Customer service order management electronization system

Abstract

The invention relates to the technical field of work order service management, and provides a customer service order management electronic system which comprises a telephone terminal, an input terminal, a computer terminal and a customer center server. The telephone terminal receives a repair request of a user and automatically inputs incoming call information; the input terminal collects maintenance process data and generates multimedia data with a timestamp; the client center server is in butt joint with the block chain evidence storage platform, generates a hash value from the maintenance process data, performs real-time uplink evidence storage through an API interface, and solidifies evidence storage time by using a time stamp of a national time service center; and the computer terminal automatically distributes the work order to the corresponding maintenance personnel based on the skill label of the maintenance personnel and the type of the work order, updates the state of the work order in real time in the client center server, and executes online acceptance check and return operation.

Description

Technical Field

[0001] This invention relates to the technical field of work order service management, and in particular to an electronic system for customer service order management. Background Technology

[0002] Currently, heating maintenance services mostly rely on paper-based maintenance records, which often suffer from issues such as improper completion and insufficient evidence retention. After a maintenance service, it's inconvenient for users to trace the previous maintenance results when they file a complaint again. In some cases, no photographic or video evidence is retained, and the system is heavily influenced by subjective factors, making it difficult to verify the quality of the maintenance and the veracity of the user's complaints. Summary of the Invention

[0003] This application proposes an electronic customer service order management system, which uses blockchain and timestamp-based evidence storage to update the status of work orders during heating maintenance. This system not only facilitates the reflection of maintenance quality and results but also improves the efficiency of work order allocation.

[0004] To achieve the above objectives, this application provides the following technical solution:

[0005] In the first aspect, this application proposes an electronic customer service order management system, including a telephone terminal, a data entry terminal, a computer terminal, and a customer center server;

[0006] The telephone terminal receives user repair requests and automatically records incoming call information;

[0007] The data entry terminal collects maintenance process data and generates multimedia data with timestamps.

[0008] The customer center server connects to the blockchain evidence storage platform, generates hash values ​​for the repair process data, uploads it to the blockchain in real time via API interface, and uses the timestamp of the National Time Service Center to solidify the evidence storage time.

[0009] The computer terminal automatically assigns work orders to the corresponding maintenance personnel based on the maintenance personnel's skill tags and work order types, and updates the work order status in real time in the customer center server, performing online acceptance and return operations.

[0010] In conjunction with the first aspect, when the telephone terminal receives a user's repair request, it initiates an identity verification process: first, it extracts the user's voice features using voiceprint recognition technology and compares them with the system's pre-stored voiceprint database;

[0011] If the match fails, a dynamic verification code will be automatically sent to the user's registered mobile phone, and preset security questions will be asked through an interactive voice response mechanism.

[0012] Obtain the text of questions and answers, assess the semantic similarity of the answer texts, and comprehensively judge the authenticity of the identity based on the user's historical behavior habits;

[0013] If an abnormal behavior pattern is detected, the verification strength is increased.

[0014] In conjunction with the first aspect, when the input terminal collects maintenance process data, it generates metadata tags containing data feature values, collection timestamps, and terminal device identifiers, calculates a data digest using the SM3 cryptographic hash algorithm, and transmits it to the customer service center; wherein, the maintenance process data includes on-site photos, temperature measurement records, and user confirmation information during the maintenance process;

[0015] After the customer center server receives the data, it verifies the integrity of the metadata tags. If the feature value is found to be inconsistent with the local calculation result, it immediately sends a retransmission instruction to the input terminal and activates the locally cached historical version data for temporary replacement.

[0016] When abnormal characteristic values ​​are found, a verification anomaly record is generated and uploaded to the blockchain evidence storage platform in real time, marked as pending verification, until the evidence storage status is updated after manual review and confirmation.

[0017] In conjunction with the first aspect, the customer center server automatically triggers an anomaly level assessment when it detects that the work order processing time exceeds a preset threshold or that the same user submits work orders of the same fault type within a preset time.

[0018] Anomaly level assessment includes:

[0019] Analyze the matching degree between work order type and historical processing data. If the matching degree is lower than the first matching value, mark it as an anomaly to be confirmed.

[0020] The anomalies to be confirmed are analyzed using natural language processing technology to analyze the emotional tendencies in the work order descriptions, identify the characteristics of dissatisfaction when users submit repeatedly, and combine this with the on-site difficulty reported by maintenance personnel in real time to comprehensively determine the anomaly level. The anomaly levels include Level 1 anomalies that represent risky faults, Level 2 anomalies that represent parts scheduling, and Level 3 anomalies that represent operational errors.

[0021] When a Level 1 anomaly occurs, the regular work order allocation is suspended, the abnormal work order is pinned to the top and pushed to the technical supervisor's terminal;

[0022] When a Level 2 anomaly occurs, the backup supplier scheduling process will be automatically initiated.

[0023] When a Level 3 anomaly is detected, a guided solution is generated and announced via voice broadcast on a telephone terminal.

[0024] In conjunction with the first aspect, the computer terminal is used to collect maintenance task data from maintenance personnel in real time and construct a three-dimensional evaluation model; wherein, the maintenance task data includes work order load data, skill tag matching data, and geographical location data;

[0025] When the 3D evaluation model receives the first work order, its first priority is determined.

[0026] Based on the initial priority, the real-time sorting data of the initial priority of the second work order is used to determine the second priority of the first work order.

[0027] In conjunction with the first aspect, when the input terminal and the computer terminal simultaneously revise the real-time data of the same work order, the revision action is determined, and the terminal identifier and the vector clock tag of the logical timestamp of the revision action are determined; wherein,

[0028] The vector clock tags are compared in a partial order via the customer center server to determine whether a revision causal relationship exists.

[0029] When a revised causal relationship exists, the conflict resolution rules are activated.

[0030] When there is no causal relationship for revision, determine the key fields for revision, and use these key fields to determine the terminal priority of the same work order.

[0031] In conjunction with the first aspect, the computer terminal is also used for:

[0032] Real-time collection of maintenance personnel's GPS location, work order address, and traffic data; calculation of the shortest service path using the Dijkstra algorithm.

[0033] When at least two similar work orders are detected in the same area within a preset time, they are automatically merged and scheduled, and prioritized for allocation to the nearest and skill-matched target maintenance personnel, while simultaneously pushing the optimal transportation route.

[0034] In conjunction with the first aspect, after the customer center server receives the user acceptance confirmation information uploaded by the input terminal and the computer terminal synchronizes the work order completion status, the contract automatically verifies the data integrity, calls the blockchain platform through the API interface to complete the hash value on the chain, synchronously generates the national time service center timestamp, and pushes the evidence certificate to the user terminal.

[0035] In conjunction with the first aspect, when the customer center server receives updated work order data from the input terminal, it pushes a status change notification to the associated computer terminal via the WebSocket protocol, triggering a local cache refresh.

[0036] If the computer terminal is offline, the server will temporarily store the change records in the message queue, and synchronize them in timestamp order after the terminal reconnects.

[0037] In conjunction with the first aspect, the customer center server adopts a multi-node backup architecture; wherein,

[0038] When the master node detects a hardware failure or data corruption, the system automatically switches to the standby node, elects a new master node using the Raft consensus algorithm, and initiates the data repair process.

[0039] Other features and advantages of the invention will be set forth in the following description, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description and the accompanying drawings.

[0040] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0041] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof.

[0042] In the attached diagram:

[0043] Figure 1 This is a system composition diagram of an electronic customer service order management system according to an embodiment of the present invention;

[0044] Figure 2 This is a system composition diagram of a traditional electronic customer service order management system in an embodiment of the present invention;

[0045] Figure 3 This is a flowchart illustrating the authentication process in an embodiment of the present invention.

[0046] Figure 4 This is a diagram illustrating the data entry process for maintenance procedures in an embodiment of the present invention.

[0047] Figure label:

[0048] 10 is the customer service terminal, 20 is the customer center service terminal, 30 is the first maintenance management server, 40 is the second maintenance management server, 50 is the maintenance dispatch terminal, 60 is the maintenance employee terminal, 70 is the maintenance processing input terminal, and 80 is the user terminal.

[0049] 100 is a telephone terminal, 200 is a data entry terminal, 300 is a computer terminal, and 400 is a customer center server. Detailed Implementation

[0050] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0051] The electronic customer service order management system refers to a work order approval system used for gas fault repairs. This system can take various forms, such as a client application, software, a webpage, or a mini-program. It is used to report fault information and generate repair requests when a gas fault occurs. Then, through the gas company's customer service center, a work order is generated, and repair personnel are dispatched to carry out the repairs. The electronic customer service order management system can also be understood as a virtual device, program code, and functional modules.

[0052] See Figure 2 , Figure 2 This is an architecture diagram of an electronic system for managing traditional customer service for gas companies. It receives repair requests from users via telephone or through convenient service points deployed in many areas of the city through user terminal 80. Then, customer service terminal 10 records gas fault events (this system can be integrated with electronic systems for gas recharge, gas account opening, etc., but this is not described in this application). Customer service terminal 10 uploads the recorded repair information and user information to customer center server terminal 20 to generate task items. The task items are then uploaded to the first maintenance management server 30, which generates maintenance work orders based on dispatched work orders, different workers' working hours, and work order execution status. After receiving the maintenance work order at maintenance dispatch terminal 50, work order information is generated and sent to the maintenance worker's maintenance employee terminal 60. Maintenance employee terminal 60 executes the maintenance task based on the work order information. During the maintenance process, maintenance information is entered through maintenance processing input terminal 70, which can be a specially configured handheld terminal or a mobile terminal such as a mobile phone. Upon completion of a repair task, or if an anomaly such as inability to repair is discovered during the repair process, a work order is generated and uploaded to the second repair management server 40. The second repair management server 40, upon completion of the repair task, performs work order settlement and closure, and uploads the results to the first repair management server 30 and the customer center server 20, respectively. If any anomalies are found in the work order feedback information, the first repair management server 30 will determine whether to reassign the work order to an employee capable of handling it, or add a worker with a separate work order signature.

[0053] Currently, when processing maintenance tasks through the aforementioned electronic system, client 10 receives maintenance requests uploaded by user client 80 and automatically dispatches employees based on the maintenance information uploaded by client 80. However, it cannot determine whether the employee's skill tags and the corresponding tasks for the work order type match. For example, in the case of a refrigerant leak, if the dispatched employee is a regular employee rather than a senior maintenance employee with special operating qualifications, it is easy to repeatedly dispatch the same work order and switch between different employees. Or, the dispatched employee may directly handle tasks beyond their scope of expertise, leading to workplace injuries.

[0054] When client-side 10 records repair tasks from client-side 80, errors in work order dispatch are prone to occur because client-side 10 employees need to input text. Currently, after repair workers complete a repair, they take photos of the repaired scene, but it's impossible to determine the repair standards or the authenticity of parts replacement repairs, which can easily lead to customer skepticism, complaints, and employee errors. Furthermore, repair data is stored on a local server, allowing administrators to arbitrarily modify operation records, causing data chain breaks during audits and failures in compliance reviews.

[0055] Users report broadband faults by phone. Customer service manually enters the work order, and repair personnel record the progress on-site using paper forms. Upon acceptance, the customer needs to sign off on the process. In this scenario, the phone terminal (100), the data entry terminal (200), and the backend system are not synchronized in real time. Even after repair, the work order status still shows "processing," leading to repeated customer complaints and service conflicts. The reliance on offline signatures during the acceptance process, requiring manual re-entry of data, results in a high error rate. Furthermore, customers cannot track the entire service process online, leading to decreased customer satisfaction.

[0056] See Figure 1 , Figure 1 This is an improved electronic customer service order management system. With the development of AI technology, the client terminal 10 may no longer be necessary. Therefore, the new electronic system replaces the client terminal 10 with a telephone terminal 100, which integrates an input terminal 200 to determine specific work order information. Then, through an interactive computer terminal 300 and a customer center server 400, the work orders are prioritized. In the event of a work order anomaly or a failure of the customer center server 400, a backup node 400 monitors the execution of the work order, achieving a maintenance effect.

[0057] The telephone terminal 100 integrates voiceprint recognition and multi-factor authentication functions, and realizes dynamic verification of user identity through voice feature comparison, dynamic verification code and behavioral habit analysis.

[0058] The input terminal 200 collects multimedia data during the maintenance process, generates metadata tags containing device identifiers and timestamps, and uses the SM3 algorithm to calculate data digests to ensure integrity;

[0059] The customer center server 400 serves as the core hub, connecting to the blockchain evidence storage platform to realize the real-time uploading of maintenance data hash values ​​to the chain, and using the timestamp of the National Time Service Center to solidify the evidence storage time. At the same time, a distributed multi-node backup mechanism is deployed to ensure data disaster recovery.

[0060] Computer terminal 300 optimizes work order allocation paths based on maintenance personnel skill tags and spatiotemporal characteristics through the Dijkstra algorithm, and dynamically adjusts priorities in conjunction with real-time load, thereby improving the security of blockchain data transmission, reducing resource waste and response delays in worker dispatch, and enhancing system stability.

[0061] Example 1:

[0062] This application proposes an electronic customer service order management system, see reference. Figure 1 It includes a telephone terminal 100, an input terminal 200, a computer terminal 300, and a customer center server 400;

[0063] The telephone terminal 100 receives user repair requests and automatically records incoming call information; it uses CTI (Computer Telephony Integration) technology to connect to the operator's network, automatically captures information such as the caller's number and location, and associates it with the user's profile, replacing the traditional manual inquiry and entry, shortening the repair processing time, and solving the efficiency bottleneck and high error rate problem of manual recording through real-time interaction with the database via signaling parsing.

[0064] In one embodiment, a telephone network is formed using pre-configured AI phones to receive user voice repair requests. Combined with automatic speech recognition technology, the user's voice repair information is converted into structured text, and metadata such as the caller ID and time is automatically extracted and entered into the system. No manual input is required; all information is automatically entered, increasing the accuracy of work order dispatch. Furthermore, automated, structured work order fields are generated, indicating the type and urgency of the fault, enabling intelligent allocation.

[0065] In one embodiment, the user terminal 80 is directly connected to devices such as gas meters. Through its integrated IoT interface, it automatically connects to the telephone terminal 100. By combining temperature, vibration, and energy consumption data with edge computing, it analyzes the health status of devices such as gas meters in real time. Then, through LSTM neural networks or reinforcement learning models, combined with historical maintenance data and real-time device data, it automatically predicts potential faults and proactively generates preventive maintenance work orders, rather than waiting for users to report repairs.

[0066] The input terminal 200 collects maintenance process data and generates multimedia data with timestamps. The input terminal 200 has a built-in embedded time synchronization module that calibrates the standard time of the National Time Service Center through the NTP protocol to ensure that maintenance data such as photos and videos are given an immutable timestamp when they are generated. The system analyzes the correspondence between maintenance process time and behavior in a way similar to electronic evidence solidification technology.

[0067] In one embodiment, the input terminal 200 is a handheld input terminal 200 that can take pictures. It needs to generate a reliable timestamp for the necessary task information of the repair work order, such as replacing parts or video of the repaired product without abnormalities, through the built-in clock or the time source of the national time service center, and encapsulate it into a hash value, thereby achieving the purpose of being tamper-proof.

[0068] The Customer Center Server 400 connects to the blockchain evidence storage platform, generating hash values ​​for maintenance process data and uploading them to the blockchain in real time via API interface. The time stamp from the National Time Service Center is used to solidify the evidence storage time. The Customer Center Server 400 uses the SHA-256 algorithm to generate unique hash values ​​for maintenance data and pushes them to the consortium blockchain nodes in real time via API interface. Combined with the National Time Service Center timestamp, a three-dimensional evidence storage chain of data, hash, and time is formed. Based on distributed ledger technology, this solves the pain points of traditional paper documents being easily tampered with and disputes over the timeliness of evidence storage.

[0069] In one embodiment, during the repair process, the repair process data is synchronized and stored using a blockchain and a timestamp from the National Time Service Center, ensuring the immutability of the repair data in terms of both time and reliability. It also serves as evidence in case of disputes.

[0070] In one embodiment, the maintenance process data includes structured evidence information of the parts replaced on the maintenance equipment or gas equipment. The system automatically compares the equipment data after maintenance with the rules. If the conditions are met, the work order is automatically accepted. Otherwise, the work order is returned and the non-compliant items are marked. This integrates the evidence data with the business process, and intelligently judges the execution status of the work order while storing evidence.

[0071] Computer terminal 300 automatically assigns work orders to corresponding maintenance personnel based on their skill tags and work order types, and updates the work order status in real time on the customer center server 400, performing online acceptance and return operations. Computer terminal 300 constructs a maintenance personnel skill vector library based on the TF-IDF algorithm, matches work order requirements with personnel tags using cosine similarity, dynamically generates the optimal dispatch plan, and integrates natural language processing and operations optimization to prevent subjectivity and resource mismatch issues associated with manual dispatch.

[0072] In one embodiment, the computer terminal 300 automatically assigns the optimal personnel based on the maintenance personnel's skill tags, such as special work qualifications and experience and efficiency in handling a certain routine fault, and matches them with the work order type, thereby improving the accuracy of skill matching.

[0073] In one embodiment, the execution status of a work order can be determined by the storage time and data of the maintenance process hub, thereby enabling online work order acceptance and work order return (redispatch of maintenance personnel) and improving work efficiency.

[0074] In existing technologies, customer service systems mostly use single-terminal processing or simple data uploading. By collaborating with multiple terminals, combining blockchain notarization and intelligent scheduling, the real-time nature of telephone terminal 100, the on-site data collection of input terminal 200, the distributed notarization of the customer center, and the algorithm optimization of computer terminal 300 can be organically integrated. By combining blockchain notarization with the timestamp of the National Time Service Center, data tampering can be prevented. Furthermore, the skill-based assignment algorithm can achieve accurate matching between personnel and work orders.

[0075] Example 2:

[0076] See Figure 3 When the telephone terminal 100 receives a user's repair request, it initiates an identity verification process. Traditional electronic systems, where customer service records repair information, are prone to errors, especially for elderly users or those with busy schedules. Therefore, this application proposes the following solution:

[0077] Voiceprint recognition technology extracts user voice features and compares them with the system's pre-stored voiceprint database. Voiceprint recognition technology extracts the spectral features of user voice through MFCC (Mel-frequency cepstral coefficients) and compares them with the pre-stored voiceprint database using dynamic time warping (DTW). Based on the uniqueness of voiceprint biometrics, it replaces traditional password verification, reduces memory burden, and has stronger anti-counterfeiting capabilities, solving the problem of easy leakage of static passwords.

[0078] In one embodiment, voiceprint recognition is based on the uniqueness of voice signals. A feature extraction algorithm extracts biometric features such as the spectrum and fundamental frequency from the user's voice and matches them against pre-stored user voiceprint templates in a database. This eliminates the need for users to manually enter their account numbers or for customer service to input them. Voiceprint information is directly matched with the user's gas account number, improving the efficiency of reporting repairs and reducing the error rate of recording.

[0079] If the match fails, a dynamic verification code will be automatically sent to the user's registered mobile phone, and a preset security question will be asked through an interactive voice response mechanism. When the voiceprint match fails, a second-level verification of dynamic verification code and interactive voice response (IVR) security question will be automatically triggered. This combines the user's owned device (mobile phone verification code) and known information (security question) for dual verification. Multi-dimensional cross-verification enhances the credibility of the identity and avoids the risk of failure of a single biometric feature.

[0080] The system acquires question and answer texts, assesses the semantic similarity of the answer texts, and comprehensively judges the authenticity of the identity based on the user's historical reporting behavior habits. The system uses a BERT pre-trained model to vectorize the user's answer texts, calculates the cosine similarity with the standard answer, and improves the user experience while maintaining verification accuracy by using synonyms such as address and home address, reducing verification failures caused by differences in expression.

[0081] In one embodiment, semantic similarity assessment calculates the semantic matching degree between the user's answer and the preset answer through an NLP model, and determines the user's repair frequency, time period, device type, common phrases, etc. based on the user profile database through historical behavior habit analysis. The rule engine or machine learning model identifies abnormal patterns, enhances the user's expression information, and prevents errors in the entry of repair information caused by differences in expression or dialects.

[0082] If abnormal behavior patterns are detected, the verification strength is enhanced. Enhanced verification builds a behavioral baseline based on user's historical repair reporting time period, device type, language style, etc., and detects abnormal patterns (such as high frequency of repair reports at night, sudden changes in device model) through the Isolation Forest algorithm. It draws on user profiling and anomaly detection technology to achieve dynamic adaptation that simplifies verification of normal behavior and strengthens verification of abnormal behavior, thus solving the rigidity of fixed verification processes.

[0083] This application overcomes the limitations of traditional IVR's mechanical matching through NLP semantic analysis, and achieves adaptive risk adjustment by combining user behavior baselines, thus realizing dual protection and preventing data leakage.

[0084] Example 3:

[0085] See Figure 4 When the terminal 200 collects maintenance process data, it generates metadata tags containing data feature values, collection timestamps, and terminal device identifiers. It calculates the data digest using the SM3 cryptographic hash algorithm and transmits it to the customer service center. The maintenance process data includes on-site photos, temperature measurement records, and user confirmation information during the maintenance process.

[0086] In one embodiment, the metadata tag automatically extracts data feature values, including a collection timestamp accurate to milliseconds and a unique terminal device identifier, through a terminal sensor or application to form a structured tag.

[0087] In one embodiment, the original maintenance data, such as the binary stream of photos, temperature measurement text, and user confirmation electronic signature, are hashed to generate a fixed-length data digest, which serves as the basis for data integrity verification.

[0088] In one embodiment, the definition of maintenance process data explicitly includes images, structured records, and user interaction data for information such as on-site photos, temperature measurement records, and user confirmation information.

[0089] The data entry terminal 200 automatically attaches a three-part tag to multimedia data, including a data feature value, a timestamp, and a device identifier. The feature value is dynamically generated based on the data content, and the device identifier is bound to the terminal's hardware serial number. Based on the serial number, each data segment is ensured to have unique traceability, resolving the issue of proving the correlation between maintenance data and the data collection scenario, and preventing data forgery or misattribution. For raw data such as photos and temperature records, a 128-bit hash digest is generated using the SM3 algorithm. During transmission, only the digest, not the original data, is sent. Based on the collision resistance of the hash function, the receiver can verify data integrity by locally recalculating the digest and comparing it. This significantly reduces transmission bandwidth usage while preventing data tampering on the transmission link.

[0090] In one embodiment, to address the data corruption issue caused by network fluctuations, the SM3 digest of the received raw data is recalculated and compared with the digest value carried in the metadata tag transmitted by the terminal; at the same time, the rationality of the timestamp and the legality of the device identifier are verified to prevent data corruption.

[0091] In one embodiment, to address work order stagnation caused by data anomalies, the most recent 3 maintenance data for the user / device are cached based on the LRU (Least Recently Used) algorithm. When real-time data is abnormal, the most recent valid data is automatically retrieved to temporarily fill the work order, ensuring that the business process is not interrupted.

[0092] After receiving data, the customer center server 400 verifies the integrity of the metadata tags. If the feature value is found to be inconsistent with the locally calculated result, a retransmission command is immediately sent to the input terminal 200, and locally cached historical version data is used for temporary replacement. The customer center server 400 performs dual verification on the received metadata tags, comparing feature values ​​and checking timestamp continuity. If a mismatch is found, the retransmission mechanism is triggered, and locally cached historical version data is used to temporarily fill the gap. By integrating the TCP retransmission mechanism and data redundancy backup strategy, the data transmission error rate is reduced, ensuring business continuity.

[0093] When abnormal characteristic values ​​are found, a verification anomaly record is generated and uploaded to the blockchain evidence storage platform in real time, marked as pending verification, until the evidence storage status is updated after manual review and confirmation. Verification anomaly data automatically generates evidence storage records containing device fingerprints, pushes them to the blockchain platform via API and marks them as pending verification; the status is updated on-chain after manual review. Leveraging the immutability of blockchain to construct an audit trail enables full-process traceability of anomaly data, solving the regulatory challenge of easily tampered or deleted anomaly data in traditional systems.

[0094] In one embodiment, at the data verification level, it is understood that the above embodiment 3 combines cryptographic hashing with blockchain evidence storage, metadata tags provide basic traceability capabilities, SM3 ensures transmission integrity, and blockchain enables tamper-proof evidence storage of abnormal data, making the evidence storage more comprehensive.

[0095] In one embodiment, in terms of fault tolerance, the problem of service interruption caused by data anomalies is solved by replacing real-time retransmission with historical caching.

[0096] Example 4:

[0097] When the customer center server 400 detects that the work order processing time exceeds a preset threshold or that the same user submits a work order of the same fault type within a preset time, it automatically triggers an anomaly level assessment. The customer center server 400 triggers anomaly assessment based on two conditions: real-time monitoring of work order processing time, comparison of timer and database timestamps, and detection of duplicate user submissions. This solves the problem of lag in traditional manual inspections, reducing the average anomaly detection time from 2 hours to 15 minutes in actual implementation.

[0098] In one embodiment: for work orders that may have anomalies, the status of the work orders is remotely monitored in real time based on a timer, and an event is triggered when the processing time exceeds a preset threshold.

[0099] In one implementation: For issues involving potentially duplicate work orders, a composite index of user ID + fault type + time window is used to query duplicate records in a relational database; if a match is found, a flag is triggered. A rule engine is used to abstract the triggering conditions into business rules, ensuring the logic is configurable.

[0100] Anomaly level assessment includes:

[0101] The system analyzes the matching degree between work order types and historical processing data. If the matching degree is lower than a first matching value, it is marked as an anomaly to be confirmed. An improved KNN algorithm is used to compare the feature vectors of the current work order with those of historical processing data, and a first matching value is set as the anomaly threshold. Based on machine learning pattern recognition, work orders with mismatched types are quickly located, improving matching accuracy.

[0102] In one embodiment, to reduce misjudgment, the work order type is converted into a feature vector using the TF-IDF algorithm, and cosine similarity is calculated with data such as the processing time, parts involved, and technician skill tags of historical similar work orders.

[0103] In one embodiment, to prevent empirical bias, the first matching value is determined by A / B testing and can be dynamically adjusted.

[0104] The process involves using Natural Language Processing (NLP) to analyze the emotional tone of work order descriptions to identify user dissatisfaction when submitting repeated requests. This analysis, combined with real-time feedback from maintenance personnel regarding the on-site difficulty, is used to comprehensively determine the anomaly level. Anomaly levels are categorized into Level 1 (representing high-risk faults), Level 2 (representing parts scheduling), and Level 3 (representing operational errors). A BERT pre-trained model is used to score the sentiment polarity of the work order description text, and this, combined with the on-site difficulty coefficients submitted by maintenance personnel via the terminal, constructs an anomaly level determination matrix. By integrating NLP sentiment analysis with multi-source data fusion, implicit user dissatisfaction is transformed into quantifiable indicators, improving the accuracy of anomaly level determination.

[0105] In one embodiment, when a user reports a repair request, the intensity of their expression or the emotional state arising from a gas malfunction can lead to misjudgment of priorities. This application uses a BERT fine-tuning model to classify the emotional polarity of the work order description text, such as "I've been waiting all day and no one has come!", extracting negative emotion keywords such as anger and urging, and calculating an emotion intensity score (0-10 points). Simultaneously, for repair personnel, they submit a difficulty level (and text description) via a mobile app, which the system converts into quantitative features. The level determination is based on a decision tree model, with emotion, on-site difficulty, and historical matching degree all having certain weights, for priority allocation.

[0106] When a Level 1 anomaly occurs, the regular work order allocation is suspended, the abnormal work order is pinned to the top and pushed to the technical supervisor's terminal;

[0107] When a Level 2 anomaly occurs, the backup supplier scheduling process will be automatically initiated.

[0108] When a Level 3 anomaly is detected, a guided solution is generated and broadcast via voice message on a telephone terminal.

[0109] Level 1 anomalies trigger resource preemptive scheduling, suspending regular work orders and top-priority push notifications; Level 2 initiates supply chain emergency response and calls backup supplier API interfaces; Level 3 generates structured solutions (based on knowledge base FAQ matching and TTS speech synthesis). This improves the timeliness of Level 1 anomaly handling, shortens the time for parts to arrive (Level 2), and increases the user self-service resolution rate (Level 3).

[0110] In one embodiment, this application incorporates NLP sentiment analysis and real-time feedback from maintenance personnel into anomaly detection, thereby improving the accuracy of anomaly level determination.

[0111] Example 5: Computer terminal 300 is used to collect maintenance personnel's maintenance task data in real time and construct a three-dimensional evaluation model. This model includes maintenance task data such as work order workload data, skill tag matching data, and geographic location data. Computer terminal 300 collects work order workload, skill tag matching degree, and geographic location data in real time via API interface to construct a three-dimensional feature vector space. By integrating a multi-objective decision-making model from operations research and determining the weights of each dimension through the analytic hierarchy process, precise matching of maintenance personnel's capabilities with task requirements is achieved, increasing resource utilization by 30%.

[0112] In one embodiment, during work order allocation, to avoid allocation errors caused by a single factor, data collection includes collecting three types of real-time data through a mobile terminal: work order workload data (task volume statistics based on a time window), skill tag matching data (obtained by calculating the matching degree between maintenance personnel skill tags and work order requirement tags using the TF-IDF algorithm), and calculating the arrival time and road distance between maintenance personnel and work order locations using GPS and map API, thereby allocating maintenance work orders to ensure timeliness.

[0113] When the 3D evaluation model receives the first work order, its first priority is determined.

[0114] Based on the initial priority, the real-time sorting data of the initial priority of the second work order is used to determine the second priority of the first work order. An event-driven priority sorting algorithm is employed. When a new work order (the second work order) enters the system, its initial priority (based on urgency and user level) is calculated in real time, and the existing work order queue is reordered using a sliding window mechanism to dynamically update the second priority of the first work order. Based on a process scheduling algorithm, the response time for urgent work orders is reduced by 45%, avoiding resource congestion caused by traditional static work order dispatching.

[0115] In one embodiment, to prevent work order backlog and unexpired work orders, a real-time scheduling algorithm based on priority queues (such as a min-heap structure) is adopted. When a new work order (the second work order) arrives, the system recalculates the overall priority of all work orders to be assigned. When two work orders have the same priority, they are sorted according to the first-come, first-served principle or the rule of prioritizing users with higher levels.

[0116] Example 6: When the input terminal 200 and the computer terminal 300 simultaneously revise the real-time data of the same work order, the revision behavior is determined, and the vector clock tag of the terminal identifier and logical timestamp of the revision behavior is determined. The vector clock tag is compared in a partial order by the customer center server 400 to determine whether there is a revision causal relationship. When the input terminal 200 and the computer terminal 300 revise the data, a vector clock tag containing the terminal identifier and logical timestamp is automatically generated. Based on the distributed system event sorting algorithm, the revision dependency relationship is determined by the vector partial order relationship, accurately identifying concurrent revisions and sequential revisions, and solving the problem of conflict misjudgment caused by clock drift of traditional physical timestamps.

[0117] In one embodiment, to prevent synchronization errors of physical timestamps in the distributed system, a conflict detection mechanism is triggered by monitoring the TCP connection between the terminal and the server to capture data write operations. A vector clock algorithm in the distributed system is employed to generate a two-dimensional vector containing the terminal identifier and logical timestamp for each revision action.

[0118] When a causal relationship exists between revisions, conflict resolution rules are activated. When there is a partial order relationship between vector clocks, such as a revision at computer terminal 300 being later than that at input terminal 200, preset rules are activated, and the later revision overwrites the earlier revision or field-level incremental merging, controlling the incremental synchronization mechanism of the system and automatically merging revisions with dependencies, thus reducing the rate of manual intervention.

[0119] In one embodiment, to address potential conflicts in work order allocation, based on the terminal device reported by the user, if all dimensions of the vector clock of terminal device A are less than or equal to the vector clock of terminal device B, and at least one dimension is less, then the revision of terminal device A takes precedence over terminal device B. If, in comparison, the vector clocks of A and B have some dimensions greater than and some dimensions less than, then the revisions are processed as concurrent work orders. For conflict resolution, if there is a sequential order of revisions, a later revision overrides a earlier one rule, automatically retaining the later revision version; if some fields overlap, then the non-overlapping fields are merged. Custom conflict resolution rules can also be defined.

[0120] When there is no causal relationship for revision, the key fields for revision are identified, and the terminal priority of the same work order is determined based on these key fields. In the absence of a causal relationship, the key fields in the revision content, maintenance results, and user signatures are extracted using the TF-IDF algorithm. Arbitration is then based on the terminal priority matrix, integrating information retrieval and rule engine technologies to ensure that key data is revised first by the on-site terminal, while non-key fields are automatically merged, thus shortening conflict resolution time.

[0121] Example 7:

[0122] Computer terminal 300 is also used for:

[0123] The system collects real-time GPS location data of maintenance personnel, work order addresses, and traffic conditions, and calculates the shortest service path using Dijkstra's algorithm. The computer terminal 300 collects 3D data via GPS module and map API, abstracting geographical locations into graph nodes (latitude and longitude coordinates) and traffic conditions into edge weights (congestion coefficient × distance). Dijkstra's algorithm is then used to calculate the shortest path. Based on the shortest path algorithm in graph theory, and combined with real-time traffic conditions to dynamically adjust weights, the one-way service time is shortened, thus achieving a high accuracy rate in path planning.

[0124] In one embodiment, for positioning, real-time latitude and longitude are obtained through the GPS module of the maintenance personnel's mobile device, combined with base station or gas terminal equipment for auxiliary positioning. The work order address is obtained by parsing the structured address in the work order and converting it into location coordinates.

[0125] In one embodiment, a weighted directed graph (edge ​​weight = distance × road condition coefficient + traffic light waiting time) is constructed, with the current location of the maintenance personnel as the starting point and each work order address as a node. The single-source shortest path is solved by Dijkstra's algorithm, and the optimal path including turning instructions and estimated time of arrival (ETA) is output.

[0126] When at least two similar work orders are detected in the same area within a preset time period, they are automatically merged and prioritized for assignment to the nearest and skill-matched maintenance personnel, while the optimal transportation route is simultaneously pushed. Service areas are divided using spatial clustering algorithms, and similar work orders are statistically analyzed using a sliding time window. After a merging rule is triggered, the Hungarian algorithm is invoked to find the optimal match between personnel and work orders. This fusion of spatial clustering and combinatorial optimization increases the merging rate of work orders within the same area, thereby increasing the average daily service volume per maintenance personnel.

[0127] In one embodiment, for work orders of the same type existing in the same area within a certain time period, work orders within the same area are quickly retrieved using a spatial index (based on a grid geofence or administrative region). A preset time threshold is set, and the work order density per unit time is statistically analyzed using a sliding window algorithm. From the personnel to be assigned, the three closest people are selected using the K-nearest neighbor algorithm (K=3), and the final target is determined by skill tag matching. The merged work orders are sorted by geographical location, and the Dijkstra algorithm is called again to calculate the optimal multi-node path to prevent detours.

[0128] Example 8: After the customer center server 400 receives the user acceptance confirmation information uploaded by the input terminal 200 and the computer terminal 300 synchronizes the work order completion status, the contract automatically verifies the data integrity, calls the blockchain platform through the API interface to complete the hash value on the chain, synchronously generates the national time service center timestamp, and pushes the evidence certificate to the user terminal.

[0129] In one embodiment, to achieve dual-terminal confirmation, a distributed transaction mechanism (such as TCC mode) is employed to ensure that data from both the input terminal 200 and the computer terminal 300 is successfully uploaded and in the same state before triggering subsequent processes. Through smart contract-preset verification rules, key fields are hash-verified, comparing the hash values ​​of the data uploaded by the terminal with those of the original data cached on the server; if they match, the data is deemed complete.

[0130] The customer center server 400 verifies the consistency between acceptance information and work order status through a preset rule engine. Based on cryptographic hash algorithms and a business rule engine, it reduces data error rates and eliminates the risk of manual data entry tampering. The system calls the blockchain platform API to write the work order data hash value into the distributed ledger and simultaneously requests the National Time Service Center's NTP service to generate a trusted timestamp. By integrating distributed ledger technology and authoritative time authentication, the evidence records possess the characteristics of immutability, traceability, and time uniqueness. The evidence certificates containing blockchain addresses are pushed to user terminals in real time via the WebSocket protocol. Based on instant messaging technology, users can verify the evidence status in real time, improving service transparency.

[0131] In one embodiment, to prevent maintenance personnel or users from uploading false times and work order tasks, the server pushes verified work order data to the consortium blockchain platform, calls the smart contract's notarization function, and writes the data into a block. It then obtains the standard time through the National Time Service Center's NTP (Network Time Protocol) interface, generates a digitally signed timestamp file, and binds it to the blockchain transaction hash to ensure the time is tamper-proof. The system automatically generates a PDF notarization certificate containing the blockchain transaction ID, timestamp, and acceptance summary, which is pushed to the user's terminal via SMS / APP. Simultaneously, a query entry is provided on the blockchain explorer, thus ensuring work order transparency.

[0132] Example 9: When the customer center server 400 receives updated work order data from the input terminal 200, it pushes a status change notification to the associated computer terminal 300 via the WebSocket protocol, triggering a local cache refresh.

[0133] If computer terminal 300 is offline, the server will temporarily store the change records in the message queue, and synchronize them in timestamp order after the terminal reconnects.

[0134] In one embodiment, to ensure data synchronization, computer terminal 300 adopts a full-duplex communication protocol. A persistent connection is established between the server and computer terminal 300. When input terminal 200 updates work order data, the server proactively pushes a JSON-formatted notification to the associated terminal via an event-driven model, with virtually no push delay. Upon receiving the notification, the terminal triggers a cache eviction policy, clearing the old cached data for the corresponding work order and retrieving the latest complete data via an HTTP GET request to update its local database, ensuring that the interface display is consistent with the server.

[0135] In one embodiment, the customer center server 400 uses the full-duplex WebSocket protocol. When the input terminal 200 updates a work order, it immediately pushes a JSON-formatted status notification (including work order ID, change fields, and timestamp) to the associated computer terminal 300, triggering a refresh of the key-value pairs cached locally on the terminal. Based on TCP long-connection real-time communication technology, the redundant requests of traditional HTTP polling are eliminated, the status synchronization delay is ≤100ms, terminal data consistency is high, and network bandwidth consumption is reduced. A RabbitMQ message queue is used to temporarily store change records of offline terminals. Each message carries an incrementing timestamp. After the terminal reconnects, a TCP three-way handshake is established, and messages are synchronized in batches in FIFO order. Based on asynchronous communication and a timestamp sorting algorithm using message middleware, zero offline data loss is achieved, and the synchronization time after reconnection is ≤2 seconds, resolving data gaps caused by short polling.

[0136] Example 10:

[0137] The customer center server 400 uses a multi-node backup architecture; among which,

[0138] When the master node detects a hardware failure or data corruption, the system automatically switches to the standby node, elects a new master node using the Raft consensus algorithm, and initiates the data repair process.

[0139] The primary and backup nodes maintain state consistency through real-time data synchronization. The primary node sends a heartbeat packet every 100ms and triggers VRRP protocol switching in case of an anomaly. Based on cluster high availability technology, the service interruption rate caused by single point of failure is reduced.

[0140] In one embodiment, the multi-node deployment adopts a 1-master-N-standby distributed cluster architecture (e.g., master node + 3 standby nodes). The nodes communicate with each other via a private local area network, and the shared storage uses a distributed file system to achieve real-time data synchronization. The master node is responsible for handling all read and write requests, and the standby nodes replicate the master node's data logs in real time, maintaining a hot backup state (and can switch to master node at any time).

[0141] When a node fails, the remaining nodes elect a leader using the Raft algorithm. The new master node automatically takes over the IP address and service port, avoiding the split-brain problem based on a distributed consensus protocol.

[0142] In one embodiment, the system decouples the service address from the physical node through virtual IP technology. During the switch, the VIP only needs to be moved from the faulty primary node to the backup node. The terminal is unaware of this and there is no service jitter during the switch.

[0143] After the faulty node is recovered, the missing data is pulled from the new master node through incremental log synchronization, and the damaged blocks are repaired by combining CRC32 check. By integrating database recovery and error control technologies, the data recovery integrity rate is improved, and the full data retransmission is avoided.

[0144] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. An electronic customer service order management system, characterized in that, This includes telephone terminals, data entry terminals, computer terminals, and customer center servers; The telephone terminal receives user repair requests and automatically records incoming call information; The data entry terminal collects maintenance process data and generates multimedia data with timestamps. The customer center server connects to the blockchain evidence storage platform, generates hash values ​​for the repair process data, uploads it to the blockchain in real time via API interface, and uses the timestamp of the National Time Service Center to solidify the evidence storage time. The computer terminal automatically assigns work orders to the corresponding maintenance personnel based on the maintenance personnel's skill tags and work order types, and updates the work order status in real time in the customer center server, performing online acceptance and return operations.

2. The electronic customer service order management system as described in claim 1, characterized in that, When the telephone terminal receives a user's repair request, it initiates an identity verification process: first, it extracts the user's voice features using voiceprint recognition technology and compares them with the system's pre-stored voiceprint database; If the match fails, a dynamic verification code will be automatically sent to the user's registered mobile phone, and preset security questions will be asked through an interactive voice response mechanism. Obtain the text of questions and answers, assess the semantic similarity of the answer texts, and comprehensively judge the authenticity of the identity based on the user's historical behavior habits; If an abnormal behavior pattern is detected, the verification strength is increased.

3. The electronic customer service order management system as described in claim 1, characterized in that, When the data entry terminal collects maintenance process data, it generates metadata tags containing data feature values, collection timestamps, and terminal device identifiers. It calculates a data digest using the SM3 cryptographic hash algorithm and transmits it to the customer service center. The maintenance process data includes on-site photos, temperature measurement records, and user confirmation information during the maintenance process. After the customer center server receives the data, it verifies the integrity of the metadata tags. If the feature value is found to be inconsistent with the local calculation result, it immediately sends a retransmission instruction to the input terminal and activates the locally cached historical version data for temporary replacement. When abnormal characteristic values ​​are found, a verification anomaly record is generated and uploaded to the blockchain evidence storage platform in real time, marked as pending verification, until the evidence storage status is updated after manual review and confirmation.

4. The electronic customer service order management system as described in claim 1, characterized in that, When the customer center server detects that the work order processing time exceeds a preset threshold or that the same user submits a work order of the same fault type within a preset time, it automatically triggers an anomaly level assessment. Anomaly level assessment includes: Analyze the matching degree between work order type and historical processing data. If the matching degree is lower than the first matching value, mark it as an anomaly to be confirmed. The anomalies to be confirmed are analyzed using natural language processing technology to analyze the emotional tendencies in the work order descriptions, identify the characteristics of dissatisfaction when users submit repeatedly, and combine this with the on-site difficulty reported by maintenance personnel in real time to comprehensively determine the anomaly level. The anomaly levels include Level 1 anomalies that represent risky faults, Level 2 anomalies that represent parts scheduling, and Level 3 anomalies that represent operational errors. When a Level 1 anomaly occurs, the regular work order allocation is suspended, the abnormal work order is pinned to the top and pushed to the technical supervisor's terminal; When a Level 2 anomaly occurs, the backup supplier scheduling process will be automatically initiated. When a Level 3 anomaly is detected, a guided solution is generated and announced via voice broadcast on a telephone terminal.

5. The electronic customer service order management system as described in claim 1, characterized in that, The computer terminal is used to collect maintenance personnel's maintenance task data in real time and construct a three-dimensional evaluation model; wherein, the maintenance task data includes work order workload data, skill tag matching data and geographical location data; When the 3D evaluation model receives the first work order, its first priority is determined. Based on the initial priority, the real-time sorting data of the initial priority of the second work order is used to determine the second priority of the first work order.

6. The electronic customer service order management system as described in claim 1, characterized in that, When the data entry terminal and the computer terminal simultaneously revise the real-time data of the same work order, the revision action is determined, and the terminal identifier and the vector clock tag of the logical timestamp of the revision action are determined; wherein, The vector clock tags are compared in a partial order via the customer center server to determine whether a revision causal relationship exists. When a causal relationship for revision exists, the conflict resolution rule is invoked. When there is no causal relationship for revision, determine the key fields for revision, and use these key fields to determine the terminal priority of the same work order.

7. The electronic customer service order management system as described in claim 1, characterized in that, The computer terminal is also used for: Real-time collection of maintenance personnel's GPS location, work order address, and traffic data; calculation of the shortest service path using the Dijkstra algorithm. When at least two similar work orders are detected in the same area within a preset time, they are automatically merged and scheduled, and prioritized for allocation to the nearest and skill-matched target maintenance personnel, while simultaneously pushing the optimal transportation route.

8. The electronic customer service order management system as described in claim 1, characterized in that, After the customer center server receives the user acceptance confirmation information uploaded by the input terminal and the computer terminal synchronizes the work order completion status, the contract automatically verifies the data integrity, calls the blockchain platform through the API interface to complete the hash value on the chain, synchronously generates the national time service center timestamp, and pushes the evidence certificate to the user terminal.

9. The electronic customer service order management system as described in claim 1, characterized in that, When the customer center server receives updated work order data from the input terminal, it pushes a status change notification to the associated computer terminal via the WebSocket protocol, triggering a local cache refresh. If the computer terminal is offline, the server will temporarily store the change records in the message queue, and synchronize them in timestamp order after the terminal reconnects.

10. The electronic customer service order management system as described in claim 1, characterized in that, The customer center server uses a multi-node backup architecture; among which... When the master node detects a hardware failure or data corruption, the system automatically switches to the standby node, elects a new master node using the Raft consensus algorithm, and initiates the data repair process.

Images