Call center work order intelligent distribution system and method supporting multi-channel access

Abstract

The invention discloses a call center work order intelligent distribution system and method supporting multi-channel access, and belongs to the field of call center service systems.The system comprises a multi-channel access engine, an intelligent processing engine, a dynamic distribution engine, a load monitoring module and a knowledge base module; the method comprises the following steps: S1, receiving work order information from multiple channels through a multi-channel access engine; s2, identifying a channel of the obtained work order information through an intelligent processing engine, determining a channel type, carrying out format unification processing and key information extraction, and generating a standard work order; and S3, the dynamic distribution engine distributes the work order to the corresponding customer service staff according to a preset intelligent distribution strategy in combination with the real-time load condition and the historical processing efficiency of the customer service staff. The method has the advantages that multi-channel unified access and automatic information processing are realized, so that the efficiency and the accuracy are greatly improved; the strategy is dynamically and intelligently distributed, and accurate matching is achieved; and real-time load monitoring and knowledge base dynamic updating support strategy continuous optimization.

Description

Technical Field

[0001] The present invention relates to the field of call center service systems, and in particular to a call center work order intelligent dispatching system and method supporting multi-channel access. Background Art

[0002] With the rapid development of information technology, communication channels between businesses and customers are becoming increasingly diverse, with telephone, apps, WeChat, and email being widely used. However, the formats of work orders across these channels vary significantly. For example, telephone work orders are typically generated through voice-to-text conversion, with relatively colloquial content and the potential for a high level of irrelevant information. App work orders are structured forms (with pre-set fields such as issue type and urgency), with concise and clearly categorized information. WeChat work orders are mostly unstructured text (including emoticons and short sentences). Email work orders may include attachments (such as screenshots and documents). Existing call center work order dispatch systems are unable to support multi-channel access. Customer service teams must frequently switch between multiple platforms (such as the phone system backend, apps, WeChat management tools, and email clients) to process these work orders. They also need to manually re-enter non-standardized information into the unified work order dispatch system, which is time-consuming and labor-intensive. This not only significantly reduces work efficiency but also makes it prone to errors during the conversion process, resulting in inaccurate work order information or missing key content, impacting subsequent processing and service quality. Furthermore, most current call center ticket dispatching systems rely on fixed rules for dispatching. Common examples include "skill tag dispatch," which assigns tickets to agents based solely on their fixed skill tags, or "round-robin dispatch," which simply assigns tickets to agents in a specific order. These rigid dispatching strategies fail to fully consider the real-time workload of agents. For example, some agents may already have a large number of unprocessed tickets, yet new tickets are still assigned to them, leading to excessive workload and delayed processing. They also fail to consider historical processing efficiency, failing to prioritize tickets assigned to agents who handle similar tickets more quickly and accurately. Furthermore, they fail to consider the crucial factor of customer priority, failing to ensure that tickets from high-value or emergency customers receive priority processing. This results in inaccurate ticket dispatching and may even result in complex tickets being assigned to novice agents, impacting problem resolution and customer satisfaction.

[0003] In view of the above-mentioned shortcomings of the existing technology, the present invention aims to propose a call center work order intelligent dispatching system and method that can effectively integrate multi-channel work orders and adopt a flexible and intelligent dispatching strategy to improve the efficiency and accuracy of work order processing and enhance customer service quality. Summary of the Invention

[0004] In order to solve the problems of the prior art, the present invention provides a call center work order intelligent dispatching system and method supporting multi-channel access.

[0005] In order to solve the above technical problems, the present invention is achieved through the following technical solutions:

[0006] An intelligent dispatching system for call center work orders that supports multi-channel access, including a multi-channel access engine, an intelligent processing engine, a dynamic dispatching engine, a load monitoring module, and a knowledge base module;

[0007] The multi-channel access engine is used to receive work order information from multiple channels, including telephone, APP, WeChat and email;

[0008] The intelligent processing engine is connected to the multi-channel access engine to perform format standardization processing on the obtained work order information and extract key information;

[0009] The dynamic dispatch engine is connected to the intelligent processing engine, the load monitoring module, and the knowledge base module, and assigns work orders to corresponding customer service personnel based on the preset intelligent dispatch strategy, combined with the real-time customer service load and historical processing efficiency;

[0010] The load monitoring module is used to monitor the current number and status of work orders processed by each customer service staff and the number of unprocessed work orders in real time;

[0011] The knowledge base module is used to store and analyze the historical work order processing data of each customer service staff, determine their processing efficiency for different types of work orders, and provide reference for the subsequent processing of the same type of work orders;

[0012] The intelligent processing engine includes an explicit judgment module, a content feature extraction and analysis module, and a standard work order generation module.

[0013] Furthermore, the multi-channel access engine includes a telephone interface unit, an APP interface unit, a WeChat interface unit and an email interface unit, and each interface unit is adapted to the work order data transmission protocol and format of the corresponding channel to achieve stable reception of work order information.

[0014] Furthermore, the explicit judgment module searches the original data to see if there is an explicit channel tag. If there is an explicit channel tag, the tag value is directly used to output the channel type.

[0015] The content feature extraction and analysis module has a lower execution priority than the explicit judgment module. The content feature extraction and analysis module includes a structural feature extraction unit, a text feature analysis unit, a media type detection unit, and a machine learning classification unit. When the explicit judgment module retrieves raw data without an explicit channel tag, the content feature extraction and analysis module extracts and detects features from the raw data, and classifies the data through the machine learning classification unit, and outputs a channel type.

[0016] The standard work order generation module includes a telephone work order processing unit, an APP work order processing unit, a WeChat work order processing unit, and an email work order processing unit. After the content feature extraction and analysis module outputs the channel type, it enters the work order processing unit of the corresponding type. The work order processing unit of the corresponding type performs format unification processing and key information extraction to generate a standard work order.

[0017] Furthermore, the dynamic dispatch engine includes a policy calculation module, a multi-source data fusion module, and a dispatch execution module, which work together to complete the entire process from policy triggering to work order allocation, including:

[0018] The multi-source data fusion module is responsible for pulling and integrating three types of key data in real time:

[0019] (1) Standard work order information from the intelligent processing engine, including work order type, urgency, basic customer information, and problem description;

[0020] (2) Real-time customer service status data from the load monitoring module, including the number of work orders currently in transit, the estimated remaining processing time for each work order, the current load rate, and whether the customer is offline or resting;

[0021] (3) Customer service capability profile data from the knowledge base module, including the average time taken to handle similar work orders in the past, first response timeliness rate, problem resolution rate, customer satisfaction score, and skill tag matching degree;

[0022] The strategy calculation module executes a three-dimensional dynamic scoring algorithm based on the fused multi-source data. The specific strategies include:

[0023] (1) Urgency priority weighting: Calculate the comprehensive urgency value of the work order based on the urgency of the work order and the customer priority;

[0024] (2) Customer service load adaptability: The customer service load tolerance for the current work order is assessed using the formula "load adaptation score = 1 / (current load rate × 0.5 + remaining processing time overtime risk coefficient × 0.5)";

[0025] (3) Historical efficiency matching: extract the customer service’s historical resolution rate and average processing time for this type of work order from the knowledge base module, and combine it with the semantic label of the current work order to calculate the skill matching score;

[0026] (4) Dynamic strategy fusion: The scores of the above three dimensions are normalized and weighted to generate a comprehensive distribution score for each candidate customer service;

[0027] The dispatch execution module selects the customer service with the highest comprehensive score as the target allocation object according to the score ranking generated by the strategy calculation module; if there are multiple customer services with the same score, further fine-tuning is performed according to secondary indicators; finally, a dispatch instruction is sent to the customer service workstation through the API, and the "current load status" and "pending work order list" of the customer service in the knowledge base module are synchronously updated.

[0028] A method for intelligently dispatching work orders in a call center, comprising the following steps:

[0029] S1. New work order arrival: The multi-channel access engine receives work order information from multiple channels, including phone, app, WeChat, and email.

[0030] S2. Channel Identification: When a new work order arrives, the intelligent processing engine identifies the channel of the work order information and determines the channel type. Based on the channel type, the work order information is formatted and key information is extracted to generate a standard work order.

[0031] S3. The dynamic dispatch engine assigns work orders to corresponding customer service personnel based on preset intelligent dispatch strategies, combined with the customer service's real-time load and historical processing efficiency.

[0032] The advantages of the present invention compared with the prior art are:

[0033] 1. Multi-channel unified access and automated information processing significantly improve efficiency and accuracy: This invention integrates the four major channels of telephone, app, WeChat, and email through a multi-channel access engine. Each interface unit adapts to the transmission format of the corresponding protocol to achieve stable reception of work order information. Combined with the explicit judgment, content feature extraction, and standard work order generation functions of the intelligent processing engine, it automatically completes channel type identification, format unification, and key information extraction, eliminating the need for manual secondary entry, greatly improving work order processing efficiency and reducing information entry errors.

[0034] 2. Dynamic, intelligent dispatching strategies replace rigid, fixed rules for precise matching: This invention achieves precise dispatching through three core strategies of a dynamic dispatching engine: Urgency priority weighting combines the urgency of work orders with customer priority to ensure that high-urgency, high-value work orders are assigned first; Customer service load adaptation quantifies the customer service's ability to handle current work orders, preventing over-assignment of work orders to high-loaded customer service representatives; and Historical efficiency matching, based on the customer service representative's historical resolution rate, average time consumption, and skill tags for similar work orders in the knowledge base module, prioritizes work orders to more efficient customer service representatives, reducing the probability of complex work orders being assigned to novice customer service representatives and shortening the average work order processing time.

[0035] 3. Real-time load monitoring and dynamic updating of the knowledge base to support continuous strategy optimization: The present invention uses the load monitoring module to collect the customer service's current number of in-transit work orders, the estimated remaining processing time of each work order, the current load rate and online status in real time to ensure that dispatch decisions are based on the latest data; at the same time, the knowledge base module stores and analyzes customer service historical processing data, dynamically updates it, and continuously calibrates it after dispatch to support subsequent dispatch decisions. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 This is an architectural diagram of a call center work order intelligent dispatching system and method supporting multi-channel access according to the present invention.

[0037] Figure 2 This is a channel type judgment flow chart of a call center work order intelligent dispatching system and method supporting multi-channel access in the present invention.

[0038] Figure 3 This is a workflow diagram of a call center work order intelligent dispatching system and method supporting multi-channel access according to the present invention. DETAILED DESCRIPTION

[0039] In order to make the purpose, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments.

[0040] Example 1, combined with the attached Figure 1-3 , a call center work order intelligent dispatching system and method supporting multi-channel access.

[0041] An intelligent dispatching system for call center work orders that supports multi-channel access, including a multi-channel access engine, an intelligent processing engine, a dynamic dispatching engine, a load monitoring module, and a knowledge base module;

[0042] The multi-channel access engine is used to receive work order information from multiple channels, including phone, APP, WeChat, and email;

[0043] The intelligent processing engine is connected to the multi-channel access engine to standardize the format of the obtained work order information and extract key information;

[0044] The dynamic dispatch engine is connected to the intelligent processing engine, load monitoring module, and knowledge base module. It assigns work orders to the corresponding customer service personnel based on the preset intelligent dispatch strategy, combined with the customer service real-time load situation and historical processing efficiency.

[0045] The load monitoring module is used to monitor the current number and status of work orders processed by each customer service staff, as well as the number of unprocessed work orders in real time;

[0046] The knowledge base module is used to store and analyze the historical work order processing data of each customer service representative, determine their processing efficiency for different types of work orders, and provide reference for the subsequent processing of the same type of work orders;

[0047] The intelligent processing engine includes an explicit judgment module, a content feature extraction and analysis module, and a standard work order generation module.

[0048] The multi-channel access engine includes a telephone interface unit, an app interface unit, a WeChat interface unit, and an email interface unit. Each interface unit is adapted to the work order data transmission protocol and format of the corresponding channel to achieve stable reception of work order information, including:

[0049] Telephone Interface Unit: Integrates with CTI (Computer Telephony Integration) middleware (such as Cisco JTAPI / Avaya DMCC / Genesys T-Server) and monitors ACD (Automatic Call Distributor) events. CTI events are triggered when an incoming call is processed by the IVR (Interactive Voice Response). The unit captures the call's ANI (Calling Number), DNIS (Dialled Number), IVR input options, call timestamp, and possible initial skill group routing information. Format Unification Preprocessing: This captured information is encapsulated into an internally defined structured data format (such as JSON / Protocol Buffers), including the following fields: call_id, caller_id, ivr_input, timestamp, and initial_skill_group.

[0050] App Interface Unit: Receives JSON-formatted ticket request data from the app via HTTPS POST. Format standardization preprocessing: Verifies the API request signature / token, parses the JSON data (such as user ID, issue type selection, issue description text, screenshot / log attachment URLs, and device information), and converts it into an internal structured data format, including the following fields: app_user_id, issue_type, description, attachment_urls, device_info, and app_version.

[0051] WeChat Interface Unit: Implements the WeChat Official Account / Enterprise WeChat message receiving interface, receiving work order information (text, images, voice, location, and mini-program form data) submitted by users through the Official Account menu, messages, or mini-programs. Format Unification Preprocessing: Parses WeChat XML / JSON message bodies, extracts the user's OpenID / UnionID, message type, and content (text content, media URLs, location coordinates, and form field values), and converts them into an internal structured data format, including the following fields: wechat_id, msg_type, content, media_urls, location, and form_data.

[0052] The Mail Interface Unit listens to the mail server via IMAP / POP3 protocol polling or listening to the SMTP port to receive emails from clients. Format Unification Preprocessing: Parsing email headers (sender address, subject, receipt time) and email bodies (HTML / plain text); extracting the body content and attachments (saving them to object storage and recording the URLs); performing basic text cleansing (removing signatures and email history); and converting to an internal structured data format, including fields such as from_email, subject, received_time, body_text, and attachment_urls.

[0053] After all interface units complete preprocessing, they output an internal common data object containing raw channel data and preliminary structured metadata, which is then passed to the intelligent processing engine. This object contains a source_raw_data field (which stores the original message / message body) and a source_metadata field (which stores the structured information extracted from each channel).

[0054] After the data is passed to the intelligent processing engine, the explicit judgment module searches the original data to see if there is an explicit channel tag. If there is an explicit channel tag, the tag value is directly used to output the channel type.

[0055] The content feature extraction and analysis module has a lower execution priority than the explicit judgment module. The content feature extraction and analysis module includes a structural feature extraction unit, a text feature analysis unit, a media type detection unit, and a machine learning classification unit. When the explicit judgment module retrieves raw data without explicit channel tags, the content feature extraction and analysis module extracts and detects features from the raw data, and classifies it through the machine learning classification unit to output the channel type.

[0056] The standard work order generation module includes telephone work order processing unit, APP work order processing unit, WeChat work order processing unit, and email work order processing unit. After the content feature extraction and analysis module outputs the channel type, it enters the work order processing unit of the corresponding type. The work order processing unit of the corresponding type performs format unification and key information extraction to generate a standard work order.

[0057] in:

[0058] For explicit judgment modules:

[0059] Implementation: Maintain an "explicit channel tag" rule base, the rules can be:

[0060] A specific field has a specific value (such as source_type="mobile_app" in APP interface data).

[0061] Specific metadata features (e.g., initial_skill_group contains "Phone_Support" in phone interface data).

[0062] Original data header information (such as the email header X-Custom-Source: WeChat).

[0063] Workflow: Sequentially match the rules in the rule library. Once a match is successful, immediately use the channel type defined by the rule (such as "Phone", "App", "WeChat", "Email") as output and skip the content feature extraction and analysis module. If all rules are not matched, it is marked as "Unknown" and the content feature extraction and analysis module is triggered.

[0064] Content feature extraction and analysis module:

[0065] Trigger condition: When the explicit judgment module outputs "unknown" or the explicit judgment module is not enabled.

[0066] Unit collaborative workflow:

[0067] 1. Extract structural feature units:

[0068] Analyze the data structure: Is it a key-value pair (JSON / XML / form)? Is it free text (email body)? Does it contain nested levels?

[0069] Identify whether there is a specific structural pattern (for example, the JSON submitted by the APP may have fixed field names).

[0070] 2. Analyze text feature units:

[0071] NLP Processing:

[0072] Word segmentation, part-of-speech tagging, named entity recognition (NER): Identify company names, product names, person names, and places.

[0073] Keyword extraction: Identify keywords related to the channel (such as "in the mini program", "APP crash", "phone call cannot be connected", "email attachment").

[0074] 3. Detect media type unit:

[0075] Identify the media types included in the data package: images (screenshots, photos), audio (voice messages), videos, documents (PDF, Word), and log files.

[0076] Extract media metadata (size, format, resolution).

[0077] 4. Machine Learning Classification Unit:

[0078] Feature engineering: Combine all the features extracted by the above units (structural features, text feature word vectors / keyword lists, media type lists, metadata features) into a feature vector.

[0079] Model: Use a pre-trained multi-classification model (such as SVM, random forest, XGBoost, or lightweight neural network). The model is trained on historical data (data with manually labeled channel types).

[0080] Input: Feature vector.

[0081] Output: Predicted channel type ("Phone", "App", "WeChat", "Email", "Unknown") and confidence score. If the confidence score is lower than the preset threshold (such as 0.7), "Unknown" is output.

[0082] Generate a standard work order:

[0083] Input: raw / pre-processed data objects from the multi-channel access engine + channel identification results (from explicit judgment or content analysis).

[0084] Processing Unit (routing by channel type):

[0085] 1. Telephone work order processing unit:

[0086] Speech-to-text (ASR): If the call is recorded, call the ASR engine (such as Alibaba Cloud Intelligent Voice Interaction / iFlytek) to convert the recording into text (call_transcript).

[0087] Key information extraction: Combine IVR input options and transcribed text, and use rules / NER to extract: customer problem type (such as "bill inquiry" or "fault repair"), account information, contact number, address, and urgency (based on keywords such as "urgent" or "unavailable" or IVR selection).

[0088] Unified format: Generates a standard work order object containing the fields channel = Phone, call_id, caller_id, ivr_input, transcript, extracted_issues[], customer_info, and urgency_level.

[0089] 2. APP work order processing unit:

[0090] Parse structured data: directly map the issue_type, description, and app_user_id fields.

[0091] Process attachments: record the attachment URL and possibly call OCR to recognize text in the screenshot (such as error codes).

[0092] Key Information Extraction: Perform NER / keyword extraction on the description text to supplement or refine the question type. Extract the model and OS version from the device information.

[0093] Unified format: Generates a standard work order object containing the fields channel = App, app_user_id, issue_type, description, attachments, device_info, and extracted_details.

[0094] WeChat work order processing unit:

[0095] Processing message type:

[0096] Text: directly used as the problem description.

[0097] Image / Voice: Record URL, OCR the image, and convert voice to text (same as phone calls).

[0098] Location: Record coordinates and convert them into addresses (e.g. for door-to-door service).

[0099] Mini Program Form: Parse form field values.

[0100] Key information extraction: Integrate text, form data, OCR / ASR results, and use rules / NER to extract core questions, user ID (OpenID / UnionID), contact information (from the form or text extraction), and address.

[0101] Unified format: Generates a standard work order object containing the fields channel = WeChat, wechat_id, msg_type, content_summary, media_attachments, location, form_data, and extracted_info.

[0102] Email work order processing unit:

[0103] Text cleaning and extraction: remove email signatures and historical correspondence content (using heuristic rules or models).

[0104] Identify the core descriptive paragraphs of the text.

[0105] Use NER / rules to extract: sender name, contact number, problem description, product serial number, expected resolution time.

[0106] Process attachments: record attachment URLs and identify attachment types (such as log files, screenshots, and contract scans).

[0107] Unified format: Generates a standard work order object with the fields channel = Email, from_email, subject, cleaned_body, attachments, extracted_customer_details, extracted_issue_details.

[0108] Output: A highly structured standard work order object. This object must contain the following core fields (specific field names are only examples):

[0109] ticket_id (unique ticket ID, system generated)

[0110] channel (source channel)

[0111] create_time (ticket creation time)

[0112] customer_info (structure: {name, phone, email, customer_id (if known), vip_level, location}) - some information may be added later

[0113] issue_category (problem category, such as "technical failure", "billing problem", "service consultation")

[0114] issue_subcategory (issue subcategory, such as "APP crash", "data fee dispute")

[0115] description (detailed description text / summary of the problem)

[0116] urgency_level (urgency level, such as P0-P4, which may be based on rules or model predictions)

[0117] attachments (list of attachment URLs)

[0118] source_metadata (keep the original pre-processing metadata for traceability)

[0119] extracted_entities (a list of extracted key entities, such as product number, error code, order number).

[0120] The dynamic dispatch engine includes a policy calculation module, a multi-source data fusion module, and a dispatch execution module. These three modules work together to complete the entire process from policy triggering to work order assignment.

[0121] The multi-source data fusion module is responsible for pulling and integrating three types of key data in real time:

[0122] (1) Standard work order information from the intelligent processing engine, including work order type, urgency, basic customer information (such as VIP level, historical complaint records), and problem description;

[0123] (2) Real-time customer service status data from the load monitoring module, including the number of work orders currently in transit, the estimated remaining processing time for each work order (based on the average time spent on similar work orders in history), the current load rate (number of work orders in transit / maximum concurrent processing capacity of the customer service), and whether the customer service is in an offline / resting state;

[0124] (3) Customer service capability profile data from the knowledge base module, including the average time taken to handle similar work orders in history, first response timeliness, problem resolution rate, customer satisfaction score, and skill label matching (e.g., whether it is marked as "excellent at after-sales" or "proficient in technical troubleshooting");

[0125] The strategy calculation module executes a three-dimensional dynamic scoring algorithm based on the fused multi-source data. The specific strategies include:

[0126] (1) Urgency priority weighting: Calculate the comprehensive urgency value of a work order based on its urgency (e.g., the weight coefficient for a P0 work order is 1.5, and for a P4 work order is 0.8) and the customer priority (an additional 0.3 weight is added to VIP customer work orders);

[0127] (2) Customer service load adaptability: The customer service load tolerance for the current work order is assessed using the formula "load adaptation score = 1 / (current load rate × 0.5 + remaining processing time overtime risk coefficient × 0.5)" (where the overtime risk coefficient is dynamically calculated based on the ratio of the urgency of the work order to the average processing time of similar work orders in the customer service history. If the expected processing time exceeds the SLA threshold of the urgent work order, the coefficient is ≥ 1).

[0128] (3) Historical efficiency matching: Extract the customer service staff’s historical resolution rate for this type of ticket (e.g., 90% or more is a high match) and average processing time (20% better than the team average is a high efficiency) from the knowledge base module, and combine it with the semantic label of the current ticket (e.g., “system failure” corresponds to technical customer service requirements) to calculate the skill matching score;

[0129] (4) Dynamic strategy fusion: The scores of the above three dimensions are normalized and weighted and summed (the weights can be configured through the management backend, such as 40% for emergency priority, 30% for load adaptability, and 30% for skill matching) to generate a comprehensive distribution score for each candidate customer service representative;

[0130] The dispatch execution module sorts the scores generated by the strategy calculation module and selects the customer service representative with the highest comprehensive score as the target assignment object; if there are multiple customer service representatives with the same score, further fine-tuning is performed based on the secondary indicators of "current idle time" and "historical customer evaluation"; finally, a dispatch instruction (including work order details, customer contact information, and SLA time limit) is sent to the customer service workstation through the API, and the "current load status" and "list of pending work orders" of the customer service representative in the knowledge base module are simultaneously updated.

[0131] A method for intelligently dispatching work orders in a call center, comprising the following steps:

[0132] S1. New work order arrival: The multi-channel access engine receives work order information from multiple channels, including phone, app, WeChat, and email.

[0133] S2. Channel Identification: When a new work order arrives, the intelligent processing engine identifies the channel of the work order information and determines the channel type. Based on the channel type, the work order information is formatted and key information is extracted to generate a standard work order.

[0134] S3. The dynamic dispatch engine assigns work orders to corresponding customer service personnel based on preset intelligent dispatch strategies, combined with the customer service's real-time load and historical processing efficiency.

[0135] The present invention and its implementation methods are described above. This description is not restrictive. If ordinary technicians in this field are inspired by it and design embodiments similar to the technical solution without creatively designing them without departing from the purpose of the invention, they should all fall within the scope of protection of the present invention.

Claims

1. A call center work order intelligent dispatching system supporting multi-channel access, characterized by: Includes multi-channel access engine, intelligent processing engine, dynamic dispatch engine, load monitoring module, and knowledge base module; The multi-channel access engine is used to receive work order information from multiple channels, including telephone, APP, WeChat and email; The intelligent processing engine is connected to the multi-channel access engine to perform format standardization processing on the obtained work order information and extract key information; The dynamic dispatch engine is connected to the intelligent processing engine, the load monitoring module, and the knowledge base module, and assigns work orders to corresponding customer service personnel based on the preset intelligent dispatch strategy, combined with the real-time customer service load and historical processing efficiency; The load monitoring module is used to monitor the current number and status of work orders processed by each customer service staff and the number of unprocessed work orders in real time; The knowledge base module is used to store and analyze the historical work order processing data of each customer service staff, determine their processing efficiency for different types of work orders, and provide reference for the subsequent processing of the same type of work orders; The intelligent processing engine includes an explicit judgment module, a content feature extraction and analysis module, and a standard work order generation module.

2. The call center work order intelligent dispatching system supporting multi-channel access according to claim 1, characterized in that: The multi-channel access engine includes a telephone interface unit, an APP interface unit, a WeChat interface unit and an email interface unit. Each interface unit is adapted to the work order data transmission protocol and format of the corresponding channel to achieve stable reception of work order information.

3. The call center work order intelligent dispatching system supporting multi-channel access according to claim 1, characterized in that: The explicit judgment module searches the original data to see if there is an explicit channel tag. If there is an explicit channel tag, the tag value is directly used to output the channel type. The content feature extraction and analysis module has a lower execution priority than the explicit judgment module. The content feature extraction and analysis module includes a structural feature extraction unit, a text feature analysis unit, a media type detection unit, and a machine learning classification unit. When the explicit judgment module retrieves raw data without an explicit channel tag, the content feature extraction and analysis module extracts and detects features from the raw data, and classifies the data through the machine learning classification unit, and outputs a channel type. The standard work order generation module includes a telephone work order processing unit, an APP work order processing unit, a WeChat work order processing unit, and an email work order processing unit. After the content feature extraction and analysis module outputs the channel type, it enters the work order processing unit of the corresponding type. The work order processing unit of the corresponding type performs format unification processing and key information extraction to generate a standard work order.

4. The call center work order intelligent dispatching system and method supporting multi-channel access according to claim 1, characterized in that: The dynamic dispatch engine includes a policy calculation module, a multi-source data fusion module, and a dispatch execution module. The three modules work together to complete the entire process from policy triggering to work order allocation, including: The multi-source data fusion module is responsible for pulling and integrating three types of key data in real time: (1) Standard work order information from the intelligent processing engine, including work order type, urgency, basic customer information, and problem description; (2) Real-time customer service status data from the load monitoring module, including the number of work orders currently in transit, the estimated remaining processing time for each work order, the current load rate, and whether the customer is offline or resting; (3) Customer service capability profile data from the knowledge base module, including the average time taken to handle similar work orders in the past, first response timeliness rate, problem resolution rate, customer satisfaction score, and skill tag matching degree; The strategy calculation module executes a three-dimensional dynamic scoring algorithm based on the fused multi-source data. The specific strategies include: (1) Urgency priority weighting: Calculate the comprehensive urgency value of the work order based on the urgency of the work order and the customer priority; (2) Customer service load adaptability: The customer service load tolerance for the current work order is assessed using the formula "load adaptation score = 1 / (current load rate × 0.5 + remaining processing time overtime risk coefficient × 0.5)"; (3) Historical efficiency matching: extract the customer service’s historical resolution rate and average processing time for this type of work order from the knowledge base module, and combine it with the semantic label of the current work order to calculate the skill matching score; (4) Dynamic strategy fusion: The scores of the above three dimensions are normalized and weighted to generate a comprehensive distribution score for each candidate customer service; The dispatch execution module selects the customer service representative with the highest comprehensive score as the target assignment object based on the score ranking generated by the strategy calculation module; if there are multiple customer service representatives with the same score, further fine-tuning is performed based on secondary indicators; finally, a dispatch instruction is sent to the customer service workstation via the API, and the "current load status" and "pending work order list" of the customer service representative in the knowledge base module are simultaneously updated.

5. A method for intelligently dispatching work orders in a call center, the method being based on a system for intelligently dispatching work orders in a call center supporting multi-channel access as claimed in any one of claims 1 to 4, characterized in that The following steps are involved: S1. New work order arrival: The multi-channel access engine receives work order information from multiple channels, including phone, app, WeChat, and email. S2. Channel Identification: When a new work order arrives, the intelligent processing engine identifies the channel of the work order information and determines the channel type. Based on the channel type, the work order information is formatted and key information is extracted to generate a standard work order. S3. The dynamic dispatch engine assigns work orders to corresponding customer service personnel based on preset intelligent dispatch strategies, combined with the customer service's real-time load and historical processing efficiency.

Images