Work order assignment method and device, computer equipment, medium and product

Abstract

The embodiment of the invention provides a work order dispatching method and device, computer equipment, a medium and a product, and relates to the field of artificial intelligence and the field of financial science and technology. The method comprises the following steps: inputting dialogue information corresponding to a work order into a natural language processing model to obtain a business type and implicit features of the work order; determining a target service scene of the work order according to the service type and the implicit features; determining at least one to-be-selected mechanism matched with the target service scene in the multiple mechanisms; and according to the historical processing data, the business type and the implicit features of each to-be-selected mechanism, determining a target assignment mechanism corresponding to the work order in the at least one to-be-selected mechanism. And through large-model semantic analysis and multi-dimensional scoring, manual misjudgment is reduced, and the mechanism matching accuracy is improved. Scenarized rules and implicit features are combined, complex business requirements are met, and work order full-process automation and resource optimization configuration are achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of artificial intelligence and the field of financial technology, and in particular to a work order dispatching method and device, a computer device, a medium and a product. BACKGROUND

[0002] In many fields such as finance, logistics, e-commerce, and public services, enterprises usually use work order systems to standardize the management of customer feedback (such as complaints, inquiries, suggestions, etc.) from multiple channels such as telephone, online chat, and email. The core link is that the specific demands of the work order must be accurately dispatched to internal institutions or departments with corresponding processing authority and ability.

[0003] Currently, work order dispatching is highly dependent on manual operation or static rules based on limited conditions. The former is inefficient and prone to errors, and the latter is inflexible and difficult to cope with complex and changing actual demands. With the development of customer service demands towards large-scale and complex, the existing methods have led to problems such as low efficiency, poor accuracy, long processing period, and cannot meet the operational needs of modern enterprise service intelligence and refinement. SUMMARY

[0004] The embodiments of the present application provide a work order dispatching method, device, computer device, medium and product to solve the problem of work order dispatching relying on manual operation, low efficiency and poor accuracy in the prior art.

[0005] In a first aspect, the embodiments of the present application provide a work order dispatching method, the method comprising:

[0006] inputting the dialogue information corresponding to the work order into a natural language processing model to obtain the business type and the implicit feature of the work order;

[0007] determining the target business scenario of the work order according to the business type and the implicit feature;

[0008] determining at least one candidate institution in the multiple institutions that matches the target business scenario;

[0009] determining the target dispatching institution corresponding to the work order from the at least one candidate institution according to the historical processing data of each candidate institution, the business type and the implicit feature.

[0010] In a possible implementation, the implicit feature includes the problem complexity and the number of collaboration departments; and determining the target business scenario of the work order according to the business type and the implicit feature comprises:

[0011] in the case that the problem complexity is greater than a preset degree and the number of collaboration departments is multiple, determining that the target business scenario is a preset business scenario;

[0012] In a case that the complexity of the problem is less than or equal to the preset degree or the number of the cooperative departments is not multiple, the target business scenario is obtained by matching the business type with the preset scenario library.

[0013] In a possible implementation, the target dispatching institution corresponding to the work order is determined from the at least one candidate institution according to the historical processing data of each candidate institution, the business type, and the implied feature, and the method comprises the following steps of:

[0014] The dispatching score of each candidate institution is determined according to the historical processing data of each candidate institution, the business type, and the implied feature.

[0015] The target dispatching institution corresponding to the work order is determined according to the dispatching score of each candidate institution.

[0016] In a possible implementation, the implied feature comprises an emergency degree; the dispatching score of each candidate institution is determined according to the historical processing data of each candidate institution, the business type, and the implied feature, and the method comprises the following steps of:

[0017] The institution evaluation index of each candidate institution is determined according to the historical processing data of each candidate institution.

[0018] The business matching degree index is determined according to the business type and the business type corresponding to each candidate institution.

[0019] The load adaptation degree is determined according to the current load of each candidate institution.

[0020] The work order priority index of each candidate institution is determined according to the emergency degree.

[0021] The dispatching score of each candidate institution is determined according to the institution evaluation index, the business matching degree index, the load adaptation degree, and the work order priority index.

[0022] In a possible implementation, the dispatching score of each candidate institution is determined according to the institution evaluation index, the business matching degree index, the load adaptation degree, and the work order priority index, and the method comprises the following steps of:

[0023] The weight of each index is determined according to the user portrait data corresponding to the work order.

[0024] The dispatching score of each candidate institution is determined according to the weight of each index, the institution evaluation index, the business matching degree index, the load adaptation degree, and the work order priority index.

[0025] In a possible implementation, the target dispatching institution corresponding to the work order is determined according to the dispatching score of each candidate institution, and the method comprises the following steps of:

[0026] verify whether a load adaptation degree of a first institution with the highest assignment score among the at least one candidate institution is less than a preset threshold value;

[0027] in a case where the load adaptation degree of the first institution is less than the preset threshold value, a second institution with a second highest assignment score among the at least one candidate institution is determined as the target assignment institution, and in a case where the load adaptation degree of the first institution is greater than or equal to the preset threshold value, the first institution is determined as the target assignment institution.

[0028] In a second aspect, the embodiments of the present application provide a work order assignment device, the device comprising:

[0029] a voice analysis module configured to input conversation information corresponding to the work order into a natural language processing model to obtain a business type and an implicit feature of the work order;

[0030] a rule engine module configured to determine a target business scenario of the work order according to the business type and the implicit feature;

[0031] a first screening module configured to determine at least one candidate institution matching the target business scenario from a plurality of institutions;

[0032] a second screening module configured to determine a target assignment institution corresponding to the work order from the at least one candidate institution according to historical processing data of each candidate institution, the business type and the implicit feature.

[0033] In a third aspect, the embodiments of the present application provide a computer device, comprising a memory and a processor; the memory stores computer execution instructions; the processor executes the computer execution instructions stored in the memory, so that the processor executes the first aspect and / or various possible implementation manners of the first aspect.

[0034] In a fourth aspect, the embodiments of the present application provide a computer readable storage medium, the computer readable storage medium stores computer execution instructions, and the computer execution instructions are executed by a processor to implement the first aspect and / or various possible implementation manners of the first aspect.

[0035] In a fifth aspect, the embodiments of the present application provide a computer program product, comprising a computer program, the computer program is executed by a processor to implement the first aspect and / or various possible implementation manners of the first aspect.

[0036] The work order dispatching method, device, computer device, medium and product provided by the embodiments of the present application relate to the fields of artificial intelligence and financial technology. The method comprises: inputting conversation information corresponding to a work order into a natural language processing model to obtain a business type and implicit features of the work order; determining a target business scenario of the work order according to the business type and the implicit features; determining at least one selected institution that matches the target business scenario from a plurality of institutions; and determining a target dispatching institution corresponding to the work order from the at least one selected institution according to historical processing data of each selected institution, the business type and the implicit features. Through semantic analysis of a large model and multi-dimensional scoring, manual misjudgment is reduced, and the accuracy of institution matching is improved. The combination of scenario-based rules and implicit features adapts to complex business requirements, realizes work order full-process automation and resource optimization configuration. BRIEF DESCRIPTION OF DRAWINGS

[0037] The accompanying drawings, which are incorporated herein and form a part of the specification, illustrate embodiments consistent with the present application and, together with the description, further serve to explain the principles of the present application.

[0038] Figure 1 Flowchart of the work order dispatching method provided by the present application Figure 1

[0039] Figure 2 Flowchart of the work order dispatching method provided by the present application Figure 2

[0040] Figure 3 Flowchart of the work order dispatching method provided by the present application Figure 3

[0041] Figure 4 Structural diagram of the work order dispatching device provided by the present application

[0042] Figure 5 Structural diagram of the computer device provided by the present application.

[0043] The specific embodiments of the present application have been shown in the above drawings, and will be described in more detail hereinafter. These drawings and textual descriptions are not intended to limit the scope of the concept of the present application by any means, but to illustrate the concept of the present application to those skilled in the art by referring to specific embodiments. DETAILED DESCRIPTION

[0044] ​​​The exemplary embodiments will be described in detail herein with reference to the attached drawings. The following description is made with reference to the accompanying drawings in which like reference numerals refer to like elements, and the term "exemplary" is used herein to mean "serving as an example, instance, or illustration." Any implementation described herein as "exemplary" is not necessarily to be construed as preferred or advantageous over other implementations. Unless specifically stated otherwise, it is appreciated that, throughout this specification, functions, steps, or actions described can be performed in an order different from the order in which they are described. For example, functions, steps, or actions can be performed in an order that is interleaved or that otherwise deviates from the order in which they are described.

[0045] In the description of the present application, the terms "first", "second", etc. are used only for the purpose of description, and should not be understood as indicating or implying relative importance or implying the number of the technical features indicated. Therefore, the features defined as "first", "second" can explicitly or implicitly include one or more of the features. In the description of the present application, the meaning of "a plurality of" is two or more, unless otherwise specifically limited.

[0046] It should be noted that the user information (including but not limited to user equipment information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in the present application are all information and data authorized by the user or authorized by all parties, and the collection, storage, use, processing, transmission, provision, disclosure and application of related data comply with relevant laws, regulations and standards of relevant countries and regions, necessary security measures are taken, do not violate public order and good customs, and provide corresponding operation portal for user to choose authorization or refusal.

[0047] And the present application involves big data analysis of user information (including but not limited to personal biological characteristics, identity data, consumption data, asset data, electronic terminal operation data, etc.), and uses artificial intelligence technology to make automatic decision, and makes technical scheme with significant influence on personal rights and interests based on automatic decision result, provides corresponding operation portal for user to choose to agree or refuse automatic decision result; if the user chooses to refuse, enter the expert decision process.

[0048] It should be noted that the work order dispatching method, device, computer equipment, medium and product provided by the present application can be used in the field of artificial intelligence, and can also be used in the field of financial technology, and can also be used in any field other than the field of artificial intelligence and the field of financial technology. The application does not limit the application field of the work order dispatching method, device, computer equipment, medium and product.

[0049] In the current enterprise customer service practice, especially in high-frequency interaction scenarios such as finance, logistics, e-commerce, public services, etc., the feedback such as complaints, inquiries, suggestions and the like submitted by customers through telephone, online chat, email and other diversified channels needs to be transferred and processed through a ticket system. A key technical link is that the specific customer demand carried by the ticket needs to be automatically and accurately matched and assigned to an institution or department with corresponding processing authority and professional ability.

[0050] However, the existing ticket assignment technology has obvious bottlenecks. First, a large number of enterprises still mainly rely on manual methods to handle assignment tasks, and customer service personnel need to read and understand the content of the ticket and select the processing department based on personal experience. This method not only has low assignment efficiency, but also has poor stability and strong dependence on personnel experience. Second, some systems use a static rule base based on keyword matching, such as assigning the ticket to the finance department if "refund" appears in the ticket. This method is difficult to understand complex context semantics and has limited processing capacity for novel or composite demands due to its lack of flexibility.

[0051] Therefore, the existing technology cannot meet the massive, heterogeneous and dynamically changing ticket assignment requirements, resulting in slow overall service response and decreased customer satisfaction, and has become a key bottleneck for improving enterprise operational efficiency. There is an urgent need for a new method that can intelligently understand ticket requirements and achieve accurate and efficient assignment.

[0052] To solve the above technical problems, the embodiments of the present application provide a ticket assignment method applied to a customer service system. As shown in Figure 1 The method comprises the following steps:

[0053] Step 101, inputting the dialogue information corresponding to the ticket into a natural language processing model to obtain the business type and implicit features of the ticket;

[0054] Step 102, determining the target business scenario of the ticket according to the business type and the implicit features;

[0055] Step 103, determining at least one candidate institution in the multiple institutions that matches the target business scenario;

[0056] Step 104, determining the target assignment institution corresponding to the ticket from the at least one candidate institution according to the historical processing data of each candidate institution, the business type and the implicit features.

[0057] The dialogue information corresponding to the work order refers to the interaction record or dialogue details of the work order, which includes the trigger of the work order and the communication process, and also carries the information of the customer triggering the dialogue, such as customer identity information, customer level information, etc. Its form can be text dialogue or voice dialogue. In the customer service system, the work order dialogue information obtained through recording, voice-to-text technology, etc. In an embodiment, the dialogue information includes problem description, urgency level (such as high / medium / low), priority (such as P0-P3), customer level, historical processing record, etc.

[0058] In simple terms, dialogue information is a digital record of the complete communication session between the customer and the customer service representative. Its core purpose is to: accurately restore the service scenario: avoid misunderstandings caused by information transmission bias. Quality monitoring and training: evaluate the professionalism, service attitude, and problem-solving ability of customer service personnel. Dispute arbitration: when customer complaints or disputes occur, it serves as the most objective evidence. Business analysis: by analyzing the dialogue content, find product problems, customer pain points, and optimize processes. This includes the "substance" of the communication, which is the most core part, containing all the details of the exchange between the two parties.

[0059] The customer service system will process the original recording through voice recognition (ASR) and natural language processing (NLP) technology. Original recording file: the most original audio data, ensuring the integrity of the information and the effectiveness of the evidence. Transcribed text: convert speech into readable text through ASR technology. This is the basis for analysis and retrieval. High-quality transcription will include: speaker separation (automatically distinguish between customer and customer service representative's speech), timestamp (each sentence corresponds to a specific time point), annotation of some emotional words or key non-verbal sounds (such as "[laughter]", "[silence for 5 seconds]"). Dialogue analysis insights (generated by NLP engine): this is the value-added part, the system will automatically analyze the text and extract key information: sentiment analysis: real-time or post-event analysis of the emotional state of the customer (positive, neutral, negative), and even the emotional fluctuation curve. Key topic / keyword extraction: automatically identify the core issues discussed in the dialogue, such as "refund", "password reset", "express delivery delay". Customer service performance indicators: speech rate / clarity: whether the customer service representative's expression is clear and understandable. Silence time: the customer service representative's excessive waiting time for the customer. Interruption frequency: whether the customer service representative frequently interrupts the customer. Forbidden language detection: whether prohibited words are used. Compliance check: automatically detect whether the customer service representative has read the necessary compliance statements (such as "this call may be recorded for service quality improvement") during the call.

[0060] The natural language processing model is used to extract keywords in the dialogue information, so as to realize business type and implicit feature extraction. The natural language processing model can be a model constructed based on BERT (Bidirectional Encoder Representations from Transformers) or GPT (Generative Pre-trained Transformer). If the dialogue information is a call recording file between a customer and a customer service, the call recording file is input into the natural language processing model, and the key entities in the dialogue are extracted, such as “logistics delay” and “account freezing”, and then the business type and the implicit features are output.

[0061] Specifically, the target needs to be clear: through a large model, the problem description of the work order is classified and features are extracted. Specifically, two main functions are to be realized: (1) business type: mapping the problem description to predefined business major and minor categories (for example, two-level classification); (2) extract implicit features: such as customer emotion, problem complexity, number of collaboration departments (that is, whether it involves multi-department collaboration).

[0062] Implementation steps: data preparation: prepare labeled data, including work order problem description, corresponding business major / minor category, and implicit feature label (emotion, complexity, etc.). Model selection: for business type, it can be regarded as a text classification task, and pre-trained models such as BERT can be used for fine-tuning. For implicit feature extraction, it can be regarded as multiple different classification tasks (or multi-label classification, depending on whether the features are independent). Model training: train the business type model and the implicit feature extraction model respectively, or design a multi-task learning model to output multiple results simultaneously. Model deployment: deploy the trained model as an API service to receive work order problem descriptions and return structured results.

[0063] Each step is described in detail below. Data preparation: there needs to be a labeled data set. Suppose we have a work order data set, each piece of data includes: problem description (text), business major category (label), business minor category (label), emotion label (such as: angry, normal, happy, etc.), complexity label (such as: simple, complex), number of collaboration departments; if there is no ready-made labeled data, manual data labeling is needed first. Model selection: pre-trained language models such as BERT can be used because they perform well on various NLP tasks.

[0064] Solution one: multiple single-task models, including training a model for business type (may need to predict major and minor categories simultaneously, can consider hierarchical classification or joint training); training a model for emotion classification; training a model for complexity classification; training a model for collaboration department quantity classification.

[0065] Solution two: Multi-task learning, design a model that shares BERT encoding layers, then has multiple output heads, corresponding to business type, sentiment classification, complexity classification, and collaboration department quantity classification.

[0066] The advantage of multi-task learning is that it can share representations and potentially improve generalization, but it requires adjusting the loss weights of multiple tasks and the training process may be more complex. Considering that these tasks are based on the same problem description and there may be correlations between tasks, we can choose multi-task learning.

[0067] Example of multi-task learning: Model structure: input: ticket problem description (text), use BERT encoder to get text representation.

[0068] Multiple output heads: Head 1: business category classification (softmax, number of categories = number of business categories), Head 2: business subcategory classification (softmax, number of categories = number of business subcategories) Note: Business subcategories may depend on business categories, but here we can simply treat them as independent classification or design a hierarchical structure, Head 3: sentiment classification (softmax, number of categories = number of sentiment labels), Head 4: complexity classification (softmax, number of categories = number of complexity labels), Head 5: collaboration department quantity (binary classification, sigmoid).

[0069] Loss function: one loss for each task, total loss is the weighted sum. Specifically, total loss = w1 × loss1 (business category) + w2 × loss2 (business subcategory) + w3 × loss3 (sentiment) + w4 × loss4 (complexity) + w5 × loss5 (collaboration department quantity). The weights w1-w5 can be set manually through grid search or according to task importance. Use labeled data to fine-tune the model.

[0070] After training, deploy the model as an API service. When there is a new ticket problem, call the API, input the problem description, the model returns:

[0071] {

[0072] "business category": "logistics",

[0073] "business subcategory": "complaint",

[0074] "sentiment": "angry",

[0075] "complexity": "complex",

[0076] "collaboration department quantity": "multiple"

[0077] }

[0078] The business type and the implicit feature of the work order can be obtained through the method. Then, the business scenario involved in the work order is determined through the business type and the implicit feature. For example, if the business type is "logistics-complaint", the matched business scenario is "complaint handling scenario"; and the scenario is dynamically adjusted based on the implicit feature, for example, if the problem complexity is high and multiple departments are involved, the scenario is upgraded to "cross-department cooperation scenario". Specifically, the matched target business scenario is output through a preset rule engine, with the business type and the implicit feature as inputs.

[0079] The candidate agency refers to an agency matched with the business scenario of the work order. The agencies are initially screened through the business scenario, and only the agencies matched with the business scenario are retained, such as the "complaint handling scenario" which only screens the complaint department.

[0080] The historical processing data refers to the historical processing of the same type of business by the agency, such as the historical processing efficiency of the same type of work order, the proportion of processing the same type of business, and customer satisfaction. It should be noted that the above-mentioned same type refers to the business type of the current work order, and is specific to the business subcategory. The dispatch score of each candidate structure is determined through the historical processing data of each candidate structure, the business type, and the implicit feature, and then the candidate structure with the highest dispatch score is selected as the target dispatch agency, and the work order is dispatched to the target dispatch agency.

[0081] In the method provided in the above embodiment, the dialogue information corresponding to the work order is input into a natural language processing model to obtain the business type and the implicit feature of the work order; the target business scenario of the work order is determined according to the business type and the implicit feature; at least one candidate agency matched with the target business scenario is determined from a plurality of agencies; and the target dispatch agency corresponding to the work order is determined from the at least one candidate agency according to the historical processing data of each candidate agency, the business type, and the implicit feature. Through semantic analysis of a large model and multi-dimensional scoring, manual misjudgment is reduced, and the accuracy of agency matching is improved. The combination of scenario-based rules and implicit features adapts to complex business needs, and realizes work order full-process automation and resource optimization.

[0082] In one of the embodiments, the implicit feature includes problem complexity and the number of collaboration departments; and the target business scenario of the work order is determined according to the business type and the implicit feature, including:

[0083] In the case where the problem complexity is greater than a preset degree and the number of collaboration departments is multiple, the target business scenario is determined as a preset business scenario;

[0084] In the case where the problem complexity is less than or equal to a preset degree or the number of collaboration departments is not multiple, the target business scenario is obtained by matching the business type with a preset scenario library.

[0085] Among them, the preset business scenario refers to a business scenario that requires special handling, in which the work order cannot be resolved by a single department. The preset level indicates that the problem is highly complex, and the number of collaborating departments is multiple, indicating that the work order involves multiple departments.

[0086] When the complexity of the problem exceeds the preset level and there are multiple collaborating departments, the target business scenario is determined as the preset business scenario, such as "cross-departmental collaboration scenario". When the complexity of the problem is less than or equal to the preset level, or there are fewer than multiple collaborating departments, the preset rule engine is matched with the preset scenario library to obtain the target business scenario corresponding to the business type, such as "logistics-complaint" → "complaint handling scenario".

[0087] The method provided in the above embodiments dynamically adjusts the business scenario through implicit features, thereby improving the flexibility and adaptability of scenario recognition and avoiding situations where assignment delays or misassignments occur, customer problems are not resolved in a timely manner, leading to an increase in the rate of repeated complaints and affecting the company's service reputation.

[0088] In one embodiment, such as Figure 2 As shown, based on the historical processing data, business type, and implicit characteristics of each candidate organization, the target dispatch organization corresponding to the work order is determined from at least one candidate organization, including:

[0089] Step 201: Determine the allocation score for each candidate institution based on its historical processing data, business type, and implicit characteristics.

[0090] Step 202: Determine the target dispatching organization corresponding to the work order based on the dispatching score of each candidate organization.

[0091] By analyzing the historical processing data, business types, and implicit characteristics of each candidate structure, the matching degree between each candidate structure and the work order is determined from both the work order and candidate organization perspectives. An assignment score characterizes the rationality of assigning the work order to each candidate organization. The assignment score is determined by multiple evaluation indicators, such as organization evaluation indicators, load adaptability, work order priority indicators, and business matching indicators. After determining the score for each indicator based on the historical processing data, business type, and implicit characteristics of each candidate structure, the assignment score is determined by the weight of each indicator.

[0092] The business matching degree index refers to the matching degree of the business subcategory output by the large model and the field of expertise of the candidate institution, such as "complaint department" being good at "complaint handling". The institution evaluation index is a pre-set capability parameter of the candidate institution, such as response speed, historical processing efficiency, professionalism, etc. The load adaptation degree is the backlog rate, average processing time, resource availability, etc. of the current institution. The work order priority index is the urgency or priority, such as high-urgency work orders being assigned to institutions with fast response speed.

[0093] In one embodiment, the assignment score = a x business matching degree index + b x institution evaluation index + g x load adaptation degree + d x work order priority index; where a, b, g, d are weight coefficients. In one embodiment, they are set to 0.4, 0.3, 0.2, 0.1 respectively, reflecting the importance of each index.

[0094] The assignment score result is used for ranking, and institutions with high scores are determined as target assignment institutions. The specific calculation process is detailed:

[0095] Business matching degree index: The semantic similarity between the business subcategory of the work order and the field of expertise of the candidate institution is calculated by natural language processing technology (such as cosine similarity), with a score range of 0-1. For example, the business subcategory of the work order is "logistics-complaint", and the field of expertise of the candidate institution is "complaint handling", with a similarity of 0.9 obtained through text matching.

[0096] Principle: The business matching degree index ensures that work orders are assigned to specialized candidate institutions, improving processing accuracy and efficiency. The weight a=0.4 is the highest, emphasizing that matching degree is the primary factor in assigning work orders, avoiding misallocation.

[0097] The institution evaluation index is a composite index, and the institution evaluation index = w1 x response speed + w2 x historical processing efficiency + w3 x professionalism + w4 x customer satisfaction;

[0098] Wherein, response speed: based on average response time (such as <2 hours = 1.0, 2-4 hours = 0.75, >4 hours = 0.5). Historical processing efficiency: based on average resolution time (such as <24 hours = 1.0, 24-72 hours = 0.6, >72 hours = 0.3). Professionalism: based on the proportion of work orders handled in this business subcategory (such as >80%=1.0, 30%-80%=0.75, <30%=0.5). Customer satisfaction: based on customer ratings (such as >4.5 / 5 = 1.0, 3.0-4.5 = 0.7, <3.0 = 0.5). w1, w2, w3, w4 are sub-weights, which need to be pre-set (such as each taking 0.25, or adjusted according to business needs).

[0099] Principle: The agency evaluation index reflects the comprehensive ability of the selected agency, including speed, efficiency, professionalism and customer feedback. The weight β = 0.3 indicates that the agency capability is an important factor, but it is less important than the business matching degree, ensuring that agencies with strong capabilities but not completely matching also have opportunities.

[0100] The load adaptation degree is dynamically calculated based on the real-time load state, and the formula is: load adaptation degree = 1-(current load rate / maximum load threshold). Current load rate = current work order number / maximum processing capacity (for example, the maximum processing capacity of the agency is 100 orders / day, and there are currently 80 orders, so the load rate = 0.8).

[0101] The maximum load threshold is usually set to 90% (adjustable), and if the load rate exceeds the threshold, the load adaptation degree is forced to 0 to avoid overloading the agency.

[0102] Principle: The load adaptation degree balances the workload of each selected agency to prevent some agencies from being overloaded while others are idle. The weight γ = 0.2 indicates that the load situation is the third most important factor, ensuring system stability and real-time performance.

[0103] The work order priority index is not a separate value, but the influence of the work order urgency on the total score is reflected by adjusting the weight δ. The value of δ is dynamically set according to the work order urgency:

[0104] High-urgency work order: δ = 0.15 (increase the weight by 10%, i.e. 0.1→0.15).

[0105] Medium-urgency work order: δ = 0.1 (weight remains unchanged).

[0106] Low-urgency work order: δ = 0.05 (weight reduced by 50%).

[0107] In the scoring formula, this item is actually δ x 1 (i.e. δ itself), so the urgency directly affects the total score as a bonus item.

[0108] Principle: The work order priority index ensures that high-urgency work orders are assigned to agencies with fast response speed, improving processing timeliness. The weight δ = 0.1 is the lowest because the urgency is a work order attribute rather than an agency attribute, avoiding excessive impact on the fairness of distribution.

[0109] In the method provided by the above embodiment, the business scenario label of the customer's appeal is extracted through natural language processing technology, and the scores are dynamically calculated in combination with multi-dimensional parameters such as agency weight, load state, and urgency, to realize precise matching and resource optimization configuration in work order distribution.

[0110] In one of the embodiments, as shown in Figure 3 , the distribution score of each selected agency is determined according to the agency evaluation index, the business matching degree index, the load adaptation degree, and the work order priority index, including:

[0111] Step 301, according to the user portrait data corresponding to the work order, determine the weight of each indicator;

[0112] Step 302, according to the weight of each indicator, and the institution evaluation indicator, the business matching degree indicator, the load adaptation degree, the work order priority indicator, determine the distribution score of each candidate institution.

[0113] Among them, the user portrait data includes customer historical complaint records, service preferences (such as preferring online / offline processing), customer levels (VIP / ordinary). Dynamically adjust the score parameters, such as the weight of the work order priority indicator of VIP customer work order is increased to 0.2, and is preferentially allocated to institutions with fast response speed. For high-frequency complaint customers, increase the professional degree weight (w3) of the institution to 0.4, and preferentially allocate to institutions with high efficiency in handling similar problems.

[0114] Through the user portrait data, dynamically adjust the indicator weight, so that the average response time of high emergency work orders is shortened, and the customer satisfaction is improved.

[0115] In one embodiment, according to the distribution score of each candidate institution, determine the target distribution institution corresponding to the work order, comprising:

[0116] Verify whether the load adaptation degree of a first institution with the highest distribution score in the at least one candidate institution is less than a preset threshold;

[0117] In the case where the load adaptation degree of the first institution is less than the preset threshold, the second institution with the second highest distribution score in the at least one candidate institution is taken as the target distribution institution, and in the case where the load adaptation degree of the first institution is greater than or equal to the preset threshold, the first institution is taken as the target distribution institution.

[0118] Generally, after determining the distribution score of each candidate structure, the first institution with the highest distribution score is taken as the target distribution institution. However, in this embodiment, the load state of the first institution needs to be verified through the load adaptation degree to determine whether the first institution can handle the work order.

[0119] Each candidate institution will be calculated a distribution score according to a comprehensive score formula (such as the aforementioned "institution score calculation" formula). The score is usually based on business matching degree, institution weight, real-time load state and work order urgency, and the higher the score, the more suitable the institution is for handling the work order.

[0120] Example: Assuming that there are three candidate institutions, the distribution scores are: institution A: score 95 (first institution), institution B: score 90 (second institution), and institution C: score 85 (third institution); among them, institution A is the first institution with the highest distribution score, and institution B is the second institution.

[0121] Load Adaptation: This is an indicator reflecting the real-time load situation of the agency, usually calculated based on a formula (such as Load Adaptation = 1 - (Current Load Rate / Maximum Load Threshold)). The indicator value ranges from 0 to 1, and the smaller the value, the heavier the load of the agency (for example, a load adaptation close to 0 indicates that the agency is almost full load).

[0122] Pre-set Threshold: A critical value (such as 0.2) preset by the system to determine whether the agency is overloaded. The threshold can be adjusted according to business needs (for example, if the system allows the agency load rate not to exceed 80%, the threshold may be set to 0.2).

[0123] Verification Logic: Check if the load adaptation of the first agency (Agency A) is less than the pre-set threshold. If it is less than the threshold, it means that Agency A is currently overloaded, and even if the score is high, it may not be able to handle new work orders in time. If it is greater than or equal to the threshold, it means that the load of Agency A is within an acceptable range, and it can be assigned work orders.

[0124] Select the target dispatching agency according to the verification result: Case one, the load adaptation of the first agency (Agency A) is less than the pre-set threshold (overloaded), the corresponding decision: do not choose the first agency (Agency A), but choose the second agency (Agency B) with the second highest dispatch score as the target dispatching agency. Avoid assigning work orders to overloaded agencies, even if their scores are high. This helps prevent Agency A from being overloaded, leading to processing delays, efficiency declines, or system bottlenecks. Selecting the second agency (Agency B) as an alternative not only ensures a relatively high score, but also achieves load balancing. For example, if the load adaptation of Agency A is 0.1 (load rate 90%), the pre-set threshold is 0.2, then Agency A is overloaded. At this time, the target agency is Agency B.

[0125] Case two, the load adaptation of the first agency (Agency A) is greater than or equal to the pre-set threshold (normal load), directly select the first agency (Agency A) as the target dispatching agency. At this time, Agency A not only has the highest score, but also has a reasonable load, making it the best choice for work order distribution. For example, if the load adaptation of Agency A is 0.3 (load rate 70%), the pre-set threshold is 0.2, then Agency A is normally loaded. At this time, the target agency is Agency A.

[0126] The method provided in the above embodiment prevents work orders from being concentrated in high-scoring but busy agencies through load adaptation checks, improving system stability. Dispatch scores ensure the professionalism and quality of work order distribution (such as business matching degree), while load adaptation ensures real-time efficiency (such as processing speed). Based on real-time load data, decisions can be made to adapt to changing system conditions, achieving intelligent load balancing.

[0127] An example is given to illustrate the above all embodiments:

[0128] The conversation information of the input work order, including:

[0129] Problem description: "East China regional customer complains about logistics delay, and involves warehouse department cooperation."

[0130] Urgency: high;

[0131] Priority: P1;

[0132] The determination process of the target assignment agency of the work order is:

[0133] S1, large model output: business subcategory: "logistics-complaint"; implied features: involves multi-department cooperation, customer emotion angry.

[0134] S2, scene matching: rule engine matches to "cross-department cooperation scene".

[0135] S3, agency scoring: candidate agencies: East China Logistics Center-Complaint Department (match degree 0.9), East China Warehouse Center-Complaint Department (match degree 0.7).

[0136] Scoring calculation:

[0137] East China Logistics Center-Complaint Department: comprehensive score = 0.4x0.9+0.3x0.85+0.2x0.3+0.1x1.15=0.79.

[0138] East China Warehouse Center-Complaint Department: comprehensive score = 0.4x0.7+0.3x0.6+0.2x0.5+0.1x1.15=0.63.

[0139] S4, output result: select East China Logistics Center-Complaint Department.

[0140] The above method improves the accuracy of agency matching through large model semantic understanding and multi-dimensional scoring. Real-time load monitoring and dynamic weight adjustment avoid agency overload. Scene rules combined with implied features adapt to complex business needs. Scoring formula parameters can be configured to support customized needs of different industries / enterprises.

[0141] It should be understood that although each step in the flowchart involved in the above embodiments is shown in sequence according to the arrow, the steps are not necessarily executed in the order indicated by the arrow. Unless otherwise specified herein, the execution of the steps is not strictly limited in sequence, and the steps can be executed in other orders. Moreover, at least part of the steps in the flowchart involved in the above embodiments can include multiple steps or multiple stages, which are not necessarily executed at the same time, but can be executed at different times, and the execution order of the steps or stages is not necessarily sequential, but can be alternately executed with other steps or steps or stages in other steps.

[0142] Based on the same inventive concept, the embodiments of the present application also provide a work order dispatching device. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme described in the above method, and therefore the specific limitations in one or more work order dispatching device embodiments provided below can refer to the limitations of the work order dispatching method described above, which will not be repeated here.

[0143] In one embodiment, as shown in FIG. 4, Figure 4 The work order dispatching device includes a voice analysis module 401, a rule engine module 402, a first screening module 403, and a second screening module 404, wherein:

[0144] The voice analysis module 401 is configured to input the dialogue information corresponding to the work order into a natural language processing model to obtain the business type and the implicit feature of the work order.

[0145] The rule engine module 402 is configured to determine the target business scenario of the work order according to the business type and the implicit feature.

[0146] The first screening module 403 is configured to determine at least one candidate institution matching the target business scenario from a plurality of institutions.

[0147] The second screening module 404 is configured to determine the target dispatching institution corresponding to the work order from the at least one candidate institution according to the historical processing data of each candidate institution, the business type, and the implicit feature.

[0148] In one possible implementation, the implicit feature includes the problem complexity and the number of cooperating departments, and the rule engine module 402 is specifically configured to:

[0149] In the case where the problem complexity is greater than a preset degree and the number of cooperating departments is multiple, the target business scenario is determined as a preset business scenario.

[0150] In a case where the complexity of the problem is less than or equal to a preset degree or the number of the cooperative departments is not multiple, the target business scenario is obtained by matching the business type with a preset scenario library.

[0151] In a possible implementation, the second screening module 404 is specifically configured to:

[0152] determine a dispatch score of each candidate institution according to the historical processing data, the business type, and the implicit feature of each candidate institution;

[0153] determine the target dispatch institution corresponding to the work order according to the dispatch score of each candidate institution.

[0154] In a possible implementation, the second screening module 404 is specifically configured to:

[0155] determine an institution evaluation index of each candidate institution according to the historical processing data of each candidate institution;

[0156] determine a business matching degree index according to the business type and the business type corresponding to each candidate institution;

[0157] determine a load adaptation degree according to the current load of each candidate institution;

[0158] determine a work order priority index of each candidate institution according to the urgency degree;

[0159] determine the dispatch score of each candidate institution according to the institution evaluation index, the business matching degree index, the load adaptation degree, and the work order priority index.

[0160] In a possible implementation, the second screening module 404 is specifically configured to:

[0161] determine the weight of each index according to the user portrait data corresponding to the work order;

[0162] determine the dispatch score of each candidate institution according to the weight of each index, and the institution evaluation index, the business matching degree index, the load adaptation degree, and the work order priority index.

[0163] In a possible implementation,

[0164] In a possible implementation, the second screening module 404 is specifically configured to:

[0165] verify whether the load adaptation degree of a first institution with the highest dispatch score in the at least one candidate institution is less than a preset threshold;

[0166] In a case where the load adaptation degree of the first institution is less than a preset threshold, a second institution with a second highest dispatch score in the at least one candidate institution is taken as the target dispatch institution; and in a case where the load adaptation degree of the first institution is greater than or equal to the preset threshold, the first institution is taken as the target dispatch institution.

[0167] The modules in the work order dispatching apparatus can be implemented by software, hardware, or a combination thereof. The modules can be embedded in a processor in a computer device or independent of the processor, or stored in a memory in the computer device to be invoked by the processor to perform operations corresponding to the modules.

[0168] Figure 5 A structural diagram of a computer device is provided in the present application. As shown in Figure 5 The computer device 50 provided in the present embodiment includes at least one processor 501 and a memory 502. Optionally, the device 50 further includes a communication component 503. The processor 501, the memory 502, and the communication component 503 are connected through a bus 504.

[0169] In the implementation process, the at least one processor 501 executes computer execution instructions stored in the memory 502, so that the at least one processor 501 performs the above method.

[0170] The specific implementation process of the processor 501 can refer to the above method embodiments, which have similar implementation principles and technical effects, and will not be described here in detail.

[0171] In the above embodiments, it should be understood that the processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the disclosed method can be directly embodied as hardware processor execution or combined execution by hardware and software modules in the processor.

[0172] The memory can include a random access memory (RAM), and can also include a non-volatile memory (NVM), such as at least one disk memory.

[0173] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. The bus can be divided into an address bus, a data bus, a control bus, etc. For the convenience of representation, the bus in the drawings of the present application does not limit to only one bus or one type of bus.

[0174] The present application also provides a computer program product, comprising a computer program, which, when executed by a processor, implements the method described above.

[0175] The present application also provides a computer readable storage medium, which stores computer execution instructions, and when a processor executes the computer execution instructions, the method described above is implemented.

[0176] The readable storage medium described above can be realized by any type of volatile or non-volatile storage device or their combination, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk or optical disk. The readable storage medium can be any available medium that can be accessed by a general or special purpose computer.

[0177] An exemplary readable storage medium is coupled to the processor, so that the processor can read information from the readable storage medium, and can write information to the readable storage medium. Of course, the readable storage medium can also be an integral part of the processor. The processor and the readable storage medium can be located in an application specific integrated circuit (ASIC). Of course, the processor and the readable storage medium can also exist as discrete components in the device.

[0178] The division of units is only a logical functional division, and in actual implementation, there can be another division manner, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the units shown or discussed can be indirect coupling or communication connection through some interfaces, devices or units, which can be electrical, mechanical or other forms.

[0179] The units described as separate components may or may not be physically separate, and the components displayed as units may or may not be physical units, i.e., may be located in one place, or may be distributed to multiple network units. Part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment scheme.

[0180] In addition, each functional unit in various embodiments of the application can be integrated into one processing unit, or each unit can exist physically, or two or more units can be integrated into one unit.

[0181] If the function is realized in the form of a software function unit and sold or used as an independent product, it can be stored in a computer readable storage medium. Based on this understanding, the technical solutions of the application essentially or the part that contributes to the prior art or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the various embodiment methods of the application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various program code storage media.

[0182] Those skilled in the art can understand that all or part of the steps of the above-mentioned method embodiments can be completed by program instruction related hardware. The aforementioned program can be stored in a computer readable storage medium. The program executes the steps including the above-mentioned method embodiments when executed; and the aforementioned storage medium includes: ROM, RAM, magnetic disk or optical disk, and various program code storage media.

[0183] Finally, it should be noted that those skilled in the art, after considering the specification and practicing the application disclosed herein, will easily think of other embodiments of the application. The application is intended to cover any variations, uses, or adaptations of the application that follow the general principles of the application and include common knowledge or conventional techniques in the art that are not disclosed by the application, and is not limited to the precise structure described above and shown in the drawings, and various modifications and changes can be made without departing from the scope thereof. The scope of the application is only limited by the appended claims.

Claims

1. A method of work order dispatching, the method comprising: The method includes: The dialogue information corresponding to the work order is input into the natural language processing model to obtain the business type and implicit features of the work order; Based on the business type and the implicit characteristics, the target business scenario of the work order is determined; Identify at least one candidate institution from among multiple institutions that matches the target business scenario; Based on the historical processing data of each candidate organization, the business type, and the implicit characteristics, the target dispatching organization corresponding to the work order is determined from the at least one candidate organization.

2. The method of claim 1, wherein, The implicit features include the complexity of the problem and the number of collaborating departments; determining the target business scenario of the work order based on the business type and the implicit features includes: When the complexity of the problem exceeds the preset level and there are multiple collaborating departments, the target business scenario is determined as the preset business scenario; If the complexity of the problem is less than or equal to the preset level, or if the number of collaborating departments is not multiple, the target business scenario is obtained by matching the business type with a preset scenario library.

3. The method of claim 1, wherein, The step of determining the target dispatching agency corresponding to the work order from the at least one candidate agency based on the historical processing data of each candidate agency, the business type, and the implicit characteristics includes: The allocation score for each candidate institution is determined based on its historical processing data, the business type, and the implicit characteristics. Based on the assignment score of each candidate organization, the target assignment organization corresponding to the work order is determined.

4. The method of claim 3, wherein, The implicit feature includes the degree of urgency; The step of determining the allocation score for each candidate institution based on its historical processing data, business type, and implicit characteristics includes: Based on the historical processing data of each candidate institution, determine the institutional evaluation indicators for each candidate institution; Based on the business type and the business type corresponding to each candidate institution, a business matching index is determined. Determine the load fit based on the current load of each candidate facility; Based on the urgency level, determine the work order priority index for each candidate organization; The assignment score for each candidate organization is determined based on the organization evaluation index, the business matching index, the load adaptability index, and the work order priority index.

5. The method of claim 4, wherein, The process of determining the assignment score for each candidate organization based on the organization evaluation indicators, the business matching indicators, the load adaptability indicators, and the work order priority indicators includes: The weight of each indicator is determined based on the user profile data corresponding to the work order; Based on the weight of each indicator, as well as the institution evaluation indicator, the business matching indicator, the load adaptability indicator, and the work order priority indicator, the assignment score for each candidate institution is determined.

6. The method of claim 4, wherein, The step of determining the target dispatching agency corresponding to the work order based on the dispatching score of each candidate agency includes: Verify whether the load adaptability of the first institution with the highest assignment score among the at least one candidate institution is less than a preset threshold. If the load adaptability of the first institution is less than the preset threshold, the second institution with the second highest assignment score among the at least one candidate institution shall be selected as the target assignment institution. If the load adaptability of the first institution is greater than or equal to the preset threshold, the first institution shall be selected as the target assignment institution.

7. A work order dispatching apparatus characterized by comprising: The device includes: The speech parsing module is used to input the dialogue information corresponding to the work order into the natural language processing model to obtain the business type and implicit features of the work order; The rules engine module is used to determine the target business scenario of the work order based on the business type and the implicit features. The first screening module is used to determine at least one candidate institution from multiple institutions that matches the target business scenario; The second screening module is used to determine the target dispatching agency corresponding to the work order from the at least one candidate agency based on the historical processing data of each candidate agency, the business type, and the implicit characteristics.

8. A computer device, comprising: include: A processor, and a memory communicatively connected to the processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory to implement the method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1 to 6.

10. A computer program product, characterised in that, Includes a computer program that, when executed by a processor, implements the method as described in any one of claims 1 to 6.

Images