Personalized work order transfer method and system based on intelligent voice AI

Abstract

The invention discloses a personalized work order transfer method and system based on intelligent voice AI, belongs to the technical field of service work order personalized distribution assisted by cloud computing and artificial intelligence, and aims to solve the technical problem of how to improve the accuracy and efficiency of a work order judgment result, reduce dependence on manual annotation data and improve the work order transfer efficiency. According to the technical scheme, a voice request is initiated, wherein a user initiates a voice service request through a user terminal; voice recognition and information extraction: recognizing a voice service request through an intelligent voice AI server, and extracting key information; matching a work order processing strategy: the intelligent voice AI server retrieves the matched work order processing strategy from a database according to the extracted key information; generating a personalized chemical list: according to a preset algorithm, generating the personalized chemical list for the user in combination with the historical service record and the implementation condition of the user; aI order transfer: transferring the personalized chemical order to a corresponding skill group and a person in charge through an intelligent voice AI server; voice is fed back to the user.

Description

Technical Field

[0001] The present invention relates to the technical field of personalized service work order allocation assisted by cloud computing and artificial intelligence, and specifically to a personalized work order transfer method and system based on intelligent voice AI. Background Art

[0002] In the existing cloud operation and maintenance work order dispatching system, as the business scale expands and customer expectations continue to rise, more personalized and immediate solutions are expected. The traditional one-size-fits-all service model can no longer meet this demand. The traditional customer service system mainly relies on manual operations, including answering calls, recording problems, and manually assigning work orders. This method is not only inefficient, but also prone to information omissions and response delays. Especially when faced with a large number of user consultations and diversified needs, service quality and customer satisfaction are often difficult to guarantee. With the development of cloud computing and big data technology, infrastructure support is provided for processing massive user data and implementing complex algorithms. This makes it possible to analyze user behavior, predict demand, and optimize resource allocation. With breakthroughs in deep learning technology, especially in the fields of speech recognition and natural language processing, AI can more accurately understand human language and extract semantic information, thus laying the foundation for the intelligence of intelligent customer service systems.

[0003] Therefore, how to improve the accuracy and efficiency of work order judgment results and reduce dependence on manually labeled data is a technical problem that needs to be solved urgently. Summary of the invention

[0004] The technical task of the present invention is to provide a personalized work order transfer method and system based on intelligent voice AI to solve the problem of how to improve the accuracy and efficiency of work order judgment results and reduce dependence on manually labeled data.

[0005] The technical task of the present invention is achieved in the following way: a personalized work order transfer method based on intelligent voice AI, the method is as follows:

[0006] Initiate voice request: The user initiates a voice service request through the user terminal;

[0007] Speech recognition and information extraction: The intelligent speech AI server recognizes the speech service request and extracts key information:

[0008] Matching work order processing strategy: The intelligent voice AI server retrieves the matching work order processing strategy from the database based on the extracted key information;

[0009] Generate personalized work orders: Generate personalized work orders for users based on preset algorithms, combined with historical service records and implementation status of users;

[0010] AI transfer: transfer personalized work orders to corresponding skill groups and persons in charge through intelligent voice AI servers;

[0011] Voice feedback to users: Track service progress through intelligent voice AI, update work order information based on service status, and provide voice feedback to users on processing results.

[0012] As a preferred method, the matching work order processing strategy is to determine the key information, combine the historical work order transfer situation, and recommend the processing skill group and person in charge of the new work order. The specific recommendation process is as follows:

[0013] Determine the skill group of the work order. Each skill group is considered a category, and the order transfer recommendation is converted into a classification problem.

[0014] Based on the classification, the specific work order is assigned to the person in charge of the corresponding skill group and then transferred to the current person on duty.

[0015] Preferably, the work order description method is as follows:

[0016] ① Directly use the original work order text as training expectation without performing word segmentation operations. The Bert model will automatically perform word segmentation and training or reasoning.

[0017] ②First use jieba to perform word segmentation and remove stop words.

[0018] More preferably, the generation of the personalized work order includes identifying specific rules in the work order conversion;

[0019] Identify specific rules in work order transfer, as follows:

[0020] ① Shield alarm work orders and directly send them to NOC for processing;

[0021] ② Capacity warning work orders are directly transferred to the cloud center;

[0022] ③Zero-trust password reset work orders are handled by NOC;

[0023] ④Whether the handler and skill group are directly specified in the work order content.

[0024] Preferably, the handler and skill group are directly specified in the work order content. The judgment method is as follows:

[0025] ① Use string matching to find the skill group name or personnel name directly in the work order description;

[0026] ② Train the named entity recognition model to extract work handlers and skill groups;

[0027] ③ Through the prompt engineering method, the large model answers through the few-shot method.

[0028] Preferably, AI transfer order is as follows:

[0029] Fill out the work order according to the problem description;

[0030] Perform word segmentation processing based on the work order content and obtain the work order content after word segmentation processing;

[0031] Perform stop word filtering on the work order content after word segmentation, remove the stop words, and obtain the filtered work order content;

[0032] Classify the filtered work order content through the classification model to obtain the classification results;

[0033] According to the classification results, the transfer order queue is recommended by the recommendation model and assigned to the corresponding on-duty personnel.

[0034] A personalized work order transfer system based on intelligent voice AI, the system comprising:

[0035] A voice request initiation module is used for a user to initiate a voice service request through a user terminal;

[0036] The speech recognition and information extraction module is used to identify the speech service request through the intelligent speech AI server and extract key information:

[0037] The work order processing strategy matching module is used to retrieve matching work order processing strategies from the database based on the extracted key information through the intelligent voice AI server;

[0038] Dynamic work order generator, which is used to generate personalized work orders for users based on preset algorithms, combined with historical service records and implementation status of users;

[0039] The work order transfer execution module is used to transfer personalized work orders to the corresponding skill groups and responsible persons through the intelligent voice AI server;

[0040] The voice feedback module is used to track the service progress through intelligent voice AI, update the work order information according to the service status, and feedback the processing results to the user in the form of voice.

[0041] Preferably, the system also includes an operation and maintenance management terminal, which uses a graphical interface. A secure data communication protocol is used between the operation and maintenance management terminal and the intelligent voice AI server and the work order processing server to ensure data security.

[0042] Preferably, the operation and maintenance management terminal working process is as follows:

[0043] (1) The operation and maintenance personnel monitor the operating status of the intelligent voice AI server and the work order processing server through the operation and maintenance management terminal;

[0044] (2) When an abnormal situation is found, the operation and maintenance personnel adjust the work order processing strategy in real time to ensure service quality;

[0045] (3) The operation and maintenance management terminal can view the progress of work order processing in real time, which is convenient for management and scheduling.

[0046] Preferably, the single description method is as follows:

[0047] ① Directly use the original work order text as training expectation without performing word segmentation operations. The Bert model will automatically perform word segmentation and training or reasoning.

[0048] ②First use jieba to perform word segmentation and remove stop words;

[0049] The generation of personalized work orders includes identifying specific rules in the work order transfer;

[0050] Identify specific rules in work order transfer, as follows:

[0051] ① Shield alarm work orders and directly send them to NOC for processing;

[0052] ② Capacity warning work orders are directly transferred to the cloud center;

[0053] ③Zero-trust password reset work orders are handled by NOC;

[0054] ④Whether the work order content directly specifies the handler and skill group;

[0055] The handler and skill group are directly specified in the work order content. The judgment method is as follows:

[0056] ① Use string matching to find the skill group name or personnel name directly in the work order description;

[0057] ② Train the named entity recognition model to extract work handlers and skill groups;

[0058] ③ Through the prompt engineering method, the large model answers through the few-shot method.

[0059] The personalized work order transfer method and system based on intelligent voice AI of the present invention has the following advantages:

[0060] (1) The present invention uses intelligent speech recognition technology, natural language processing technology and big data analysis technology to achieve efficient recognition and personalized dispatch of customer service requests;

[0061] (ii) The present invention receives voice input from customers and intelligently analyzes customer needs; dynamically generates work orders based on customer needs and preset business rules; realizes personalized transfer of work orders through the constructed user portrait and resource optimization algorithm, and has intelligent voice recognition and processing functions, which can significantly improve the accuracy and efficiency of work order processing, while reducing the reliance on manually labeled data, and is particularly suitable for the cloud operation and maintenance industry, which can greatly improve the customer service experience;

[0062] (III) The present invention ensures that user needs are captured quickly and accurately by introducing voice recognition and in-depth intent understanding capabilities, and realizes intelligent allocation and transfer of work orders; the multi-dimensional resource matching algorithm combined with dynamic adaptability realizes the precise matching between the processor and the work order requirements, and automatically matches the most suitable engineer to handle it, which greatly reduces the time and labor cost of manual scheduling and improves overall work efficiency;

[0063] (iv) The present invention ensures that work orders can be processed efficiently and professionally within a specified time by tracking the progress of work order processing in real time, which helps to improve customer satisfaction and enhance corporate reputation;

[0064] (V) The present invention uses artificial intelligence to transfer work orders, which can reduce the enterprise's dependence on manual scheduling to a certain extent, thereby reducing labor costs; the rules and weights of work order transfer can be customized according to the business needs of the enterprise, making the system more flexible and adaptable. BRIEF DESCRIPTION OF THE DRAWINGS

[0065] The present invention is further described below in conjunction with the accompanying drawings.

[0066] Attached Figure 1 This is a flowchart of a personalized work order transfer method based on intelligent voice AI;

[0067] Attached Figure 2 This is a structural diagram of a personalized work order dispatching system based on intelligent voice AI;

[0068] Attached Figure 3 It is the workflow diagram of the operation and maintenance management terminal;

[0069] Attached Figure 4 This is a schematic diagram of the AI ​​order transfer process. DETAILED DESCRIPTION

[0070] The personalized work order dispatching method and system based on intelligent voice AI of the present invention are described in detail below with reference to the drawings and specific embodiments of the specification.

[0071] Embodiment 1:

[0072] As attached Figure 1As shown, this embodiment provides a personalized work order transfer method based on intelligent voice AI, and the method is specifically as follows:

[0073] S1. Initiate voice request: The user initiates a voice service request through the user terminal;

[0074] S2. Voice recognition and information extraction: The intelligent voice AI server recognizes voice service requests and extracts key information:

[0075] S3, matching work order processing strategy: the intelligent voice AI server retrieves the matching work order processing strategy from the database based on the extracted key information;

[0076] S4. Generate personalized work orders: Generate personalized work orders for users based on preset algorithms, combined with historical service records and implementation status of users;

[0077] S5, AI transfer: transfer personalized work orders to corresponding skill groups and persons in charge through intelligent voice AI server;

[0078] S6. Voice feedback to users: Track service progress through intelligent voice AI, update work order information based on service status, and feedback processing results to users in voice form.

[0079] The matching work order processing strategy in step S3 of this embodiment is to determine the key information, combine the historical work order transfer situation, and recommend the processing skill group and person in charge of the new work order. The specific recommendation process is as follows:

[0080] S301. Determine the skill group of the work order. Each skill group is determined as a category, and the order transfer recommendation is converted into a classification problem.

[0081] S302. Determine the specific work order according to the classification and assign it to the person in the corresponding skill group, transfer it to the current person on duty, etc.

[0082] The work order description method in this embodiment is as follows:

[0083] ① Directly use the original work order text as training expectation without performing word segmentation operations. The Bert model will automatically perform word segmentation and training or reasoning.

[0084] ②First use jieba to perform word segmentation and remove stop words.

[0085] The generation of the personalized work order in step S4 of this embodiment includes identifying specific rules in the work order transfer;

[0086] Identify specific rules in work order transfer, as follows:

[0087] ① Shield alarm work orders and directly send them to NOC for processing;

[0088] ② Capacity warning work orders are directly transferred to the cloud center;

[0089] ③Zero-trust password reset work orders are handled by NOC;

[0090] ④Whether the handler and skill group are directly specified in the work order content.

[0091] In this embodiment, the work order content directly specifies the handler and the skill group, and the judgment method is as follows:

[0092] ① Use string matching to find the skill group name or personnel name directly in the work order description;

[0093] ② Train the named entity recognition model to extract work handlers and skill groups;

[0094] ③ Through the prompt engineering method, the large model answers through the few-shot method.

[0095] As attached Figure 4 As shown, the AI ​​transfer order in step S5 of this embodiment is specifically as follows:

[0096] S501. Fill out the work order according to the problem description;

[0097] S502, performing word segmentation processing according to the work order content, and obtaining the work order content after word segmentation processing;

[0098] S503, filtering the work order content after word segmentation processing by stop words, removing the stop words, and obtaining the filtered work order content;

[0099] S504, classifying the filtered work order content through a classification model to obtain a classification result;

[0100] S505. According to the classification result, the order transfer queue is recommended by the recommendation model and assigned to the corresponding on-duty personnel.

[0101] Embodiment 2:

[0102] This embodiment provides a personalized work order transfer system based on intelligent voice AI, which includes:

[0103] A voice request initiation module is used for a user to initiate a voice service request through a user terminal;

[0104] The speech recognition and information extraction module is used to identify the speech service request through the intelligent speech AI server and extract key information:

[0105] The work order processing strategy matching module is used to retrieve matching work order processing strategies from the database based on the extracted key information through the intelligent voice AI server;

[0106] Dynamic work order generator, which is used to generate personalized work orders for users based on preset algorithms, combined with historical service records and implementation status of users;

[0107] The work order transfer execution module is used to transfer personalized work orders to the corresponding skill groups and responsible persons through the intelligent voice AI server;

[0108] The voice feedback module is used to track the service progress through intelligent voice AI, update the work order information according to the service status, and feedback the processing results to the user in the form of voice.

[0109] The system also includes an operation and maintenance management terminal, which uses a graphical interface. A secure data communication protocol is used between the operation and maintenance management terminal and the intelligent voice AI server and work order processing server to ensure data security.

[0110] As attached Figure 3 As shown, the operation process of the operation and maintenance management terminal in this embodiment is as follows:

[0111] (1) The operation and maintenance personnel monitor the operating status of the intelligent voice AI server and the work order processing server through the operation and maintenance management terminal;

[0112] (2) When an abnormal situation is found, the operation and maintenance personnel adjust the work order processing strategy in real time to ensure service quality;

[0113] (3) The operation and maintenance management terminal can view the progress of work order processing in real time, which is convenient for management and scheduling.

[0114] The working process of the system is as follows:

[0115] a) The user initiates a service request through the user terminal, and the request is input in the form of voice.

[0116] b) After receiving the user request, the intelligent voice AI server recognizes the voice and extracts key information.

[0117] c) The intelligent voice AI server retrieves the matching processing skill group and person in charge from the database based on the extracted key information

[0118] After determining the key information, combined with the historical work order transfer situation, recommend the processing skill group and person in charge of the new work order. In the recommendation process, the recommendation process can be divided into two steps:

[0119] The first step is to determine the skill group of the work order. Each skill group is considered a category, so order transfer recommendations can be considered a classification problem.

[0120] Step 2: Determine the person who will handle the specific work order in a certain skill group based on the rules, for example: transfer it to the person currently on duty.

[0121] d) The intelligent voice AI server transfers the work order to the corresponding work order processing server.

[0122] e) After receiving the work order, the work order processing server performs corresponding processing according to the content of the work order.

[0123] f) The processing results are fed back to the user in the form of voice.

[0124] Among them, the intelligent voice AI server uses deep learning technology to achieve voice recognition and key information extraction. The database can be personalized according to user needs. The work order processing server can automatically call the corresponding processing flow according to the type and content of the work order.

[0125] Intelligent voice AI includes speech recognition and natural language processing; the ASR system analyzes the acoustic features of speech signals (such as frequency, amplitude, pitch, etc.) and maps them to predefined words or phrases to generate corresponding text output.

[0126] The model training is as follows:

[0127] Classification training: BERT is a pre-training method proposed by Google AI in 2018. It is a variant of the Transformer architecture and is mainly used for natural language processing (NLP) tasks. Thanks to its powerful bidirectional context encoding capabilities and pre-training-fine-tuning training paradigm, it is mainly used in the following scenarios: sentence classification, named entity recognition (NER), machine translation, etc.

[0128] Construct sample sets and test sets: including data, labels and corresponding relationships, where the ratio of sample sets to test sets is approximately 8 to 2.

[0129] The results of different parameters in the training process are compared:

[0130] Batch size: 16, 32;

[0131] Learning rate (Adam): 5e-5, 3e-5, 2e-5;

[0132] epochs:2,3,4;

[0133] Select the model with the highest accuracy for inference;

[0134] Multi-classification training is performed based on Bert (pre-trained model) and Xgboost respectively. The accuracy of the first version (2023 / 07) is much better than that of traditional machine learning.

[0135] Recommended skill group accuracy Recommender accuracy XGBoost model 58.11% 41.25% Bert model 80.42% 72.39%.

[0136] Database, data analysis, dataset construction;

[0137] Specifically, data analysis involves collecting historical samples of current work order transfers, with a transfer accuracy of 99.19%. If the accuracy of historical training data is insufficient, manual labeling is required to improve the accuracy.

[0138] For the constructed data set, the following judgments or filters are performed:

[0139] Each skill group needs to have a related work order. The work distribution between skill groups may be uneven, resulting in an imbalance in samples of different skill groups. This situation is not currently considered.

[0140] It is recommended to remove the template document in the work order (through regular expression or replace).

[0141] The body of each work order description needs to have certain semantics. It is recommended to set a word limit (e.g. discard work orders with less than 5 Chinese characters).

[0142] Construct training set and test set, including two fields:

[0143] The work order description text;

[0144] Tag(skill group ID);

[0145] Example:

[0146] OSS system submits change application to execute change test change authority OSS system executes change authority Kunming OSS change 0 Chengdu owner requests the exchange system to conduct data volume exchange test name Chengdu exchange system test Chengdu request 1.

[0147] The delimiter between the work order description and the label is (\t, which can be customized and then split according to the delimiter when loading)

[0148] df=pd.read_csv(path,names=["desc","label"],sep="\t")

[0149] The work order description can be provided in two ways:

[0150] 1) Directly use the original work order text as training corpus without performing word segmentation and other related operations. Bert can perform word segmentation and training or reasoning internally.

[0151] 2) Perform word segmentation first (such as using jieba), remove stop words, etc. (the above example is the corpus after word segmentation)

[0152] The 2) method is used in work order processing, which first performs word segmentation (Inspur internal special terms are also added during word segmentation).

[0153] The generation of personalized work orders includes identifying specific rules in work order transfer; identifying specific rules in work order transfer, such as:

[0154] ① Shield alarm work orders and directly send them to NOC for processing;

[0155] ② Capacity warning work orders are directly transferred to the cloud center;

[0156] ③Zero-trust password reset work orders are handled by NOC;

[0157] ④Whether the handler and skill group are directly specified in the work order content.

[0158] The work order content directly specifies the handler and skill group. In this case, the judgment method is as follows:

[0159] ① Use rules and string matching to directly find (skill group name or personnel name) in the work order description;

[0160] ② Train the named entity recognition model to extract work handlers or skill groups;

[0161] ③ Through the prompt engineering method, the large model answers through the few-shot method.

[0162] The work order transfer rules are as follows:

[0163] ① The skill group name (ID) may change. You can map the old and new skill groups in the recommendation results, or directly replace the old skill groups in the training data with the new skill groups.

[0164] ② The job responsibilities of a skill group may be fine-tuned: a skill group may increase or decrease its job responsibilities, but this information is not included in the historical training samples, resulting in inaccurate recommendations. The historical work orders can be manually annotated and retrained (the big model can be used for annotation, and a prompt can be constructed to return the job responsibilities and work order content, which will be judged by the big model).

[0165] ③The division of labor between the two skill groups may not be clear, such as: the second line of the platform and the technical service group.

[0166] In conjunction with the business side’s clear definition of business rules, it is possible to manually determine whether a work order belongs to different skill groups based on the language;

[0167] Collect more training data for these two business groups to allow the big model to learn the boundaries of division of labor;

[0168] After identifying the two types of skill groups through classification, further judgment is made by combining other technologies; for example: machine learning classification model (constructing binary classification); constructing a few-shot method to organize prompts for judgment by the big model; adding a description of each team's responsibilities in the prompt, and the big model determines the skill group based on the content of the work order.

[0169] ④ The work assignment rules from the current skill group to the handler are as follows:

[0170] Directly transferred to the skill group duty officer;

[0171] Assign to the person in charge of the skill group in the region / cloud center (e.g., Jinan Cloud Center: Zhang San is the platform person in charge, and Li Si is the network person in charge);

[0172] Differentiate between working hours and off-time (8:30-17:30), for example: working hours are handled by the person in charge of the skill group in the region / cloud center; non-working hours are handled by the person on duty;

[0173] Distinguish between working days and non-working days: for example, non-working days are handled by the person on duty, etc.

[0174] ⑤Update business-related stop words or professional terms;

[0175] ⑥Extended solution: Use COT and other agent technologies to achieve automatic orchestration.

[0176] Embodiment 3:

[0177] As attached Figure 2 As shown, this embodiment provides a personalized work order transfer system based on intelligent voice AI, which includes:

[0178] Speech recognition and processing module, the specific functions are as follows:

[0179] (1) Technical core: The speech recognition module relies on deep neural networks (DNN), recurrent neural networks (RNN), especially long short-term memory networks (LSTM), and the recent Transformer architecture. These technologies can effectively capture the temporal features in speech signals and achieve high-precision speech-to-text conversion. Through training on large-scale speech data sets, the model can learn the subtle differences in different languages, accents, and speech speeds, thereby improving the recognition rate.

[0180] (2) Intent understanding: Natural language processing (NLP) technology plays a vital role in this link. Using technologies such as word embedding, syntactic analysis, and semantic understanding, the system can parse the intent, sentiment, and specific entity information (such as product name, question type) in the user's voice. This step usually involves advanced NLP tasks such as named entity recognition (NER) and sentiment analysis to ensure that the system can accurately understand the user's real needs.

[0181] User data parsing module, the specific functions are as follows:

[0182] Key information extraction: Using pattern recognition and keyword extraction algorithms, the system can extract elements directly related to the creation of the work order from the recognized text, such as the user's specific request, the urgency of the problem, the type of product or service involved, etc. This process combines rule-based and machine learning methods to achieve both accurate and flexible information capture.

[0183] Resource matching algorithm module, the specific functions are as follows:

[0184] Multi-dimensional considerations: In algorithm design, multiple factors are considered, including the professional skills of the skill group, historical processing efficiency, current workload, geographic location (for situations where on-site service is required), etc. This usually involves multi-objective optimization problems, using advanced algorithms such as genetic algorithms, particle swarm optimization, and reinforcement learning to solve the best matching solution.

[0185] Dynamic adaptability: Resource matching is not static, but is dynamically adjusted over time and as external conditions change. The system needs to monitor the status of the handler and the work order pool in real time, and continuously optimize the matching logic through continuous learning and feedback mechanisms to ensure efficient resource scheduling even during peak hours.

[0186] Dynamic work order generator, the specific functions are as follows:

[0187] Personalization: Based on the extracted key information and matching results, the dynamic ticket generator not only generates ticket content, but also adds personalized content based on the specific situation of each user. For example, it adjusts the transfer strategy based on the user's historical interaction records, or presets priority tags based on urgency.

[0188] Information integration: The work order will integrate service content details, estimated processing time, user contact information, recommended work order processing skill group and person in charge, to ensure that the team or individual receiving the work order can immediately obtain all necessary information and quickly start the service process.

[0189] Artificial intelligence algorithm module: Apply advanced machine learning models to achieve intelligent work order transfer decisions. Specific implementation steps include:

[0190] (1) Model construction and training: Select or customize a deep learning framework (such as RNN, Transformer) to build a matching model that can understand the complex characteristics of work orders and employee capabilities. During the training process, a large amount of historical work order and processing result data is used to enable the model to learn to identify the best matching mode between different work order characteristics (such as problem type, urgency) and employee expertise.

[0191] (2) Algorithm parameter tuning: Fine-tune model parameters through hyperparameter optimization algorithms (such as grid search, random search, Bayesian optimization, etc.) to ensure that the algorithm can take into account the urgency and difficulty of work orders while also taking into account employees' work preferences and current workload, ultimately achieving the goal of optimizing transfer decisions.

[0192] (3) Dynamic adaptability: The algorithm design must have the ability to self-learn and adjust so that it can automatically optimize the matching logic and improve the efficiency and accuracy of transfers over time based on new work scenarios and employee skill upgrades.

[0193] The work order transfer execution module is designed to seamlessly execute the transfer plan generated by the AI ​​algorithm to ensure a smooth process:

[0194] (1) Automated process design: Develop a highly automated work order flow system that eliminates the need for human intervention from receiving analysis results to executing transfer instructions, greatly improving work efficiency.

[0195] (2 Real-time monitoring and exception handling: Equipped with real-time monitoring tools to track the flow status of each work order. Once a transfer error or delay is found, an alarm is immediately triggered and remedial measures are automatically taken, such as reallocation or priority adjustment.

[0196] The following preparations were performed in this embodiment:

[0197] (1) Data collection: Periodically or in real time collect the work status data of staff members.

[0198] (2) Algorithm training: Regularly train AI algorithms to adapt to new data features and work requirements.

[0199] (3) Transfer decision: Make transfer decisions quickly through algorithm models based on real-time data.

[0200] (4) Optimization and adjustment: Based on the feedback results, adjust the algorithm parameters to achieve higher transfer accuracy and efficiency.

[0201] When there are multiple work orders or workers of the same type, this embodiment distinguishes them by numbering (such as "01", "02") to avoid confusion.

[0202] The “artificial intelligence algorithm” involved in this embodiment can select a suitable algorithm according to the specific implementation situation, such as a neural network, a decision tree, etc., and the algorithm model should be able to adaptively learn and optimize.

[0203] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A personalized work order transfer method based on intelligent voice AI, characterized in that: The method is as follows: Initiate voice request: The user initiates a voice service request through the user terminal; Speech recognition and information extraction: The intelligent speech AI server recognizes the speech service request and extracts key information: Matching work order processing strategy: The intelligent voice AI server retrieves the matching work order processing strategy from the database based on the extracted key information; Generate personalized work orders: Generate personalized work orders for users based on preset algorithms, combined with historical service records and implementation status of users; AI transfer: transfer personalized work orders to corresponding skill groups and persons in charge through intelligent voice AI servers; Voice feedback to users: Track service progress through intelligent voice AI, update work order information based on service status, and provide voice feedback to users on processing results.

2. The personalized work order transfer method based on intelligent voice AI according to claim 1 is characterized in that: The matching work order processing strategy is to determine the key information, combine the historical work order transfer situation, and recommend the processing skill group and person in charge of the new work order. The specific recommendation process is as follows: Determine the skill group of the work order. Each skill group is considered a category, and the order transfer recommendation is converted into a classification problem. Determine the person who will handle the specific work order in the corresponding skill group based on the classification.

3. The personalized work order transfer method based on intelligent voice AI according to claim 1 or 2 is characterized in that: The work order description is as follows: ① Directly use the original work order text as training expectation, and the Bert model will automatically segment words and perform training or inference; ②First use jieba to perform word segmentation and remove stop words.

4. The personalized work order transfer method based on intelligent voice AI according to claim 3 is characterized in that: The generation of personalized work orders includes identifying specific rules in the work order transfer; Identify specific rules in work order transfer, as follows: ① Shield alarm work orders and directly send them to NOC for processing; ② Capacity warning work orders are directly transferred to the cloud center; ③Zero-trust password reset work orders are handled by NOC; ④Whether the handler and skill group are directly specified in the work order content.

5. The personalized work order transfer method based on intelligent voice AI according to claim 4 is characterized in that: The handler and skill group are directly specified in the work order content. The judgment method is as follows: ① Use string matching to find the skill group name or personnel name directly in the work order description; ② Train the named entity recognition model to extract work handlers and skill groups; ③ Through the prompt engineering method, the large model answers through the few-shot method.

6. The personalized work order transfer method based on intelligent voice AI according to claim 5 is characterized in that: The details of AI order transfer are as follows: Fill out the work order according to the problem description; Perform word segmentation processing based on the work order content and obtain the work order content after word segmentation processing; Perform stop word filtering on the work order content after word segmentation, remove the stop words, and obtain the filtered work order content; Classify the filtered work order content through the classification model to obtain the classification results; According to the classification results, the transfer order queue is recommended by the recommendation model and assigned to the corresponding on-duty personnel.

7. A personalized work order transfer system based on intelligent voice AI, characterized in that: The system includes: A voice request initiation module is used for a user to initiate a voice service request through a user terminal; The speech recognition and information extraction module is used to identify the speech service request through the intelligent speech AI server and extract key information: The work order processing strategy matching module is used to retrieve matching work order processing strategies from the database based on the extracted key information through the intelligent voice AI server; Dynamic work order generator, which is used to generate personalized work orders for users based on preset algorithms, combined with historical service records and implementation status of users; The work order transfer execution module is used to transfer personalized work orders to the corresponding skill groups and responsible persons through the intelligent voice AI server; The voice feedback module is used to track the service progress through intelligent voice AI, update the work order information according to the service status, and feedback the processing results to the user in the form of voice.

8. The personalized work order dispatching system based on intelligent voice AI according to claim 7 is characterized in that: The system also includes an operation and maintenance management terminal, which uses a graphical interface. A secure data communication protocol is used between the operation and maintenance management terminal and the intelligent voice AI server and work order processing server to ensure data security.

9. The personalized work order transfer system based on intelligent voice AI according to claim 8 is characterized in that: The working process of the operation and maintenance management terminal is as follows: (1) The operation and maintenance personnel monitor the operating status of the intelligent voice AI server and the work order processing server through the operation and maintenance management terminal; (2) When an abnormal situation is found, the operation and maintenance personnel adjust the work order processing strategy in real time to ensure service quality; (3) The operation and maintenance management terminal can view the progress of work order processing in real time, which is convenient for management and scheduling.

10. The personalized work order dispatching system based on intelligent voice AI according to claim 9 is characterized in that: The single description method is as follows: ① Directly use the original work order text as training expectation, and the Bert model will automatically segment words and perform training or inference; ②First use jieba to perform word segmentation and remove stop words; The generation of personalized work orders includes identifying specific rules in the work order transfer; Identify specific rules in work order transfer, as follows: ① Shield alarm work orders and directly send them to NOC for processing; ② Capacity warning work orders are directly transferred to the cloud center; ③Zero-trust password reset work orders are handled by NOC; ④Whether the work order content directly specifies the handler and skill group; The handler and skill group are directly specified in the work order content. The judgment method is as follows: ① Use string matching to find the skill group name or personnel name directly in the work order description; ② Train the named entity recognition model to extract work handlers and skill groups; ③ Through the prompt engineering method, the large model answers through the few-shot method.

Images