A manual service management method and system

Abstract

The present invention relates to the field of service management technology, and in particular to a manual service management method and system. The method comprises the following steps: obtaining customer demand data; obtaining service personnel data, and matching demand according to customer personnel data and customer demand data to obtain service personnel matching data; performing service scheduling according to service personnel matching data to obtain service scheduling data, and performing task allocation according to service scheduling data to obtain service task allocation data; performing service process monitoring according to service task allocation data to obtain service process monitoring data, so as to perform manual service management auxiliary operations. The present invention accurately captures customer needs, especially potential needs, matches service personnel, improves resource utilization, and continuously optimizes service processes and innovates business models through data-driven.

Description

Technical Field

[0001] The present invention relates to the technical field of service management, and in particular to a manual service management method and system. Background Art

[0002] With the urbanization process and the increasing intensity of modern work, the manual service industry (such as massage, health therapy, foot therapy, etc.) has gradually emerged. However, as the scale of the industry increases, efficiency problems of traditional manual service management have begun to emerge. Traditional service management methods mainly rely on manual operations or simple static rule design. Traditional methods rely on customers filling out forms or verbal descriptions, which makes it difficult to fully capture the real needs of customers, especially when customers fail to clearly express their needs. There is a lack of close coordination between different links, and the management process relies on manual intervention, which is prone to information islands and inefficiency. Summary of the invention

[0003] In order to solve the above technical problems, the present invention proposes a manual service management method and system to solve at least one of the above technical problems.

[0004] The present application provides a manual service management method, the method comprising:

[0005] S1. Obtain customer demand data;

[0006] S2. Obtain service personnel data, and perform demand matching according to customer personnel data and customer demand data to obtain service personnel matching data;

[0007] S3. Perform service scheduling according to the service personnel matching data to obtain service scheduling data, and perform task allocation according to the service scheduling data to obtain service task allocation data;

[0008] S4. Monitor the service process according to the service task allocation data to obtain service process monitoring data for manual service management auxiliary operations.

[0009] The present invention can accurately capture the explicit needs of customers (such as service type, duration, technician gender) and implicit needs (such as preferences and special requirements) through a systematic data acquisition process, providing high-quality data support for matching and scheduling. Multimodal input (such as text, voice or click behavior) is used to improve the flexibility and coverage of demand acquisition and adapt to the usage habits of different customers. Combined with automated verification and structured processing, the integrity and standardization of input data are ensured to reduce the impact of human errors. By integrating data such as service personnel's skills, experience and free time, resources are reasonably allocated to avoid technicians being idle or overloaded, and the efficiency of human resource utilization is improved. Algorithm-based intelligent matching methods (such as feature weight matching or rule matching) are adopted to maximize customer needs and improve customer experience. The status of service personnel is updated in real time to ensure that the matching process has dynamic adjustment capabilities to adapt to real-time demand changes. Service scheduling and task allocation are automatically completed by the system, reducing manual intervention and greatly improving work efficiency and accuracy. Combined with a multi-objective optimization algorithm, the tasks of service personnel are reasonably allocated to balance workloads and avoid resource waste or service delays. The scheduling system dynamically adjusts task allocation according to real-time data to improve the response speed to emergencies. Through real-time monitoring of the service process (such as technician behavior, customer feedback, completion time, etc.), we ensure that the service quality meets expectations and reduce service quality fluctuations. Monitoring data provides a quantitative basis for improving the service process, supporting service process optimization and technician training. Abnormal situations in the service process (such as delays, customer complaints, etc.) are discovered in a timely manner and quickly corrected to improve service stability.

[0010] Optionally, the customer demand data includes customer conventional demand data and customer potential demand data, and S1 includes:

[0011] S11. Obtain basic customer data;

[0012] S12. Perform regular demand analysis based on customer basic data to obtain customer regular demand data;

[0013] S13. Conduct potential demand analysis based on customer basic data to obtain customer potential demand data.

[0014] The present invention supports multi-channel and multi-modal data input (such as mobile applications, online forms or voice interactions) to ensure the flexibility and coverage of data collection. Customer basic data can be dynamically updated over time (such as historical service records increase with the number of services) to provide the latest and most relevant data for demand analysis. By analyzing customer basic data, explicit needs (such as service type, time preference, technician gender, etc.) are mined to achieve rapid and accurate extraction of conventional needs. Conventional demand data generated based on customer historical service records and preferences provide direct support for personalized recommendations and service matching to improve customer satisfaction. With the help of computing technology (such as machine learning and graph neural networks), implicit needs that are not clearly expressed by customers but exist are analyzed to expand service scenarios. For example, by analyzing service frequency, preferences for specific services, etc., it is predicted that customers may be interested in new service items. Potential demand analysis helps companies provide more forward-looking service solutions, thereby forming a differentiated competitive advantage and improving customer stickiness. Through the dual analysis of conventional and potential needs, it can not only meet the current explicit needs of customers, but also tap into implicit needs to form a complete customer demand portrait.

[0015] Optionally, S12 includes:

[0016] S121, extracting multi-dimensional basic features according to the customer basic data to obtain multi-dimensional basic feature data of the customer;

[0017] S122, generating dynamic labels according to the multi-dimensional basic feature data of the customer to obtain dynamic label data of the customer;

[0018] S123, obtaining all customer basic data from the database;

[0019] S124, performing cluster calculation based on all customer basic data and customer dynamic tag data to obtain customer tag demand tendency data;

[0020] S125, extracting behavioral features from the multi-dimensional basic feature data of the customer to obtain user behavioral feature data;

[0021] S126, performing behavior feature fusion according to the user behavior feature data to obtain user demand behavior fusion data;

[0022] S127, deriving service industry rules based on the fusion data of user demand behavior to obtain user regular demand tendency data;

[0023] S128. Obtain customer historical data corresponding to the customer basic data;

[0024] S129: Perform historical correlation inference based on the customer's historical data and the user's regular demand tendency data to obtain the customer's regular demand data.

[0025] The present invention utilizes multi-dimensional feature extraction technology to cover basic information such as customer attributes, preferences, and behaviors, and provide high-quality input. Dynamic label generation ensures that customer labels are updated in real time with time and behavior changes, reflecting the current status of customers. Labeled feature data helps to analyze customer needs more finely and lay the foundation for personalized services. By uniformly extracting all customer data, the globality and accuracy of the analysis are ensured to avoid one-sided decisions. Cluster analysis reveals the group tendencies of customer needs and provides data support for market segmentation and precision services. Mining customers' implicit behavior patterns from multi-dimensional features reveals behavioral tendencies that are not clearly expressed. Behavioral feature fusion transforms isolated behaviors into comprehensive behavior patterns, reflecting the overall demand tendencies of customers. Combined with service industry rules, it is deduced whether customer needs meet existing service standards and improves the industry adaptability of demand derivation. Obtaining customer historical data supports time series analysis and identifies demand change trends. Historical correlation inference combines time and behavior patterns to dynamically predict customers' regular needs.

[0026] Optionally, S13 includes:

[0027] S131, extracting implicit features according to the customer basic data to obtain implicit feature data;

[0028] S132, predicting points of interest based on implicit feature data to obtain point of interest prediction data;

[0029] S133, performing cross-domain knowledge association according to the interest point prediction data to obtain cross-domain knowledge association data;

[0030] S134, deriving service behavior according to the cross-domain knowledge association data to obtain service behavior derivation data;

[0031] S135. Generate situational demand based on service behavior derivation data to obtain customer potential demand data.

[0032] Implicit feature extraction in the present invention goes beyond explicit data analysis, and uses embedded modeling methods (such as Autoencoder and knowledge graph embedding) to mine deep attributes that customers have not explicitly expressed but are hidden in the data, such as consumption potential, unexplored areas of interest, etc. The prediction data of interest points is updated in real time based on customer behavior, enabling the system to dynamically adjust recommendations and service strategies to adapt to rapidly changing needs. Through cross-domain knowledge graph association, customer interest points are linked to multiple service areas (such as health management, social activities, etc.). For example, the customer's potential demand for massage meditation is inferred from the customer's preference for massage services. Cross-domain knowledge association effectively matches customer needs with enterprise resources, such as linking needs with specific technician skills, venue resources, etc. through prediction to optimize service arrangements. Combined with cross-domain knowledge, the possible service behavior path of the customer is derived. For example, a specific environment preference (such as a quiet room) or a specific technician demand is derived from the interest in health massage. Scenario-based demand generation combines the customer's behavior derivation results with the real-time environment (such as time, weather, festivals, etc.) to provide customers with demand suggestions that meet the current scenario. For example, push warm massage services in cold weather.

[0033] Optionally, S131 includes:

[0034] Extract customer text data and customer behavior data based on customer basic data to obtain customer text data and customer behavior data respectively;

[0035] Perform historical feature correlation mining based on customer text data and customer behavior data to obtain correlation feature data;

[0036] Perform multimodal feature fusion on the correlation feature data to obtain fused feature data;

[0037] Perform feature rarity evaluation on the fused feature data to obtain rare feature data;

[0038] Hidden features are generated for the rare feature data to obtain hidden feature data.

[0039] The present invention extracts customer text data (such as document input, service comments, questionnaire feedback, etc.) and behavioral data (such as click records, purchase frequency, etc.) to capture customers' explicit needs and preferences. The implicit relationship between text data and behavioral data is mined to reveal the potential pattern of customer needs. Text features and behavioral features are integrated into a unified data representation to eliminate the limitations of isolated data and enhance analysis consistency. Through Zipf distribution or frequency analysis, rare features in customer data are mined to reveal customers' unique needs and preferences. High-frequency but valueless features are filtered out during the evaluation process to reduce data noise. Implicit features can reflect customers' unexpressed demand tendencies and enhance the system's ability to predict customers' potential needs.

[0040] Optionally, S133 includes:

[0041] Perform graph mapping according to the preset cross-domain knowledge graph model and the POI prediction data to obtain initial graph mapping data;

[0042] Obtain basic service data;

[0043] Perform service semantic expansion based on the initial graph mapping data and service basic data to obtain service semantic expansion data;

[0044] Perform cross-domain relationship mining based on service semantic extension data to obtain service cross-domain relationship data;

[0045] Perform relationship embedding learning based on cross-domain relationship data to obtain service embedding relationship data;

[0046] Potential association inference is performed based on service embedding relationship data to obtain cross-domain knowledge association data.

[0047] The cross-domain knowledge graph in the present invention unifies different fields (such as customer interest points and service options) in a logical structure, constructs a basic association framework, and breaks the limitations of single-field analysis. Service basic data (such as service type, technician skills, service time, etc.) are used as supplements to enrich the content of the graph to provide more details. Service semantic extension mines potential service scenarios by analyzing the semantic relationship between interest points and service basic data, such as expanding from "relaxation" to "meditation service". Cross-domain relationship mining uses graph search algorithms (such as random walk algorithms) and path optimization to reveal the potential connection between customer needs and services, such as from "health therapy" to "nutrition consultation". Through embedded learning (such as TransE, RotatE), high-dimensional and complex cross-domain relationships are converted into easy-to-compute low-dimensional vector representations to improve computing efficiency. The inference results can provide data basis for enterprises to design new service combinations, such as expanding from "massage + yoga" to "massage + meditation".

[0048] Optionally, S2 includes:

[0049] S21. Obtain service personnel data;

[0050] S22, performing service category stratification according to the customer demand data to obtain demand service category stratification data;

[0051] S23, extracting demand keywords according to the demand service category hierarchical data to obtain demand keyword data;

[0052] S24, performing semantic analysis on the demand keyword data to obtain demand semantic data;

[0053] S25, performing service type preference mapping and service environment preference mapping according to the demand semantic data to obtain service type preference data and service environment preference data;

[0054] S26. Matching is performed according to the customer personnel data, the service type preference data, and the service environment preference data to obtain service personnel matching data.

[0055] The service personnel data in the present invention includes multi-dimensional data such as skills, experience, and available time periods, which provide a basis for accurate matching. Layering customer needs according to service types (such as massage, physical therapy, foot therapy, etc.) helps to clarify the scope of needs and reduce interference from irrelevant services. Extract core keywords (such as "deep massage" and "quiet environment") from the demand description to improve the accurate understanding of customer needs. Through semantic analysis, the contextual meaning of customer needs is extracted from keywords to reveal implicit needs. Service type preference mapping identifies customers' inclinations towards specific service content (such as inclination towards deep massage or relaxing massage), while environmental preference mapping focuses on customers' requirements for the service environment (such as quiet, warm, etc.). Combining multi-source data (customer personnel data, service type preferences, environmental preferences), accurate matching of customer needs with technicians' skills and styles is achieved.

[0056] Optionally, S3 includes:

[0057] S31. Decompose the service tasks according to the service personnel matching data to obtain customer service task data;

[0058] S32, performing dispatch path planning on the customer service task data to obtain service dispatch data;

[0059] S33. Perform time allocation according to the service scheduling data to obtain service task allocation data.

[0060] The present invention decomposes complex service requirements into detailed task units (such as skill matching and service duration calculation), and clarifies the execution objectives and responsibilities of each task. Based on the scheduling path planning data, the time of each task is reasonably allocated to ensure that the service personnel's time utilization is maximized while meeting the time requirements of customer needs. Time allocation optimizes the workload of service personnel and avoids task backlogs or resource waste caused by time conflicts.

[0061] Optionally, S4 includes:

[0062] S41, generating service monitoring parameters according to the service task allocation data to obtain service monitoring parameter data;

[0063] S42, collecting service real-time data according to the service monitoring parameter data to obtain service real-time data;

[0064] S43. Perform manual service management auxiliary operations based on real-time service data.

[0065] In the present invention, monitoring parameters for different service tasks (such as service duration, technician behavior, customer feedback, etc.) are dynamically generated according to service task allocation data to ensure that the monitoring scope accurately covers key links. Real-time service data (such as service start time, service quality, technician behavior, etc.) are collected through IoT devices (such as cameras, wearable devices) and client feedback tools to fully grasp the service process. Real-time service data can promptly detect anomalies that occur during the service process (such as timeouts, customer dissatisfaction, etc.), buy time for problem handling, and avoid a decline in service quality. Based on real-time data, the management system can quickly identify specific problems in the service process (such as technician operation deviations or customer feedback dissatisfaction) and achieve precise intervention.

[0066] Optionally, the present application further provides a manual service management system for executing the manual service management method as described above, the manual service management system comprising:

[0067] Customer demand data collection module, used to obtain customer demand data;

[0068] The service personnel matching module is used to obtain service personnel data and perform demand matching according to customer personnel data and customer demand data to obtain service personnel matching data;

[0069] A service task allocation module is used to perform service scheduling according to the service personnel matching data to obtain service scheduling data, and to perform task allocation according to the service scheduling data to obtain service task allocation data;

[0070] The service process monitoring module is used to monitor the service process according to the service task allocation data and obtain the service process monitoring data to perform manual service management auxiliary operations.

[0071] The purpose of the present invention is that the present invention covers the whole process from customer demand acquisition to task allocation and service monitoring, and realizes end-to-end coordination of manual service management. Each link is both independent and highly connected, forming an efficient closed-loop management system. Demand acquisition is combined with service personnel matching to form a dynamic adjustment mechanism for supply and demand, avoiding the problem of data islands in traditional manual service management. The sub-steps of obtaining customer demand data (such as demand stratification, keyword extraction, and semantic analysis) can accurately identify customers' explicit needs and implicit preferences. Service personnel matching uses multi-dimensional data (such as skills, experience, free time, etc.) to achieve accurate customer demand and personnel skill matching. Service scheduling and task allocation decompose complex service requirements into executable task units, while balancing the load of service personnel's human resources to avoid excessive concentration or waste of resources. Service process monitoring realizes full visualization, from task initiation to service completion, monitoring service quality, time and customer feedback. Real-time monitoring data can quickly discover anomalies (such as service timeouts, customer complaints, etc.), and adjust tasks in time through auxiliary operation functions. BRIEF DESCRIPTION OF THE DRAWINGS

[0072] Other features, objects and advantages of the present application will become more apparent by reading the detailed description of non-limiting implementations made with reference to the following drawings:

[0073] Figure 1 A flowchart showing the steps of a manual service management method according to an embodiment is shown;

[0074] Figure 2 A flowchart showing the steps of a method for collecting customer demand data according to an embodiment is shown;

[0075] Figure 3 A flowchart showing the steps of a service personnel matching method according to an embodiment is shown;

[0076] Figure 4 A flowchart showing the steps of a service task allocation method according to an embodiment is shown;

[0077] Figure 5 A flowchart showing a method for monitoring a service process according to an embodiment of the present invention is provided;

[0078] The realization of the purpose, functional features and advantages of the present invention will be further explained in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0079] The following is a clear and complete description of the technical method of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by technicians in this field without creative work are within the scope of protection of the present invention.

[0080] In addition, the accompanying drawings are only schematic illustrations of the present invention and are not necessarily drawn to scale. The same reference numerals in the figures represent the same or similar parts, and their repeated description will be omitted. Some of the block diagrams shown in the accompanying drawings are functional entities and do not necessarily correspond to physically or logically independent entities. The functional entities can be implemented in software form, or implemented in one or more hardware modules or integrated circuits, or implemented in different networks and / or processor methods and / or microcontroller methods.

[0081] It should be understood that, although the terms "first", "second", etc. may be used herein to describe various units, these units should not be limited by these terms. These terms are used only to distinguish one unit from another unit. For example, without departing from the scope of the exemplary embodiments, the first unit may be referred to as the second unit, and similarly the second unit may be referred to as the first unit. The term "and / or" used herein includes any and all combinations of one or more of the listed associated items.

[0082] See also Figures 1 to 5 , the present application provides a manual service management method, the method comprising:

[0083] S1. Obtain customer demand data;

[0084] In one embodiment, the customer submits a demand form through a terminal device (such as a mobile application or website), including the service type, service duration, time, and special requirements (such as technician gender and skill level). The system extracts the customer's preference data (such as service frequency and commonly used technicians) from historical service data. , Service Complexity , Customer Level , calculate demand priority, ; For demand priority, is the urgency weight of the demand time, is the service complexity weight, It is the customer level weight. Add high priority requirements to the task queue.

[0085] S2. Obtain service personnel data, and perform demand matching according to customer personnel data and customer demand data to obtain service personnel matching data;

[0086] In one embodiment, the system obtains relevant data from the service personnel database, including the technician's skill category, current status (free or busy), location, and historical evaluation. Technicians who meet the customer service category requirements are selected from the technician database. Technicians are sorted according to their scores (technicians with high scores are given priority).

[0087] S3. Perform service scheduling according to the service personnel matching data to obtain service scheduling data, and perform task allocation according to the service scheduling data to obtain service task allocation data;

[0088] In one embodiment, technicians who are available are given priority and it is checked whether the technicians have enough time to complete the task. The system compares the timetable to ensure that the assigned task time period does not conflict with existing tasks. The shift is optimized based on the technician score and task time requirements. Detailed task data is generated for each customer demand, including: service time period; technician ID; service type.

[0089] S4. Monitor the service process according to the service task allocation data to obtain service process monitoring data for manual service management auxiliary operations.

[0090] In one embodiment, the technician signs in at the beginning of the task through a handheld device or an in-store terminal. During the service process, the technician updates the task status (such as "service started", "service completed") through the system. If the task service time exceeds 20% of the expected duration, the system generates an abnormal reminder. Check whether the technician signs in and completes the task on time. If the task is not completed, a reminder will be sent to the management end. After the task is completed, the customer scores and evaluates the technician's service. The system automatically records the score and updates the technician's service quality indicator.

[0091] Optionally, the customer demand data includes customer conventional demand data and customer potential demand data, and S1 includes:

[0092] S11. Obtain basic customer data;

[0093] In one embodiment, the customer's historical registration information, membership information and past order records are obtained from the in-store system database, including: basic personal information (such as name, gender, contact information); membership level and account status (such as VIP, ordinary customer); historical service data (such as service type, frequency, order amount, evaluation).

[0094] S12. Perform regular demand analysis based on customer basic data to obtain customer regular demand data;

[0095] In one embodiment, regular demand refers to the service demand that is explicitly expressed by the customer or repeatedly selected in historical orders. The number of occurrences of each service type is counted from the customer's historical service data. The service preference is calculated by combining the evaluation score and service frequency. The service types that meet the following conditions are selected as regular demand: service frequency >0.5 (used at least once a month); preference ≥4.0 (high customer rating).

[0096] S13. Conduct potential demand analysis based on customer basic data to obtain customer potential demand data.

[0097] In one embodiment, potential demand is a possible demand that is not explicitly expressed by the customer but is mined through historical behavior data or association rules. Extract the customer's service time preferences (such as morning and evening peaks). Calculate the similarity with other customers based on the customer's service frequency and preferences. Count the high-frequency service types that are not selected in the customer group with high similarity to infer the potential demand of the current customer. Generate customer potential demand data, including recommended service categories and reasons (such as "80% of similar customers choose this service").

[0098] Optionally, S12 includes:

[0099] S121, extracting multi-dimensional basic features according to the customer basic data to obtain multi-dimensional basic feature data of the customer;

[0100] In one embodiment, the basic features include demographic features, service preference features, geographic location features, historical consumption features, etc. Multidimensional features are extracted from the customer basic data: demographic features: gender, age, membership level; service preference features: service category selection frequency, evaluation score; historical consumption features: average consumption amount, total consumption amount.

[0101] S122, generating dynamic labels according to the multi-dimensional basic feature data of the customer to obtain dynamic label data of the customer;

[0102] In one embodiment, dynamic tags are dynamically generated based on customer characteristics for classification and matching. For example: activity tags: high activity, medium activity, low activity; preference tags: preference for specific service categories. Customer classification tags are calculated based on feature data, such as the number of orders divided by month / quarter / year, and the high, medium and low activity levels are divided by thresholds. Service category preferences are processed similarly, with the corresponding service categories divided by time and the preference levels divided by thresholds.

[0103] S123, obtaining all customer basic data from the database;

[0104] In one embodiment, the customer database is connected to extract basic data fields of all customers.

[0105] S124, performing cluster calculation based on all customer basic data and customer dynamic tag data to obtain customer tag demand tendency data;

[0106] In one embodiment, based on the basic characteristics and dynamic tags of customers, customers are grouped by similarity calculation or clustering calculation to identify groups with similar demand tendencies. According to the similarity matrix, customers are divided into K groups to obtain the demand tendency labels of each group.

[0107] S125, extracting behavioral features from the multi-dimensional basic feature data of the customer to obtain user behavioral feature data;

[0108] In one embodiment, behavior-related features are extracted from basic features of customers, such as order time period distribution and service frequency.

[0109] S126, performing behavior feature fusion according to the user behavior feature data to obtain user demand behavior fusion data;

[0110] In one embodiment, multiple behavior features are fused into a unified feature representation in a weighted manner.

[0111] S127, deriving service industry rules based on the fusion data of user demand behavior to obtain user regular demand tendency data;

[0112] In one embodiment, the customer's regular demand tendency is derived based on the association of fused data and industry rules. For example, if a customer frequently chooses massage services and the service time is concentrated in the evening, "nighttime deep massage" is derived as his regular demand.

[0113] S128. Obtain customer historical data corresponding to the customer basic data;

[0114] In one embodiment, customer behavior data is extracted from historical order records, including service categories, evaluation records, order amounts, etc.

[0115] S129: Perform historical correlation inference based on the customer's historical data and the user's regular demand tendency data to obtain the customer's regular demand data.

[0116] In one embodiment, based on customer historical data and conventional demand trend data , infer the actual needs of customers. Get the regular demand data of customers .

[0117] Optionally, S13 includes:

[0118] S131, extracting implicit features according to the customer basic data to obtain implicit feature data;

[0119] In one embodiment, implicit features are customer characteristics that cannot be directly observed but are derived through basic data analysis, such as consumption habits, potential preferences, etc. Keywords are extracted from customer reviews and messages, and the importance is calculated using word frequency statistics. Behavioral data (such as consumption frequency, service time preferences) are extracted from customer order records. Text features and behavioral features are fused into implicit features according to weights.

[0120] S132, predicting points of interest based on implicit feature data to obtain point of interest prediction data;

[0121] In one embodiment, the services or behaviors that the customer is interested in are inferred based on the implicit features, such as new service categories or high-frequency services. The customer's interest score for each potential service is calculated through linear regression based on the implicit features, and the scores are sorted by score. >The threshold service is used as the point of interest to obtain the point of interest prediction data .

[0122] S133, performing cross-domain knowledge association according to the interest point prediction data to obtain cross-domain knowledge association data;

[0123] In one embodiment, the interest points are associated with other domain knowledge to mine more potential service information. For example, by associating industry trends, additional services that customers may need are predicted. The interest point data is combined with the industry knowledge base (such as service category association, market demand) to build a knowledge graph. The correlation between the interest point node and the industry knowledge node is calculated, and the high correlation nodes are selected as potential service association items to obtain cross-domain knowledge association data.

[0124] S134, deriving service behavior according to the cross-domain knowledge association data to obtain service behavior derivation data;

[0125] In one embodiment, the service behavior taken by the customer is inferred based on the cross-domain knowledge association data, such as booking a new service or adjusting the service frequency. The service behavior probability is calculated based on the association data through probability calculation or multivariate regression calculation. If the point of interest and the industry knowledge base show that "the demand for new service A has increased among similar customers", it is inferred that the customer chooses new service A.

[0126] In one embodiment, more importantly, environment modeling is performed based on cross-domain knowledge association data to obtain a cross-domain knowledge environment model; behavior classification is performed on customer basic data based on the cross-domain knowledge environment model to obtain customer behavior classification data; scenario matching is performed on the customer behavior classification data to obtain scenario matching data; service behavior hierarchical decision is performed on the scenario matching data to obtain service behavior derivation data;

[0127] The service behavior derivation data includes first service behavior derivation data and second service behavior derivation data, and the service behavior hierarchical decision includes:

[0128] Performing hierarchical decision making on the first service behavior on the scenario matching data to obtain first service behavior derivation data;

[0129] A method for performing a second service behavior hierarchical decision on the scenario matching data to obtain second service behavior derivation data, wherein the first service behavior hierarchical decision and the second service behavior hierarchical decision are different.

[0130] The first service behavior hierarchical decision includes:

[0131] Performing service category mapping on the scenario matching data to obtain service category mapping data;

[0132] Calculating the service demand priority of the service category mapping data to obtain first service behavior derivation data;

[0133] The second service behavior hierarchical decision includes:

[0134] Perform global behavior deduction on the scenario matching data to obtain global deduction data of service behavior;

[0135] Perform local behavior optimization on the scenario matching data to obtain local derivation data of service behavior;

[0136] Constructing a service tree according to the global derivation data of the service behavior and the local derivation data of the service behavior to obtain second service behavior derivation data;

[0137] The system builds an environmental model through existing cross-domain knowledge association data. For example, environmental dependency information of specific service categories is extracted from the industry knowledge graph, such as the impact of weather on service demand, changes in holiday service types, etc. Based on cross-domain knowledge analysis, the demand distribution of different service categories under specific environmental conditions is quantified and modeled. The system generates a cross-domain knowledge environment model to describe the association between service categories and environmental factors (such as time, weather, geographic location, etc.). The system uses the cross-domain knowledge environment model to analyze the customer's historical data (such as order records, behavioral characteristics): Customer A's data shows that he prefers short massages on weekdays and deep foot therapy on weekends; Customer B's data shows that his demand is greatly affected by the season, and he prefers hot stone massage in winter. The system classifies customer behavior based on these historical data: Customer A is classified as "time-sensitive"; Customer B is classified as "seasonal adaptation".

[0138] The system matches customer behavior classification data with current situational information (such as weather and time): the current time is Saturday morning and the weather is sunny; customer A is "time-sensitive", and the system infers that he may prefer short-term relaxation services; customer B is "seasonally adaptable", and since the weather is sunny, the system infers that he may choose outdoor-related services.

[0139] After scenario matching, the system generates a set of service demand matching data for the current scenario. Based on the scenario matching data, the system makes hierarchical decisions on the matching results. The specific process includes: First-level decision: Determine the service category. The system analyzes the scenario matching data and recommends shoulder and neck massage based on the historical preferences of customer A and the current time period. For customer B, based on his environmental adaptation behavior, the recommended service category is foot therapy. Second-level decision: Optimize service details. The system further refines the specific options within the service category: For customer A, it is recommended to choose "shoulder and neck relaxation massage (30 minutes)"; for customer B, it is recommended to choose "foot care combined with hot stones (60 minutes)".

[0140] The system integrates the decision results to generate complete service behavior derivation data: the derivation data for customer A includes: the service category is shoulder and neck massage, and the service details are 30 minutes of relaxing massage; the derivation data for customer B includes: the service category is foot therapy, and the service details are hot stone foot therapy.

[0141] For the first service behavior hierarchical decision, the system receives scenario matching data, which includes the customer's current demand information (such as the need for relaxation services), scenario conditions (such as weekday evenings), and customer historical behavior data (such as choosing massage services many times in the past). The system analyzes the customer's scenario matching data and matches it with the options in the service category standard library: the customer's historical behavior shows that he tends to choose "shoulder and neck relaxation" or "foot massage" in similar scenarios; combined with the current weekday evening time period, the system infers that the customer needs a fast, fatigue-relieving service. After the system prioritizes the service categories, it selects the highest priority service category: the customer is recommended "massage service" as his service category. The first service behavior derivation data generated includes the recommended service category "massage service" and indicates the customer's current preference background and historical behavior support.

[0142] The system receives the first service behavior derivation data and limits the service category to "massage service". The system further analyzes the customer's scenario matching data to optimize the specific details of the service: in historical data, customers gave a high score to the "shoulder and neck massage" item; the customer's free time on the current weekday evening is 30 minutes, which is suitable for short-term service; the customer's most recent choice of "hot stone therapy" was good, but the preference is for a technician with moderate intensity. The system further deduces "shoulder and neck massage" as the main recommended option, and combined with the customer's special preferences, it recommends that the technician choose a person with "hot stone massage" skills, and the time is set to 30 minutes. Generate the second service behavior derivation data, including the recommended service as "shoulder and neck massage (hot stone therapy, 30 minutes)", and the supporting basis for the recommendation.

[0143] First, in the implementation of hierarchical decision making for service behavior, the system receives scenario matching data, including the customer's current status (e.g., Friday night, tired, needing to relax quickly) and historical service records (e.g., preferring massage services, having a high opinion of foot therapy). The system maps the customer scenario data with the standard service category library and preliminarily selects the service categories: the customer's records in similar scenarios indicate that they tend to choose "massage" or "foot therapy" as their preferred service. The system generates service category mapping data, marking "massage service" and "foot therapy service" as candidate categories.

[0144] Based on the mapping results, the system evaluates the demand priority of each candidate category: "Massage service" has a high priority because customers are more concerned about quickly recovering from physical fatigue; "Foot massage service" has a slightly lower priority, but is suitable for long-term service needs. The system determines "massage service" as the highest priority recommended category. Output the first service behavior derivation data, including: Recommended service category: "massage service"; derivation basis: supported by the customer's current scenario and historical preferences.

[0145] The implementation of the second service behavior hierarchical decision-making, global behavior deduction, the system limits the service category to "massage service" based on the first service behavior deduction data. Analyze customer needs from a global dimension: customers need a short and efficient physical recovery, and "shoulder and neck massage" appears frequently in historical data; the current time period is Friday night, and the system believes that soothing services are more suitable for this scenario. The system derives the global deduction data of service behavior and recommends "shoulder and neck massage". Local behavior optimization, based on the global deduction results, further optimize the service details: the customer's historical orders show that he prefers shoulder and neck massage combined with hot stone therapy; the customer's evaluation record mentions that he prefers technicians with moderate strength; the current time allows customers to choose a 30-minute service. The system generates local deduction data of service behavior and recommends "shoulder and neck massage (hot stone therapy, 30 minutes)". Service tree construction, the system combines global and local deduction data to construct a service behavior tree: the root node is the service category "massage service"; the first-level node is the specific direction "shoulder and neck massage"; the second-level node is the optimization solution "hot stone therapy, 30 minutes". Output the second service behavior derivation data, including a service behavior tree and a recommended complete service solution.

[0146] S135. Generate situational demand based on service behavior derivation data to obtain customer potential demand data.

[0147] In one embodiment, the derived service behavior is combined with customer-specific scenarios (such as time, holidays) to generate personalized demand data. Scenario variable definition: time dimension (weekdays / weekends, holidays); environmental dimension (season, weather). Scenario matching logic: Build a scenario demand model based on behavior data and scenario variables , To meet the needs of the situation, Derive data for service behavior, is the scenario weight matrix. Combined with historical data, it generates recommended requirements (such as "recommended additional technician services during holidays").

[0148] Optionally, S131 includes:

[0149] Extract customer text data and customer behavior data based on customer basic data to obtain customer text data and customer behavior data respectively;

[0150] In one embodiment, unstructured text data such as customer service evaluation, message records, and notes. The text is segmented to remove stop words and meaningless words. Keywords are extracted using statistical methods, the word frequency (TF) and inverse document frequency (IDF) are calculated, the keywords are mapped to vector representations, and text features are generated. The customer's historical order records (service category, amount, frequency, etc.) are counted by time period, the frequency is calculated, and the mean and variance of each order amount are counted to obtain customer behavior data.

[0151] Perform historical feature correlation mining based on customer text data and customer behavior data to obtain correlation feature data;

[0152] In one embodiment, the feature correlation is calculated using a covariance method, feature pairs whose absolute correlation values ​​are greater than a threshold are screened, and correlation feature data is generated.

[0153] Perform multimodal feature fusion on the correlation feature data to obtain fused feature data;

[0154] In one embodiment, features of different modalities (text and behavior) in the correlation feature data are fused into a unified feature representation.

[0155] Perform feature rarity evaluation on the fused feature data to obtain rare feature data;

[0156] In one embodiment, rare features refer to features that are uncommon in all customer data but have unique value. The probability distribution of the fused features is calculated using Gaussian distribution. If it is less than a threshold, it is considered a rare feature. The rarity score of the feature (1-probability value obtained by probability distribution) is calculated to obtain rare feature data.

[0157] Hidden features are generated for the rare feature data to obtain hidden feature data.

[0158] In one embodiment, rare features are combined into implicit features to reflect the special needs or potential preferences of customers. Frequently occurring feature pairs or feature combinations are selected from the rare features. The regularity of rare feature combinations is detected, such as whether they are concentrated in a certain service category or time period.

[0159] Optionally, S133 includes:

[0160] Perform graph mapping according to the preset cross-domain knowledge graph model and the POI prediction data to obtain initial graph mapping data;

[0161] In one embodiment, the knowledge graph is composed of nodes (entities, such as service type, time period) and edges (relationships, such as "includes", "related"), which represent the association information between different fields. The predicted data of the points of interest correspond to specific nodes in the graph. The similarity is calculated based on the features of the predicted data of the points of interest and the nodes in the graph, and the potential relationship edges between the points of interest and the target nodes are mapped to the initial graph data.

[0162] Obtain basic service data;

[0163] In one embodiment, the service information in the system includes service classification, service description, historical evaluation, etc. The service category is mapped to a standard category number; and the time period characteristics of the service (such as peak period) are extracted.

[0164] Perform service semantic expansion based on the initial graph mapping data and service basic data to obtain service semantic expansion data;

[0165] In one embodiment, the graph mapping node is combined with the service basic data to expand the semantic information of the service node. According to the service basic data, the keyword attribute is added to the graph node (in logical expression, it is a union relationship). According to the keyword similarity, the edges between the nodes are added to obtain the service semantic extension data.

[0166] Perform cross-domain relationship mining based on service semantic extension data to obtain service cross-domain relationship data;

[0167] In one embodiment, potential associations between nodes in different fields are mined (such as the implicit connection between service categories and customer behaviors). Nodes and edges in semantic extension data are clustered to identify high-frequency cross-domain patterns. Association rules with high support and confidence are extracted through frequent pattern mining to obtain service cross-domain relationship data.

[0168] Perform relationship embedding learning based on cross-domain relationship data to obtain service embedding relationship data;

[0169] In one embodiment, cross-domain relational data is mapped to a low-dimensional vector space to facilitate model learning and inference.

[0170] Potential association inference is performed based on service embedding relationship data to obtain cross-domain knowledge association data.

[0171] In one embodiment, unknown potential cross-domain associations are inferred based on embedded relationship data.

[0172] Optionally, S2 includes:

[0173] S21. Obtain service personnel data;

[0174] In one embodiment, the service personnel's personal profile data, skill level, historical service records, available time periods, ratings and evaluations, etc. are stored in a service personnel database.

[0175] S22, performing service category stratification according to the customer demand data to obtain demand service category stratification data;

[0176] In one embodiment, customer demand data is classified and hierarchically divided according to service types. Hierarchical rules: First-level classification: major categories of services (such as "massage", "foot therapy"). Second-level classification: detailed to service types (such as "shoulder and neck massage", "full body massage"). According to the service category field in the demand, it is mapped with the service category standard library.

[0177] S23, extracting demand keywords according to the demand service category hierarchical data to obtain demand keyword data;

[0178] In one embodiment, keywords related to the service category are extracted from the customer demand text (such as notes, selection tags), the customer demand text is segmented and stop words are removed, and high-weight keywords are extracted using the word frequency statistics (TF-IDF) method.

[0179] S24, performing semantic analysis on the demand keyword data to obtain demand semantic data;

[0180] In one embodiment, the semantic information of the demand keywords is analyzed to understand the customer's preference and specific demand for the service. The keywords are converted into vector representations through a semantic model. The semantic vectors are matched with the semantic vectors in the service category tag library, and the similarity is calculated to obtain the demand semantic data.

[0181] S25, performing service type preference mapping and service environment preference mapping according to the demand semantic data to obtain service type preference data and service environment preference data;

[0182] In one embodiment, the customer's specific preference for the service type is mapped according to the demand semantic data. The service type preference score is calculated by combining the demand semantic data and the customer's historical order records through weight calculation. The customer's demand preference for the service environment (such as room type, background music) is mapped. The environment keywords are analyzed according to the customer's notes and historical records, and the service environment preference data is obtained through full connection calculation or activation calculation.

[0183] S26. Matching is performed according to the customer personnel data, the service type preference data, and the service environment preference data to obtain service personnel matching data.

[0184] In one embodiment, a comprehensive match is performed between the customer personnel data (such as the preferred technician gender and experience level) and the preference data to select a suitable service personnel. A matching score is calculated for each service personnel, and the service personnel with the highest score are selected according to the ranking of the scores.

[0185] Optionally, S3 includes:

[0186] S31. Decompose the service tasks according to the service personnel matching data to obtain customer service task data;

[0187] In one embodiment, after associating the customer's service demand with the matching service personnel data, it is refined into specific service tasks. Each task contains customer demand details, service personnel information and time requirements. According to the customer service demand and service personnel matching data, the task information is initialized. If the demand contains multiple service types or time periods, it is divided into independent tasks. The tasks are integrated according to the service personnel's skill coverage to obtain customer service task data.

[0188] S32, performing dispatch path planning on the customer service task data to obtain service dispatch data;

[0189] In one embodiment, in a single location (such as a store), the path planning goal is to optimize the order of task execution, reduce waiting time, and improve service efficiency. The tasks are initially sorted according to their priorities, and the sorting is adjusted to minimize the total waiting time. An execution path is generated for each service staff, including the task order and time arrangement.

[0190] S33. Perform time allocation according to the service scheduling data to obtain service task allocation data.

[0191] In one embodiment, a specific time period is allocated to each task according to the task scheduling order and the available time window of the service personnel. The overlap between the service personnel's available time and the task time requirement is checked. If there is no intersection, the task order is adjusted or other service personnel are assigned. Check whether there is a time conflict. If the time periods of task i and task j overlap, they are reallocated. Minimize the total idle time of task allocation and obtain service task allocation data, including the start time, end time and service personnel of each task.

[0192] Optionally, S4 includes:

[0193] S41, generating service monitoring parameters according to the service task allocation data to obtain service monitoring parameter data;

[0194] In one embodiment, monitoring parameters are used to define key monitoring points in the service process, including: service personnel status (check-in, completion status); service time (estimated start and end time); service execution details (such as task links). Monitoring-related fields are extracted from the service task allocation data to set normal ranges for monitoring parameters, such as the allowable range of time deviation. According to the service type of task allocation, a corresponding monitoring parameter template is generated, which is applicable to different service scenarios.

[0195] S42, collecting service real-time data according to the service monitoring parameter data to obtain service real-time data;

[0196] In one embodiment, dynamic data during the service process is collected for real-time monitoring of service execution. Service personnel clock in through the terminal device at the beginning of the service and upload service status and location data. During the service process, data points are collected regularly, including: current time; service status (such as "in progress" or "paused"). The data collection frequency is set to f=1 / T (for example, once every 1 minute). After the service is completed, data such as completion time and customer feedback are uploaded.

[0197] S43. Perform manual service management auxiliary operations based on real-time service data.

[0198] In one embodiment, the service process is dynamically optimized or exceptions are handled based on real-time data to assist manual service management decisions. If monitoring parameters detect an anomaly (such as timeout or service interruption), the system generates an alarm and recommends adjusting the task. Task priorities are adjusted or tasks are reallocated based on the service status. For example, if a service task is delayed, the system recalculates the time window for subsequent tasks. Real-time data is aggregated to generate service reports, including: service personnel efficiency; customer satisfaction; and anomaly rate.

[0199] Optionally, the present application further provides a manual service management system for executing the manual service management method as described above, the manual service management system comprising:

[0200] Customer demand data collection module, used to obtain customer demand data;

[0201] The service personnel matching module is used to obtain service personnel data and perform demand matching according to customer personnel data and customer demand data to obtain service personnel matching data;

[0202] A service task allocation module is used to perform service scheduling according to the service personnel matching data to obtain service scheduling data, and to perform task allocation according to the service scheduling data to obtain service task allocation data;

[0203] The service process monitoring module is used to monitor the service process according to the service task allocation data and obtain the service process monitoring data to perform manual service management auxiliary operations.

[0204] Therefore, from any point of view, the embodiments should be regarded as illustrative and non-restrictive, and the scope of the present invention is limited by the attached application documents rather than the above description, and it is intended that all changes falling within the meaning and scope of equivalent elements of the application documents are included in the present invention.

[0205] The above description is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.

Claims

1. A manual service management method, characterized in that: The method comprises: S1. Obtain customer demand data; S2. Obtain service personnel data, and perform demand matching according to customer personnel data and customer demand data to obtain service personnel matching data; S3. Perform service scheduling according to the service personnel matching data to obtain service scheduling data, and perform task allocation according to the service scheduling data to obtain service task allocation data; S4. Monitor the service process according to the service task allocation data to obtain service process monitoring data for manual service management auxiliary operations; The customer demand data includes customer regular demand data and customer potential demand data. S1 includes: S11. Obtain basic customer data; S12. Perform regular demand analysis based on customer basic data to obtain customer regular demand data; S13, analyzing potential needs based on basic customer data to obtain customer potential needs data; S12 includes: S121, extracting multi-dimensional basic features according to the customer basic data to obtain multi-dimensional basic feature data of the customer; S122, generating dynamic labels according to the multi-dimensional basic feature data of the customer to obtain dynamic label data of the customer; S123, obtaining all customer basic data from the database; S124, performing cluster calculation based on all customer basic data and customer dynamic tag data to obtain customer tag demand tendency data; S125, extracting behavioral features from the multi-dimensional basic feature data of the customer to obtain user behavioral feature data; S126, performing behavior feature fusion according to the user behavior feature data to obtain user demand behavior fusion data; S127, deriving service industry rules based on the fusion data of user demand behavior to obtain user regular demand tendency data; S128. Obtain customer historical data corresponding to the customer basic data; S129, performing historical association inference based on the customer's historical data and the user's conventional demand tendency data to obtain the customer's conventional demand data; S3 includes: S31. Decompose the service tasks according to the service personnel matching data to obtain customer service task data; S32, performing dispatch path planning on the customer service task data to obtain service dispatch data; S33. Perform time allocation according to the service scheduling data to obtain service task allocation data.

2. The method according to claim 1, characterized in that S13 includes: S131, extracting implicit features according to the customer basic data to obtain implicit feature data; S132, predicting points of interest based on implicit feature data to obtain point of interest prediction data; S133, performing cross-domain knowledge association according to the interest point prediction data to obtain cross-domain knowledge association data; S134, deriving service behavior according to the cross-domain knowledge association data to obtain service behavior derivation data; S135. Generate situational demand based on service behavior derivation data to obtain customer potential demand data.

3. The method according to claim 2, characterized in that S131 includes: Extract customer text data and customer behavior data based on customer basic data to obtain customer text data and customer behavior data respectively; Perform historical feature correlation mining based on customer text data and customer behavior data to obtain correlation feature data; Perform multimodal feature fusion on the correlation feature data to obtain fused feature data; Perform feature rarity evaluation on the fused feature data to obtain rare feature data; Hidden features are generated for the rare feature data to obtain hidden feature data.

4. The method according to claim 2, characterized in that: S133 includes: Perform graph mapping according to the preset cross-domain knowledge graph model and the POI prediction data to obtain initial graph mapping data; Obtain basic service data; Perform service semantic expansion based on the initial graph mapping data and service basic data to obtain service semantic expansion data; Perform cross-domain relationship mining based on service semantic extension data to obtain service cross-domain relationship data; Perform relationship embedding learning based on cross-domain relationship data to obtain service embedding relationship data; Potential association inference is performed based on service embedding relationship data to obtain cross-domain knowledge association data.

5. The method according to claim 1, characterized in that S2 include: S21. Obtain service personnel data; S22, performing service category stratification according to the customer demand data to obtain demand service category stratification data; S23, extracting demand keywords according to the demand service category hierarchical data to obtain demand keyword data; S24, performing semantic analysis on the demand keyword data to obtain demand semantic data; S25, performing service type preference mapping and service environment preference mapping according to the demand semantic data to obtain service type preference data and service environment preference data; S26. Matching is performed according to the customer personnel data, the service type preference data, and the service environment preference data to obtain service personnel matching data.

6. The method according to claim 1, characterized in that S4 include: S41, generating service monitoring parameters according to the service task allocation data to obtain service monitoring parameter data; S42, collecting service real-time data according to the service monitoring parameter data to obtain service real-time data; S43. Perform manual service management auxiliary operations based on real-time service data.

7. A manual service management system, characterized in that: Used to execute the manual service management method according to claim 1, the manual service management system comprises: Customer demand data collection module, used to obtain customer demand data; The service personnel matching module is used to obtain service personnel data and perform demand matching according to customer personnel data and customer demand data to obtain service personnel matching data; The service task allocation module is used to perform service scheduling according to the service personnel matching data to obtain service scheduling data, and to perform task allocation according to the service scheduling data to obtain service task allocation data; wherein the service scheduling is performed according to the service personnel matching data to obtain service scheduling data, and the task allocation is performed according to the service scheduling data to obtain service task allocation data, including: performing service task decomposition according to the service personnel matching data to obtain customer service task data; performing scheduling path planning on the customer service task data to obtain service scheduling data; performing time allocation according to the service scheduling data to obtain service task allocation data; The service process monitoring module is used to monitor the service process according to the service task allocation data and obtain the service process monitoring data to perform manual service management auxiliary operations.

Images