover the supervision efficiency, service quality, and service satisfaction, etc. It is difficult to evaluate the supervisors comprehensively and accurately, cannot deeply analyze their work profile, and cannot meet the actual needs.

[0006] The technology application is backward: relying on traditional experience judgment, when facing a large amount of complex data, the processing capacity is insufficient, it is difficult to accurately mine the core influencing factors in each dimension, it is difficult to provide a scientific improvement direction for the supervisors, leading to the limitation of the scientificity and accuracy of the supervision evaluation.

[0007] The management mode is not closed-loop: there is no closed-loop management from evaluation to improvement, even if it is found that the supervision personnel has problems, it cannot be timely and effectively supervised through system means to learn and improve, which is not conducive to the continuous improvement of the professional and service ability of the supervision team and the steady improvement of the service quality. SUMMARY

[0008] The purpose of the present application is to overcome the existing defects, and provide a decoration supervision service quality optimization method and system based on eight-dimensional feature evaluation.

[0009] In order to solve the above technical problems, the present application provides the following technical solutions: The first purpose of the present application is to provide a decoration supervision service quality optimization method based on eight-dimensional feature evaluation, which comprises: (1) collecting multi-source heterogeneous data related to supervision service through a distributed data processing platform, including structured supervision log, unstructured text report and time series engineering data; (2) integrating multi-source heterogeneous data by using feature alignment and space-time association algorithm, and cleaning and labeling the multi-source heterogeneous data; (3) using cross-modal semantic mapping technology based on neural network to correlate text description and numerical indicators in high-dimensional space, constructing a three-dimensional data cube covering personnel-project-environment, and deeply integrating multi-source heterogeneous data; (4) generating eight-dimensional feature vectors related to supervision efficiency, service quality and satisfaction based on dynamic weight distribution mechanism and text sentiment analysis method, the eight-dimensional feature vectors including supervision personnel efficiency dimension, service ability dimension, professional ability dimension, customer complaint dimension, customer satisfaction dimension, NPS score dimension, customer praise dimension and customer return single dimension; (5) combining deep feature engineering framework for causal reasoning analysis, generating detailed evaluation report containing eight-dimensional specific performance and overall service level, analyzing each dimension problem according to the evaluation result, and generating targeted improvement suggestions; (6) based on real-time data flow and dynamic model adjustment mechanism, continuously updating the evaluation result; (7) pushing the evaluation result and improvement suggestions to the supervision personnel in real time through the mobile terminal and tracking the improvement progress, forming a closed-loop management.

[0010] Further, the quantitative indicators of the supervisor performance dimension include the amount completed per unit of time and the average time spent on-site, the quantitative indicators of the service capability dimension include the hidden danger follow-up rate and the communication and coordination ability assessment, the quantitative indicators of the professional ability dimension include the professional knowledge examination results and the log writing professionalism, the quantitative indicators of the customer complaint dimension include the number of complaints, the proportion of complaint problem classification, and the sentiment analysis of complaint problems, the quantitative indicators of the customer satisfaction dimension include the revisit satisfaction score and the app satisfaction score, the quantitative indicators of the customer praise dimension include the number of praises and the praise content classification, and the quantitative indicators of the customer return single dimension include the return single rate and the return single type score. The calculation method of the NPS score dimension is: NPS = (number of recommenders / total number of samples) x 100% - (number of detractors / total number of samples) x 100%.

[0011] Further, in step (5), the specific steps of conducting causal reasoning analysis combined with the deep feature engineering framework include: Algorithm selection and model construction: Selecting random forest as the causal reasoning algorithm framework, taking the eight-dimensional feature vector as the input and the supervision service effect index as the output label, and constructing a random forest model based on the bootstrap sampling method; Model training: Using training data to train the model, and randomly selecting part of the features to find the best split point when each decision tree splits the node, and through the ensemble learning of multiple decision trees, the complex relationship between each feature and the effect in the data is learned; Causal relationship analysis: Based on the trained model, the influence weight of each feature on the supervision service effect is quantified by analyzing the frequency of feature use in decision tree splitting and the impact of feature changes on prediction results, and the key causal relationship is determined; Result output: Output includes feature importance ranking and quantitative evaluation report of the impact on supervision service quality.

[0012] Further, in step (6), based on real-time data flow and dynamic model adjustment mechanism, the specific steps of continuously iterating and updating the evaluation results include: Real-time capture of new supervision records, customer feedback and project progress data through a distributed platform, real-time analysis of preprocessed data using stream computing, monitoring of the instantaneous state of the eight-dimensional feature indicators, and triggering automatic early warning when an abnormal threshold is detected; Start the parameter adaptive system, dynamically adjust the eight-dimensional feature weights and the causal reasoning model parameters according to the real-time index changes, and migrate the historical project knowledge to the new project scenario through the dynamic weight migration learning mechanism; Re-substitute the updated model parameters into the deep feature engineering framework to generate an updated version of the evaluation report containing the latest index data and causal relationships.

[0013] Another object of the present application is to provide an eight-dimensional feature evaluation-based decoration supervision service quality optimization system, comprising: A multi-source heterogeneous data collection module is configured to collect multi-source heterogeneous data related to supervision services through a distributed data processing platform, including structured supervision logs, unstructured text reports, and time-series engineering data. A data preprocessing module is configured to integrate multi-source heterogeneous data using a feature alignment and spatio-temporal correlation algorithm, and to clean and label the multi-source heterogeneous data. A multi-modal data fusion module uses a neural network-based cross-modal semantic mapping technology to correlate text descriptions and numerical indicators in a high-dimensional space, constructs a three-dimensional data cube covering personnel, projects, and the environment, and performs deep fusion of multi-source heterogeneous data. A deep feature engineering module includes an eight-dimensional feature generator, a text sentiment analysis submodule, and an adaptive feature screening submodule. The eight-dimensional feature generator generates an eight-dimensional feature vector related to supervision efficiency, service quality, and satisfaction based on a dynamic weight distribution mechanism. The text sentiment analysis submodule uses a Transformer architecture to perform sentiment classification on customer feedback text. The adaptive feature screening submodule identifies key influencing factors in each dimension through causal reasoning. A real-time dynamic update module uses a hybrid architecture of stream computing and batch processing, combined with a dynamic weight adjustment mechanism, to achieve real-time data updating and model parameter adaptation. A multi-dimensional evaluation output module generates an evaluation report containing the specific performance of the eight dimensions and the overall service level based on the results of deep feature engineering. An improvement direction generation module analyzes problems in each dimension based on the evaluation results and generates targeted improvement suggestions. A supervision and monitoring module pushes the evaluation results and improvement suggestions to supervision personnel in real time through a mobile terminal and tracks improvement progress.

[0014] Further, the eight-dimensional feature vector includes: A supervision personnel efficiency dimension for measuring the ability and efficiency of supervision personnel to complete effective supervision work within a unit of time; A service capability dimension for measuring the comprehensive ability exhibited by supervision personnel in providing supervision services; A professional capability dimension for evaluating the level of supervision personnel in terms of decoration professional knowledge and skills; A customer complaint dimension for reflecting customer dissatisfaction with supervision services; A customer satisfaction dimension for measuring the degree of customer satisfaction with supervision services; An NPS score dimension for evaluating the spread and influence of supervision services among the customer group; A customer praise dimension is used to reflect the recognition and appreciation degree of the customer to the service of the supervisor. A customer return sheet dimension is used to measure the feeling of the customer to the supervision service and reflect the willingness of the customer to recommend new customers.

[0015] Another object of the present application is to provide an electronic device comprising a processor and a memory storing a computer program, wherein the processor implements the eight-dimension feature evaluation-based decoration supervision service quality optimization method provided by the first object of the present application when executing the computer program.

[0016] Another object of the present application is to provide a computer-readable storage medium storing a computer program, wherein the computer program is executed by a processor to implement the eight-dimension feature evaluation-based decoration supervision service quality optimization method provided by the first object of the present application.

[0017] Another object of the present application is to provide a server comprising at least one processor and a memory in communication connection with the processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the processor to enable the at least one processor to execute the eight-dimension feature evaluation-based decoration supervision service quality optimization method provided by the first object of the present application.

[0018] In combination with the above technical solutions, the present application has the following beneficial effects compared with the prior art: A comprehensive evaluation system is constructed: an evaluation system based on supervision efficiency, service quality and service satisfaction is created, and is refined to eight dimensions such as supervisor efficiency and service ability, and multi-modal data fusion processing technology is used to deeply fuse multi-source heterogeneous data such as structured, unstructured and time-series data, so as to comprehensively and accurately evaluate the supervisors, deeply analyze the work panorama, set a multi-dimensional evaluation benchmark for the industry, and solve the problem of one-sidedness of the existing evaluation system.

[0019] Frontier technologies are integrated: machine learning and large model analysis technology are deeply integrated into supervision evaluation, advanced technologies are used to quickly process complex data, accurately find out the core influencing factors of each dimension, and real-time output scientific improvement direction, so as to improve the scientificity and accuracy of supervision evaluation, and solve the problem of backward application of the existing technology.

[0020] A closed-loop management mode is created: a mixed architecture of stream computing and batch processing, a dynamic weight adjustment mechanism and a parameter self-adaptive system are realized through real-time dynamic updating module, and a closed-loop management mode from evaluation to improvement is created. Through the APP of the supervisor and the management means, the supervisor is prompted and urged in real time to learn and improve according to the evaluation results, the professional and service ability of the supervision team is ensured to be continuously improved, the service quality is ensured to be steadily improved, and the problem of non-closed loop of the existing management mode is solved.

[0021] The application solves many existing problems in the decoration industry through innovative technical means and a comprehensive evaluation system. The system evaluates the supervision personnel from eight dimensions such as supervision efficiency, service quality and satisfaction, uses advanced technologies such as machine learning and large model analysis, accurately finds out the core influencing factors of each dimension, outputs evaluation and improvement direction, and uses the supervision personnel APP and management means to supervise their learning and improvement, and is committed to building a professional and efficient service team, and bringing new development opportunities and reform power to the decoration industry. BRIEF DESCRIPTION OF DRAWINGS

[0022] The accompanying drawings are included to provide a further understanding of the application, and constitute a part of the specification, together with the embodiments of the application, to explain the application, and do not constitute a limitation on the application. In the drawings: Figure 1 is a flowchart of the decoration supervision service quality optimization method based on eight-dimensional feature evaluation provided by the embodiment of the application; Figure 2 is a data processing flowchart provided by the embodiment of the application; Figure 3 is an eight-dimensional feature output principle diagram provided by the embodiment of the application; Figure 4 is a structure principle diagram of the decoration supervision service quality optimization system based on eight-dimensional feature evaluation provided by the embodiment of the application. DETAILED DESCRIPTION

[0023] The preferred embodiments of the application are described below in conjunction with the accompanying drawings, and it should be understood that the preferred embodiments described herein are only used to illustrate and explain the application, and do not limit the application.

[0024] Embodiment 1: As shown in Figure 1 is an embodiment of the decoration supervision service quality optimization method based on eight-dimensional feature evaluation provided by the application, which specifically includes the following steps: S1: Collecting multi-source heterogeneous data related to supervision service through a distributed data processing platform, including structured supervision log, unstructured text report and time sequence engineering data; S2: Integrating multi-source heterogeneous data by using feature alignment and space-time association algorithm, and cleaning and labeling the multi-source heterogeneous data; S3: Using cross-modal semantic mapping technology based on neural network to associate text description and numerical indicators in high-dimensional space, constructing a three-dimensional data cube covering personnel-project-environment, and deeply integrating multi-source heterogeneous data; S4: Generate an eight-dimensional feature vector related to supervision efficiency, service quality and satisfaction based on a dynamic weight allocation mechanism and a text sentiment analysis method, the eight-dimensional feature vector including a supervisor efficiency dimension, a service capability dimension, a professional ability dimension, a customer complaint dimension, a customer satisfaction dimension, an NPS score dimension, a customer praise dimension and a customer return single dimension; S5: Conduct causal reasoning analysis combined with a deep feature engineering framework to generate a detailed evaluation report containing eight-dimensional specific performance and overall service level, analyze each dimension problem according to the evaluation result, and generate targeted improvement suggestions; S6: Based on real-time data flow and dynamic model adjustment mechanism, continuously update the evaluation results; S7: Real-time push the evaluation results and improvement suggestions to the supervision personnel through the mobile terminal and track the improvement progress to form a closed-loop management.

[0025] Specifically, the embodiments of the present application focus on three core links of data processing, analysis and evaluation, and feedback and improvement, realize comprehensive and accurate evaluation and continuous improvement of supervision personnel through multi-dimensional data collection, leading technology fusion analysis and closed-loop management mode.

[0026] (I) Data collection layer: 1. Multi-source heterogeneous data collection: a distributed data processing platform is built to collect various types of data related to supervision services. It covers structured supervision logs, which record the time, place, inspection items and results of each supervision work in detail; unstructured text reports, such as customer evaluation of supervision services and complaint content; and time-series engineering data, such as time nodes of engineering progress and quality detection data at different stages. These data reflect the supervision service process and effect from multiple angles.

[0027] 2. Data integration and preprocessing: feature alignment and spatio-temporal correlation algorithm is used to integrate multi-source heterogeneous data, eliminate data islands, and ensure data consistency and integrity. At the same time, the original data is preprocessed such as cleaning and labeling to provide high-quality data basis for subsequent analysis.

[0028] (II) Data analysis layer: 1. Multi-modal data fusion processing technology: cross-modal semantic mapping technology based on neural network is used to correlate text description and numerical indicators in high-dimensional space, and a three-dimensional data cube covering "personnel-project-environment" is constructed. Deep fusion of different types of data is realized to mine potential relationships and provide comprehensive data support for multi-dimensional evaluation.

[0029] 2. Deep feature engineering framework: Eight-dimensional feature generator: based on a dynamic weight allocation mechanism, closely combined with the characteristics of supervision business, eight-dimensional features closely related to supervision efficiency, service quality and satisfaction are generated to comprehensively and meticulously depict the whole picture of the work of the supervision personnel, such as Figure 3 as shown.

[0030] The eight-dimensional features are specifically composed as follows: ①Supervision personnel efficiency dimension: This dimension mainly measures the ability and efficiency of the supervision personnel to complete effective supervision work within a unit of time. The specific quantitative indicators include: Unit time completion quantity: the actual number of supervision visits completed by the supervision personnel within a natural month is counted.

[0031] Average visit time: the average time spent on each visit is calculated. For example, 56 visits were completed within a week, and the total time spent was 42 hours, so the average task processing time was 0.75 hours per visit. The shorter the average visit time, the more tasks the supervision personnel can handle within a unit of time, and the higher the personnel efficiency.

[0032] ②Service ability dimension: This dimension focuses on the comprehensive ability exhibited by the supervision personnel in providing supervision services. The specific indicators are as follows: Hidden danger follow-up rate: refers to the proportion of supervision personnel who timely follow up after discovering decoration problems. For example, 20 decoration problems were discovered within a month, and 18 of them were given solutions within the specified time (such as 24 hours), so the problem solution timely rate was 90%. Timely solving problems helps to ensure the smooth progress of the decoration project, reflecting the service ability of the supervision personnel.

[0033] Communication and coordination ability assessment: by collecting homeowners' feedback questionnaires and evaluations of the supervision personnel during the service process, the text analysis module assesses the communication and coordination ability of the supervision personnel. For example, a questionnaire survey is conducted to ask customers to score the supervision personnel in terms of coordinating relationships, conveying information accuracy and timeliness, etc., with a full score of 10 points. The average score is used to measure their communication and coordination ability. Good communication and coordination ability is one of the key factors to ensure the quality of supervision services.

[0034] ③Professional ability dimension: The professional ability dimension is used to assess the level of supervision personnel in terms of decoration professional knowledge and skills. The specific quantitative indicators include: Professional knowledge examination score: supervision personnel are regularly organized to take decoration professional knowledge examinations, covering building structure, electrical installation, material identification, etc. The examination score is calculated in percentage, for example, a supervision personnel scored 85 points in one examination, and a higher score reflects a better mastery of professional knowledge.

[0035] Log writing professionalism: A professional checks the supervision service log released by the supervisor, and scores the service log according to the inspection standard. For example, in a natural month, 60 logs are checked, with an average score of 85.5. The higher the score, the more professional this item is.

[0036] ④ Customer complaint dimension: This dimension mainly reflects the customer's dissatisfaction with the supervision service. The relevant quantitative indicators are as follows: Complaint times: Directly count the number of complaints initiated by customers against the supervision personnel service within a certain period of time. For example, a customer complained about a certain supervisor 3 times in a quarter. The more complaints, the more likely the customer is dissatisfied with the service.

[0037] Complaint problem classification ratio: The important degree of customer complaints is classified, such as feedback, one type of complaint, two types of complaint, etc., and the proportion of each type of problem in the total complaint problem is calculated. For example, among the 10 complaints, 8 are feedback, accounting for 80%. By analyzing the complaint problem classification ratio, the problems of the supervision personnel can be found more accurately for targeted improvement.

[0038] Complaint problem sentiment analysis: Through the complaint text sentiment analysis module, the complaint content is scored, with a level of 10. The stronger the negative emotions of the customer, the higher the level, and the lower the score of the supervision personnel. For example, in a certain period of time, the module analyzes that a customer is strongly dissatisfied with a certain supervisor, with a level of 9, and serves 40 customers, with a score of 77.5.

[0039] ⑤ Customer satisfaction dimension: The customer satisfaction dimension aims to measure the customer's satisfaction with the supervision service. The following indicators are quantified: Follow-up satisfaction score: Regularly visit customers for questionnaire survey, and the questionnaire content covers various aspects of supervision service, such as service quality, communication effect, problem solving ability, etc. Let the customer score according to the actual experience, usually using 5-point system (1 point for very dissatisfied, 5 points for very satisfied). For example, 50 questionnaires are collected, and the average score is calculated. If the average score is 4.9, it indicates that the customer's satisfaction with the service of the supervisor is high. If there is negative evaluation, it is dynamically weighted. The less negative evaluation in the period, the higher the weight.

[0040] Mini-program satisfaction score: Customers evaluate the supervision personnel each time they come to the mini-program, with a full score of 5. Negative evaluations are weighted. For example, there are 50 evaluations, 2 of which are very dissatisfied after weighting. Then calculate the average score, which is 4.1 points, indicating that the customer's satisfaction with the service of the supervisor is low.

[0041] ⑥NPS score dimension: The NPS (Net Promoter Score) score is an important indicator of customer loyalty and reputation, and is used in this dimension to evaluate the spread and influence of supervision services among the customer group. Its calculation method is: NPS=(number of recommenders / total sample number) x 100%-(number of detractors / total sample number) x 100%. For example, 200 customers were surveyed, of which 120 were recommenders and 30 were detractors, then NPS=(120 / 200) x 100%-(30 / 200) x 100%=45%. A higher NPS score indicates that the supervision service has a good reputation among customers, and customers are more willing to recommend it to others, which has a positive impact on the brand promotion and business expansion of the enterprise.

[0042] ⑦Customer praise dimension: This dimension reflects the degree of recognition and appreciation of customers for the services of supervision personnel. The specific quantitative indicators are as follows: Praise times: The number of times a customer praises a supervision personnel in writing or other forms within a statistical period. For example, a customer praised a supervision personnel 5 times in a month, and more praise times reflect that the supervision personnel performed well in the service process and was affirmed by the customer.

[0043] Praise content grading: The content of customer praise is classified by the sentiment analysis module, the stronger the positive emotion, the higher the score of the praise content, which is divided into 10 levels, and the proportion of each type of praise content in the total praise content is calculated. For example, in 15 praises, the total emotional score is 120 points, and the score is 80 points. Through praise grading, the strengths of supervision personnel can be identified, serving as a role model for other supervision personnel.

[0044] ⑧Customer return order dimension: The customer return order dimension mainly focuses on the customer's feelings about the supervision service and whether they are willing to recommend the supervision service to friends. The specific indicators include: Return order rate: The number of new customers recommended by a supervision personnel to a client within a cycle. For example, serving 100 customers, a total of 32 new customers were recommended, and the return order rate was 32%. A higher return order rate indicates that the customer is very satisfied with the supervision service and is willing to recommend it to friends, which helps the enterprise to survive better.

[0045] Return order type score: The type of order recommended by a supervision personnel to a client within a cycle is calculated, and the higher the order version, the higher the weight. For example, serving 100 customers, a total of 15 basic version orders, 10 standard version orders, 5 upgraded version orders, and 1 premium version order were recommended, and the return order type score was 12 points. A higher score indicates a greater contribution to the enterprise.

[0046] Through the above eight dimensions, each dimension is quantified by a series of specific indicators, providing a comprehensive and in-depth portrayal of the work performance of supervision personnel.

[0047] In the eight-dimensional feature generation process, the dynamic weight distribution mechanism plays a key role. This mechanism dynamically adjusts the weight of each dimension based on the business weight and the real-time score results of each dimension of the eight-dimensional score of the current supervision personnel.

[0048] In the initial stage of system operation, the following business weight proportions are set for each dimension of the eight-dimensional score: supervision personnel efficiency dimension weight proportion 25%, this dimension focuses on the efficiency of supervision personnel in completing effective supervision work within a certain period of time, and has an important influence on the overall assessment; service ability dimension weight proportion 5%, aiming to evaluate the comprehensive ability of supervision personnel in providing services; professional ability dimension weight proportion 25%, emphasizing the level of supervision personnel in decoration professional knowledge and skills; customer complaint dimension weight proportion 10%, reflecting the customer dissatisfaction with supervision services; customer satisfaction dimension weight proportion 5%, mainly measuring the degree of customer satisfaction with supervision services; NPS score dimension weight proportion 5%, evaluating the spread and influence of supervision services in the customer group through the net recommendation value; customer praise dimension weight proportion 5%, reflecting the degree of customer recognition and appreciation for supervision personnel services; customer return single dimension weight proportion 20%, focusing on the enthusiasm of customer feedback on supervision services and the degree of attention to service details.

[0049] With the continuous operation of the system, in order to more accurately reflect the work performance of supervision personnel, the weight of each dimension is dynamically adjusted according to the weakness strengthening principle. Specifically, the adjustment range of the weight of a single dimension is set to +30% of the upper limit and -30% of the lower limit. For example, if it is found that a supervision personnel has a significantly low score in the customer complaint dimension at a certain stage, indicating that there is a large improvement space in this dimension, according to the weakness strengthening principle, the system will appropriately increase the weight of the customer complaint dimension, with a maximum increase of 30%; conversely, if the score of a certain dimension is consistently good, the weight of that dimension may be appropriately reduced, but at most by 30%. Through this dynamic adjustment method, the evaluation system can be more flexible and accurate to adapt to the work characteristics and actual situation of different supervision personnel, ensuring the scientificity and fairness of the evaluation results.

[0050] (Three) Evaluation and feedback layer: 1. Multi-dimensional evaluation output: the system evaluates supervision personnel from eight dimensions based on data analysis results, outputs detailed evaluation report, and clearly defines the performance of supervision personnel in each dimension and the overall service level.

[0051] The causal reasoning analysis aims to accurately output the evaluation results of the supervisors. Through in-depth analysis of the collected data, the internal causal relationship between the data and the supervisors' ability is mined. Specifically, the multi-dimensional data covering the supervisor's performance, service ability, professional ability, customer complaints, customer satisfaction, NPS score, customer praise, and customer return sheet are comprehensively analyzed to understand how each factor interacts and affects the supervisor's work performance.

[0052] Based on this causal reasoning process, it can be clearly identified that the supervisors' ability in which aspects needs to be strengthened or which work methods need to be improved. For example, if the data analysis shows that the number of customer complaints is closely related to the supervisor's mastery of professional knowledge, and the number of complaints is high, it indicates that the supervisor may need to strengthen the relevant professional ability.

[0053] 2. Improvement direction generation: based on the evaluation results, analyze the problems in each dimension, and determine the key influencing factors based on the causal analysis to generate specific improvement directions and suggestions for the supervisors.

[0054] After the analysis is completed, the results of the causal reasoning are output in the form of a report. The report can intuitively present the current work status and ability level of the supervisors, and also clearly indicate the directions that need to be improved. It provides targeted and operable guidance suggestions for the supervisors to improve the quality of work and optimize the work methods, thereby helping to continuously improve the quality of supervision services.

[0055] 3. Supervision and improvement: through the supervisor's APP and management means, the evaluation results and improvement directions are pushed to the supervisors in real time, and they are supervised to learn and improve through task reminders and training course recommendations. The management layer can monitor the improvement progress through the management platform to improve the service quality.

[0056] As an option, in step S5, the specific steps of the causal reasoning analysis combined with the deep feature engineering framework include: Algorithm selection and model construction: select the causal reasoning algorithm framework - Random Forest. Take the eight-dimensional feature vector as the input feature, and take the supervision service effect (such as comprehensive evaluation score, customer satisfaction level, etc.) as the output label to construct the Random Forest model. Each decision tree in the model is constructed based on bootstrapsampling from the original data set with replacement.

[0057] Model training: train the Random Forest model using the prepared training data. During the training process, each decision tree randomly selects some features for the best split point search when splitting the node. Through the construction of multiple decision trees, the complex relationship between each feature and the supervision service effect in the data is learned.

[0058] Causal relationship analysis: After the model training is completed, the importance of each feature in the decision tree and the dependency between the feature and the output are analyzed to determine the impact of each factor on the quality of supervision services. For example, by observing which features are frequently used in the splitting process of the decision tree and how the changes of these features affect the final prediction results, the causal relationship between these features and the supervision service effect is inferred. For example, if it is found that the number of customer complaints plays a key role in predicting customer satisfaction level in the decision tree, it indicates that there is a strong causal relationship between the number of customer complaints and customer satisfaction.

[0059] Result output: The results of causal reasoning analysis are output, including the importance ranking of each feature, the quantitative evaluation of the impact on the quality of supervision services, and other information. The report is presented in the form of a report to provide intuitive information to supervision personnel and management, helping them understand which factors have a key impact on the supervision service effect, so as to develop targeted improvement measures. For example, the report indicates that the professional knowledge examination results in the professional competence dimension have a significant positive impact on the improvement of the quality of supervision services, and suggests strengthening the training and examination of the professional knowledge of supervision personnel.

[0060] As shown in Figure 2 , the data processing flow of the present application is as follows: Data collection stage: When a new supervision service project is started, the system automatically collects relevant data from various data sources, such as supervision logs, customer feedback texts, decoration engineering data, etc., and aggregates them to a distributed data processing platform.

[0061] Data analysis stage: On the distributed data processing platform, the data is analyzed in sequence by multi-modal data fusion processing and sentiment analysis module. Key information is mined to determine the impact of each factor on supervision services, providing real-time data support for evaluation.

[0062] Evaluation and feedback stage: According to the data analysis results, the system generates an eight-dimensional evaluation report to determine the improvement direction of the supervision personnel. Through the APP and management platform, it is conveyed to the supervision personnel to start the supervision mechanism, track the improvement process, and form a closed-loop management.

[0063] As a preferred embodiment, in step S6, the specific steps of continuously iterating and updating the evaluation results based on real-time data streams and dynamic model adjustment mechanism include: Real-time capture of new supervision records, customer feedback and engineering progress data through a distributed platform, real-time analysis of preprocessed data using stream computing, monitoring of the instantaneous state of eight-dimensional feature indicators, and triggering of automatic early warning when an abnormal threshold is detected; Start the parameter adaptive system, dynamically adjust the eight-dimensional feature weights and causal reasoning model parameters according to the real-time index changes, and migrate the historical project knowledge to the new project scenario through the dynamic weight migration learning mechanism; The updated model parameters are re-substituted into the deep feature engineering framework to generate an updated evaluation report containing the latest indicator data and causal relationships.

[0064] Specifically, the real-time dynamic updating step of the embodiments of the present application includes real-time dynamic updating at the data level and real-time dynamic updating at the model level.

[0065] I. Real-time dynamic updating at the data level: 1. Real-time data collection: The system maintains real-time connection with various data sources through a distributed data processing platform. For example, it interacts with supervision log recording systems, customer feedback platforms, and engineering progress monitoring systems in real time. When new supervision work records are generated, customers submit new feedback, or engineering progress data is updated, the platform immediately captures these data. This ensures that the data obtained by the system is always up-to-date, reflecting the actual situation of supervision work in a timely manner.

[0066] 2. Fast data preprocessing: After the newly collected data enters the system, it is first subjected to fast preprocessing. Feature alignment algorithms are used to unify data from different sources in terms of time, space, and business dimensions, such as time format, geographic coordinate systems, etc. At the same time, temporal and spatial correlation algorithms are used to mine the potential relationships between data in time and space. In addition, real-time data cleaning is performed to remove duplicate, erroneous, or incomplete data records, and labeled according to pre-set rules, providing a high-quality data foundation for subsequent analysis.

[0067] II. Real-time dynamic updating at the model level: 1. Real-time analysis using stream computing: Stream computing technology is used to analyze the pre-processed data in real time. For example, real-time monitoring of changes in indicators such as task completion rate, average task processing time, etc. in the supervision efficiency dimension. By calculating these indicators in real time, the system can immediately understand the current work efficiency of the supervision personnel, and compare it with historical data and pre-set standards. Once an abnormal fluctuation in the indicators is detected, such as a sudden and significant decrease in task completion rate, the system immediately issues a warning signal.

[0068] 2. Model parameter dynamic adjustment: Based on real-time analysis results, the system starts the parameter adaptive system. Taking the eight-dimensional feature generator in the deep feature engineering framework as an example, when it is found that the influence of a certain dimension of feature on the supervision service effect changes, the system automatically adjusts the weight of the dimension feature. For example, if the customer complaint dimension has a more significant impact on the overall service quality in a certain period, the system will appropriately increase the weight of the customer complaint dimension within the allowed adjustment range (the upper and lower limits of the adjustment range for a single dimension weight are +30% and -30%, respectively) according to the dynamic weight distribution mechanism. Similarly, for models used for causal reasoning analysis (such as random forest models), the parameters of the model, such as the number of decision trees and node splitting conditions, will also be adjusted in real time according to the characteristics of new data to ensure that the model can always accurately analyze the causal relationship between data and supervision personnel capabilities.

[0069] 3. Dynamic weight transfer learning: When new project data flows in, the system uses dynamic weight transfer learning mechanism. First, the new project data is analyzed to identify its similarities and differences with historical project data. Then, the general knowledge and model parameters trained in previous projects are retained, and local fine-tuning is performed according to the characteristics of the new project. For example, the new project may differ from previous projects in terms of project type, customer group, etc. The system will adjust the parameters related to these characteristics in the model to make the model quickly adapt to the needs of the new project, maintain the accuracy and effectiveness of the evaluation.

[0070] Embodiment 2: As shown in Figure 4 The embodiment of the present application provides a decoration supervision service quality optimization system based on eight-dimensional feature evaluation, which comprises: A multi-source heterogeneous data collection module is used to collect multi-source heterogeneous data related to supervision services through a distributed data processing platform, including structured supervision logs, unstructured text reports, and time-series engineering data; A data preprocessing module is used to integrate multi-source heterogeneous data using feature alignment and spatio-temporal association algorithms, and to clean and label the multi-source heterogeneous data; A multi-modal data fusion module uses neural network-based cross-modal semantic mapping technology to associate text descriptions and numerical indicators in high-dimensional space, constructs a three-dimensional data cube covering personnel, projects, and environment, and deeply fuses multi-source heterogeneous data; The deep feature engineering module includes an eight-dimension feature generator, a text sentiment analysis submodule, and an adaptive feature screening submodule, the eight-dimension feature generator generates an eight-dimension feature vector related to supervision efficiency, service quality, and satisfaction based on a dynamic weight distribution mechanism, the text sentiment analysis submodule performs sentiment grading on customer feedback text using a Transformer architecture, and the adaptive feature screening submodule identifies key influencing factors of each dimension through causal reasoning; The real-time dynamic updating module is used for adopting a mixed architecture of stream computing and batch processing, combining a dynamic weight adjustment mechanism to realize real-time data updating and model parameter adaptation; The multi-dimension evaluation output module generates an evaluation report containing eight-dimension specific performance and overall service level based on the results of deep feature engineering; The improvement direction generation module analyzes each dimension problem according to the evaluation results and generates targeted improvement suggestions; The supervision and monitoring module pushes the evaluation results and improvement suggestions to the supervision personnel in real time through a mobile terminal and tracks the improvement progress.

[0071] Preferably, the eight-dimension feature vector in the embodiment of the application includes: A supervision personnel efficiency dimension for measuring the ability and efficiency of the supervision personnel to complete effective supervision work in a unit of time; A service ability dimension for measuring the comprehensive ability exhibited by the supervision personnel in providing supervision services; A professional ability dimension for evaluating the level of the supervision personnel in decoration professional knowledge and skills; A customer complaint dimension for reflecting the dissatisfaction of the customer with the supervision services; A customer satisfaction dimension for measuring the satisfaction degree of the customer with the supervision services; An NPS score dimension for evaluating the propagation force and influence of the supervision services in the customer group; A customer praise dimension for embodying the recognition and appreciation degree of the customer for the supervision personnel services; A customer return single dimension for measuring the feelings of the customer for the supervision services and reflecting the willingness of the customer to recommend new customers.

[0072] Embodiment 3: The embodiment of the application provides an electronic device, including a processor and a memory storing a computer program, and the processor implements the eight-dimension feature evaluation-based decoration supervision service quality optimization method provided in embodiment 1 of the application when executing the computer program.

[0073] Embodiment 4: The embodiment of the present application provides a computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the eight-dimension feature evaluation based decoration supervision service quality optimization method provided in the embodiment 1 of the present application.

[0074] Embodiment 5: The embodiment of the present application provides a server, comprising at least one processor, and a memory connected with the processor, wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the processor to enable the at least one processor to execute the eight-dimension feature evaluation based decoration supervision service quality optimization method provided in the embodiment 1 of the present application.

[0075] It should be understood that, although each step in the flowchart of each embodiment of the present application is displayed in sequence according to the arrow, these steps are not necessarily executed in sequence according to the arrow. Unless otherwise specified in the present application, the execution of these steps is not strictly limited in sequence, and these steps can be executed in other sequences. Moreover, at least part of the steps in each embodiment can include multiple sub-steps or multiple stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution sequence of these sub-steps or stages is not necessarily sequential, but can be executed in rotation or alternation with at least part of other steps or sub-steps or stages of other steps.

[0076] It can be understood by those skilled in the art that all or part of the processes in the above-mentioned embodiment methods can be completed by a computer program instructing related hardware, and the program can be stored in a non-volatile computer readable storage medium. When the program is executed, it can include the processes of the above-mentioned embodiments of each method. Any reference to memory, storage, database or other medium used in each embodiment of the present application can include non-volatile and / or volatile memory. Non-volatile memory can include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory can include random access memory (RAM) or external cache memory. As an illustration but not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), Rambus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).

[0077] Finally, it should be noted that the above only describes the preferred embodiments of the present application and is not intended to limit the present application. Although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art will appreciate that modifications can be made to the technical solutions described in the foregoing embodiments, or some of the technical features thereof can be replaced by equivalent features. Any modifications, equivalent replacements, improvements, and the like made within the spirit and principle of the present application should be included in the protection scope of the present application.

Claims

1. A method for optimizing the quality of decoration supervision services based on eight-dimensional feature evaluation, characterized in that, The method includes: (1) Collect multi-source heterogeneous data related to supervision services through a distributed data processing platform, including structured supervision logs, unstructured text reports and time-series engineering data; (2) Feature alignment and spatiotemporal correlation algorithms are used to integrate multi-source heterogeneous data, and the multi-source heterogeneous data are cleaned and labeled. (3) Using neural network-based cross-modal semantic mapping technology, text descriptions and numerical indicators are correlated in a high-dimensional space to construct a three-dimensional data cube covering personnel, projects and environment, and to deeply fuse multi-source heterogeneous data; (4) Generate an eight-dimensional feature vector related to supervision efficiency, service quality and satisfaction based on a dynamic weight allocation mechanism and text sentiment analysis. The eight-dimensional feature vector includes the supervisor efficiency dimension, service capability dimension, professional capability dimension, customer complaint dimension, customer satisfaction dimension, NPS score dimension, customer praise dimension and customer return order dimension. (5) Combine the deep feature engineering framework to conduct causal reasoning analysis, generate a detailed evaluation report containing eight dimensions of specific performance and overall service level, analyze the problems in each dimension based on the evaluation results, and generate targeted improvement suggestions; (6) The evaluation results are continuously updated based on real-time data streams and dynamic model adjustment mechanisms; (7) The evaluation results and improvement suggestions are pushed to the supervisors in real time through mobile terminals and the improvement progress is tracked to form a closed-loop management.

2. The method for optimizing the quality of decoration supervision services based on eight-dimensional feature evaluation according to claim 1, characterized in that, The quantitative indicators for the supervisor's efficiency dimension include the amount completed per unit time and the average on-site time; the quantitative indicators for the service capability dimension include the hazard follow-up rate and communication and coordination ability assessment; the quantitative indicators for the professional capability dimension include professional knowledge assessment scores and log writing professionalism; the quantitative indicators for the customer complaint dimension include the number of complaints, the proportion of complaint problem categories, and complaint problem sentiment analysis; the quantitative indicators for the customer satisfaction dimension include the follow-up visit satisfaction score and the mini-program satisfaction score; the quantitative indicators for the customer praise dimension include the number of praises and the praise content classification; and the quantitative indicators for the customer return order dimension include the return order rate and return order type score. The calculation method for the NPS score dimension is as follows: NPS = (Number of recommenders / Total sample size) × 100% - (Number of detractors / Total sample size) × 100%.

3. The method for optimizing the quality of decoration supervision services based on eight-dimensional feature evaluation according to claim 1, characterized in that, In step (5), the specific steps for causal reasoning analysis using the deep feature engineering framework include: Algorithm selection and model construction: Random forest was selected as the causal inference algorithm framework, with eight-dimensional feature vectors as input and supervision service effect indicators as output labels. A random forest model was constructed based on the bootstrap sampling method. Model training: The model is trained using training data. Each decision tree randomly selects some features when splitting at a node to find the best split point. The complex relationship between the features and the effect in the learning data is learned by integrating multiple decision trees. Causal relationship analysis: Based on the trained model, by analyzing the frequency of feature usage in decision tree splitting and the degree of influence of feature changes on prediction results, the influence weight of each feature on the effect of supervision services is quantified to determine key causal relationships; Output results include a ranking of feature importance and a quantitative assessment report on the impact on the quality of supervision services.

4. The method for optimizing the quality of decoration supervision services based on eight-dimensional feature evaluation according to claim 1, characterized in that, In step (6), the specific steps for continuously iteratively updating the evaluation results based on real-time data streams and dynamic model adjustment mechanisms include: The distributed platform captures new supervision records, customer feedback and project progress data in real time, and uses streaming computing to analyze the preprocessed data in real time, monitors the instantaneous status of eight-dimensional characteristic indicators, and triggers automatic warnings when abnormal thresholds are detected. The parameter adaptive system is activated to dynamically adjust the weights of the eight-dimensional features and the parameters of the causal inference model based on real-time indicator changes, and to transfer historical project knowledge to new project scenarios through a dynamic weight transfer learning mechanism. The updated model parameters are then fed back into the deep feature engineering framework to generate an updated evaluation report that includes the latest indicator data and causal relationships.

5. A system for optimizing the quality of decoration supervision services based on eight-dimensional feature evaluation, characterized in that, The system includes: The multi-source heterogeneous data collection module is used to collect multi-source heterogeneous data related to supervision services through a distributed data processing platform, including structured supervision logs, unstructured text reports, and time-series engineering data. The data preprocessing module is used to integrate multi-source heterogeneous data using feature alignment and spatiotemporal correlation algorithms, and to clean and label the multi-source heterogeneous data. The multimodal data fusion module uses neural network-based cross-modal semantic mapping technology to associate text descriptions with numerical indicators in a high-dimensional space, constructing a three-dimensional data cube covering people, projects, and environment, and deeply fusing multi-source heterogeneous data. The deep feature engineering module includes an eight-dimensional feature generator, a text sentiment analysis submodule, and an adaptive feature selection submodule. The eight-dimensional feature generator generates eight-dimensional feature vectors related to supervision efficiency, service quality, and satisfaction based on a dynamic weight allocation mechanism. The text sentiment analysis submodule uses the Transformer architecture to perform sentiment classification on customer feedback text. The adaptive feature selection submodule identifies key influencing factors in each dimension through causal reasoning. The real-time dynamic update module is used to achieve real-time data updates and adaptive model parameters by adopting a hybrid architecture of streaming computing and batch processing, combined with a dynamic weight adjustment mechanism. The multi-dimensional evaluation output module generates an evaluation report based on the results of deep feature engineering, which includes eight dimensions of specific performance and overall service level. The improvement direction generation module analyzes the problems in each dimension based on the evaluation results and generates targeted improvement suggestions; The monitoring module pushes evaluation results and improvement suggestions to supervisors in real time via mobile terminals and tracks the progress of improvements.

6. The decoration supervision service quality optimization system based on eight-dimensional feature evaluation according to claim 4, characterized in that, The eight-dimensional feature vector includes: The supervisor efficiency dimension is used to measure the ability and efficiency of supervisors in completing effective supervisory work within a unit of time. The service capability dimension is used to measure the comprehensive capabilities demonstrated by supervisors in providing supervisory services; The professional competence dimension is used to assess the level of professional knowledge and skills of the supervisors in the decoration industry; The customer complaint dimension is used to reflect customer dissatisfaction with the supervision services. Customer satisfaction is a metric used to measure customer satisfaction with the supervision services. The NPS score dimension is used to assess the reach and influence of the supervision service among the client group; Customer praise dimension is used to reflect the degree of recognition and appreciation that customers have for the services provided by the supervisors. The customer feedback dimension is used to measure customers' feelings about the supervision services and reflect their willingness to recommend new customers.

7. An electronic device comprising a processor and a memory storing a computer program, characterized in that, When the processor executes the computer program, it implements the method for optimizing the quality of decoration supervision services based on eight-dimensional feature evaluation as described in any one of claims 1 to 4.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the method for optimizing the quality of decoration supervision services based on eight-dimensional feature evaluation as described in any one of claims 1 to 4.

9. A server, characterized in that: It includes at least one processor and a memory communicatively connected to the processor, the memory storing instructions executable by the at least one processor, the instructions being executed by the processor to cause the at least one processor to perform the decoration supervision service quality optimization method based on eight-dimensional feature evaluation as described in any one of claims 1 to 4.

Images