Customer service robot system integrating knowledge base question and answer retrieval and work order processing

Abstract

The invention discloses a customer service robot system integrating knowledge base question and answer retrieval and work order processing, which comprises a natural language understanding and generating module, a knowledge retrieval module, a business system integration module, a dialogue management and decision module and a cue word management module, and relates to the technical field of artificial intelligence. Through cooperation of a large language model and a dialogue management module, the system realizes multiple rounds of deep understanding of a natural language, and can actively guide a user to clarify a demand, such as gradually inquiring an equipment model and fault details during repair, memorizing a dialogue context and avoiding information omission. Compared with traditional regular customer service, the method has the advantages that the accuracy of understanding complex problems is remarkably improved, open dialogue scenes such as technical consultation and fault diagnosis are supported, user interaction experience is closer to manual service seats, and the problems that multi-round dialogue understanding is limited, progress data cannot be obtained in real time, multiple knowledge bases are fused, and the automation degree of the business process is low are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of artificial intelligence, and specifically to a customer service robot system that integrates knowledge base retrieval Q&A and work order processing. Background Art

[0002] With the development of natural language processing technology, how to enable a computer to smoothly understand human language and conduct conversations has always been an important topic in the field of artificial intelligence. Most of the existing customer service robots on the market are based on keyword matching or simple intent recognition, and can only handle fixed-format inquiries. They often have poor understanding of informal spoken language or complex questions and are difficult to conduct continuous multi-round conversations. This limitation causes the robot to easily give inappropriate answers when the user's question changes slightly, making it difficult to meet actual customer service needs.

[0003] In traditional customer service scenarios, voice call centers usually adopt an IVR system with a preset process. Users need to press buttons according to prompts to make selections, lacking flexible natural language communication and unable to handle complex problems. On the other hand, online chat customer service can provide a certain degree of automatic Q&A, but often relies on preset FAQ scripts or simple retrieval, has limited ability to handle non-standard questions, and lacks in-depth understanding of the context. In addition, these systems are usually not integrated with the enterprise's internal business systems: for example, when a user asks about the progress of work order processing, traditional robots cannot directly obtain data from the background work order system, and the user still needs to query manually; when a user submits a repair request, the robot cannot automatically generate a work order and requires manual intervention. This fragmentation limits the role of automatic customer service in real business processing. Under the traditional solution, if new knowledge needs to be extended or the answering style of the robot needs to be adjusted, it usually requires retraining or rule configuration, lacking a fast-configurable optimization mechanism. Summary of the Invention

[0004] In view of the deficiencies of the prior art, the present invention provides a customer service robot system that integrates knowledge base retrieval Q&A and work order processing, and solves the problems of limitations in multi-round conversation understanding, inability to obtain progress data in real time, integration of multiple knowledge bases, and low degree of business process automation.

[0005] To achieve the above objectives, the present invention is realized through the following technical solutions: A customer service robot system that integrates knowledge base retrieval Q&A and work order processing, including: a multi-channel access module, a natural language understanding and generation module, a knowledge retrieval module, a business system integration module, a dialogue management and decision module, and a prompt word management module, where: The natural language understanding and generation module is built based on a pre-trained large language model, and is used to parse the intent in the user's natural language text, maintain the conversation history to support multi-round interaction, and generate a natural language reply containing professional answers or work order information in combination with external knowledge; The knowledge retrieval module communicates with the enterprise knowledge base, which stores semantic vectors of PDF, DOCX documents, video transcription texts, and Q&A comparison data. This module generates query vectors based on the user's intent through a vector database, retrieves document fragments with similar semantics, and outputs them to the natural language understanding and generation module; The business system integration module communicates with the work order management system through an API interface, and is used to automatically create work orders, query work order status and handler information according to the user's intent, and feedback real-time business data to the natural language understanding and generation module; Based on the dialogue context state and the user's intent, the dialogue management and decision-making module dynamically invokes the knowledge retrieval module, the business system integration module, or triggers an artificial transfer process, and maintains the dialogue logic through a state machine, such as guiding the repair process and jumping to knowledge Q&A.

[0006] Preferably, the natural language understanding and generation module adopts the Retrieval-Augmented Generation (RAG) technology, which concatenates the document fragments returned by the knowledge retrieval module with the model input to generate responses that integrate the enterprise's exclusive knowledge.

[0007] Preferably, the business system integration module supports the following functions: automatically filling in work order fields according to the user's repair information, including user identity, device model, fault description, marking the work order emergency level based on emergency keywords in the user input, such as "on fire" and "water leakage", and pushing work order progress change notifications to the user, including work order number, handler contact information, and estimated completion time.

[0008] Preferably, the prompt word management module provides a backend management interface for configuring the prompt templates of the large language model to adjust the robot's response style, inject domain-specific knowledge, or specify the priority response strategy for specific questions without retraining the model.

[0009] Preferably, the dialogue management and decision-making module receives information from the external multi-channel access module. The multi-channel access module supports embedding the system into web pages, APPs, mini-programs, WeChat official accounts, and call center phone lines through standard SDKs and APIs, and realizes the unified processing of voice streams and text messages through a voice gateway, ensuring a consistent cross-platform interaction experience.

[0010] Preferably, the dialogue management and decision-making module includes a memory cache. When identifying the user's emergency repair intent, it automatically raises the priority of the work order, triggers the emergency notification process of the work order system, generates an artificial transfer request, and at the same time pushes the dialogue history, such as fault description and collected information, to the artificial seat interface.

[0011] Preferably, the natural language understanding and generation module has a knowledge precipitation function. It counts high-frequency questions based on conversation logs, and automatically integrates and generates standardized FAQ content in combination with knowledge base documents, which is released to external channels such as the enterprise official website and WeChat official account after manual review.

[0012] The present invention provides a customer service robot system that integrates knowledge base retrieval Q&A and work order processing. It has the following beneficial effects: 1. Through the collaboration of the large language model and the dialogue management module, the system realizes multi-round in-depth understanding of natural language, can actively guide users to clarify their needs. For example, when reporting a repair, it gradually asks about the device model and fault details, and remembers the conversation context to avoid information omission. Compared with traditional rule-based customer service, the accuracy of complex problem understanding is significantly improved, supporting open dialogue scenarios such as technical consultation and fault diagnosis, and the user interaction experience is closer to that of a human operator.

[0013] 2. The system can automatically generate work orders according to user intentions and fill in key information, synchronously connect to the work order system to query the status in real time, reducing the manual intervention link. In emergency scenarios, such as identifying keywords like "water leakage" and "power outage", it automatically marks the priority of the work order and triggers the notification mechanism, realizing a closed-loop from "dialogue consultation" to "transaction processing". Through the Retrieval-Augmented Generation (RAG) technology, it retrieves enterprise knowledge base documents in real time and integrates professional knowledge into the reply to avoid "hallucination" answers. At the same time, the system automatically precipitates high-frequency questions to generate FAQs, which are released to multiple channels after review, forming a self-circulation of knowledge utilization and update.

[0014] 3. Adopting a modular architecture design, it can be quickly embedded into web pages, APPs, WeChat official accounts, and call center phone lines through standard SDK / APIs, supporting unified processing of voice and text interactions, ensuring a consistent cross-platform experience, and reducing the enterprise's technology docking cost. It provides a prompt word management background, and operation personnel can independently configure the reply style, inject domain knowledge, or adjust emergency words without retraining the model, further shortening the cycle of responding to business requirement changes. At the same time, it supports cloud or local deployment to meet the data security and compliance requirements of different enterprises. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 is a flowchart of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0016] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0017] Please refer to Figure 1 , the present invention provides a technical solution: a customer service robot system integrating knowledge base retrieval Q&A and work order processing, including: a natural language understanding and generation module, a knowledge retrieval module, a business system integration module, a dialogue management and decision-making module, and a prompt word management module, wherein: The natural language understanding and generation module is built based on a pre-trained large language model, used to parse the intent in the user's natural language text, maintain the dialogue history to support multi-round interaction, and generate a natural language reply containing professional answers or work order information in combination with external knowledge; The knowledge retrieval module communicates with the enterprise knowledge base. The knowledge base stores semantic vectors of PDF, DOCX documents, video transcription texts, and Q&A comparison data. This module generates query vectors according to the user's intent through a vector database, retrieves semantically similar document fragments, and outputs them to the natural language understanding and generation module; The business system integration module communicates with the work order management system through an API interface, used to automatically create work orders, query work order status and handler information according to the user's intent, and feedback real-time business data to the natural language understanding and generation module; The dialogue management and decision-making module dynamically calls the knowledge retrieval module, the business system integration module or triggers the manual transfer process based on the dialogue context state and the user's intent, and maintains the dialogue logic through a state machine.

[0018] The natural language understanding and generation module adopts retrieval-augmented generation technology, splices the document fragments returned by the knowledge retrieval module with the model input, and generates a reply integrating the enterprise's exclusive knowledge.

[0019] The business system integration module supports the following functions: automatically filling work order fields according to the user's repair information, marking the work order emergency level based on the emergency keywords in the user's input, and pushing work order progress change notifications to the user.

[0020] The prompt word management module provides a backend management interface for configuring the prompt templates of the large language model to adjust the robot's reply style, inject domain-specific knowledge, or specify the priority response strategy for specific questions, without retraining the model.

[0021] The multi-channel access module supports embedding the system into web pages, APPs, mini-programs, WeChat official accounts, and call center phone lines through standard SDKs and APIs, realizes the unified processing of voice streams and text messages through a voice gateway, and ensures a consistent cross-platform interaction experience.

[0022] The dialogue management and decision-making module contains a memory cache. When it recognizes the user's emergency repair intent, it automatically raises the work order priority and triggers the emergency notification process of the work order system, generates a manual transfer request, and at the same time pushes the dialogue history to the manual seat interface.

[0023] The natural language understanding and generation module has a knowledge precipitation function. It statistically analyzes high-frequency questions based on conversation logs, and automatically integrates and generates standardized FAQ content in combination with knowledge base documents. After manual review, it is published to external channels. The natural language understanding and generation module NLU&NLG based on large language models is used to parse user intents and generate multi-turn conversation responses, and supports context memory, such as recording the user's device model and fault phenomenon. The knowledge retrieval module is used to connect to the enterprise knowledge base, store document semantic vectors through a vector database, and adopt the retrieval-augmented generation (RAG) technology to retrieve relevant knowledge fragments in real time and feedback them to the NLU&NLG module.

[0024] The business system integration module communicates with the work order management system through API interfaces to achieve automatic creation of work orders, status query, emergency level marking, and feedback of handler information. The conversation management and decision-making module is used to maintain the conversation state machine and dynamically schedule knowledge retrieval, work order processing, or manual transfer processes. For example, when a user reports a repair, it guides the collection of key information, such as device model and fault duration, and determines whether to trigger an emergency response.

[0025] Its core functions include multi-turn conversation guidance. When the user describes that "the air conditioner is not cooling", the robot actively asks "Is it completely not cold or is the cooling effect decreasing?" to gradually collect repair details.

[0026] When the user asks about "the water heater error code E05", the system retrieves the relevant paragraphs in the "Water Heater Fault Manual" in the knowledge base and generates a response containing troubleshooting steps. According to the confirmed repair information provided by the user, it automatically fills in the work order fields and generates a work order number, which is pushed to the user synchronously.

[0027] When the keyword "leakage" is recognized, it is regarded as an emergency situation. The work order is immediately marked as high priority, triggering a text message to notify the person in charge and transferring to the manual customer service, while transmitting the conversation history to the manual agent.

[0028] The system regularly counts high-frequency questions, automatically generates FAQ content in combination with the knowledge base, and after manual review, publishes it to the official website help center to reduce repeated consultations.

[0029] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprising", "including" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or further includes elements inherent to such process, method, article or device. Without further limitation, an element defined by the statement "comprising an..." does not exclude the presence of additional identical elements in the process, method, article or device comprising said element.

[0030] Although the embodiments of the present invention have been shown and described, it will be understood by those of ordinary skill in the art that various changes, modifications, substitutions and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A customer service robot system that integrates knowledge base retrieval for question answering and work order processing, characterized in that, It includes: A multi-channel access module, a natural language understanding and generation module, a knowledge retrieval module, a business system integration module, a dialogue management and decision-making module, and a prompt management module. Among them: The natural language understanding and generation module is built based on a pre-trained large language model, which is used to parse the intention in the user's natural language text, maintain the dialogue history to support multi-round interaction, and generate natural language responses containing professional answers or work order information in combination with external knowledge; The knowledge retrieval module communicates with the enterprise knowledge base, and the knowledge base stores the semantic vectors of PDF, DOCX documents, video transcription texts, and Q&A comparison data. This module generates query vectors according to the user's intention through a vector database, retrieves semantically similar document fragments, and outputs them to the natural language understanding and generation module; The business system integration module communicates with the work order management system through an API interface, which is used to automatically create work orders, query work order status and handler information according to the user's intention, and feedback real-time business data to the natural language understanding and generation module; The dialogue management and decision-making module dynamically calls the knowledge retrieval module, the business system integration module or triggers the manual transfer process based on the dialogue context state and the user's intention, and maintains the dialogue logic through a state machine.

2. The customer service robot system integrating knowledge base retrieval Q&A and work order processing according to claim 1, characterized in that: The natural language understanding and generation module adopts retrieval-augmented generation technology, splices the document fragments returned by the knowledge retrieval module with the model input, and generates responses integrating the enterprise's exclusive knowledge.

3. A customer service robot system integrating knowledge base retrieval Q&A and work order processing according to claim 1, characterized in that: The business system integration module supports the following functions: automatically filling work order fields according to the user's repair information, marking the work order emergency level based on the emergency keywords in the user's input, and pushing work order progress change notifications to the user.

4. The customer service robot system integrating knowledge base retrieval Q&A and work order processing according to claim 1, characterized in that: The prompt management module provides a backend management interface for configuring the prompt templates of the large language model.

5. The customer service robot system integrating knowledge base retrieval Q&A and work order processing according to claim 1, characterized in that: The dialogue management and decision-making module receives information from the external multi-channel access module. The multi-channel access module supports embedding the system into web pages, APPs, mini-programs, WeChat official accounts, and call center phone lines through standard SDKs and APIs, realizes the unified processing of voice streams and text messages through a voice gateway, and ensures a consistent cross-platform interaction experience.

6. The customer service robot system integrating knowledge base retrieval Q&A and work order processing according to claim 1, characterized in that: The dialogue management and decision-making module contains a memory cache. When it recognizes the user's intention of urgent repair, it automatically raises the priority of the work order and triggers the emergency notification process of the work order system, generates a manual transfer request, and at the same time pushes the dialogue history to the manual agent interface.

7. A customer service robot system integrating knowledge base retrieval Q&A and work order processing, characterized in that: The natural language understanding and generation module has a knowledge precipitation function, counts high-frequency questions based on dialogue logs, and automatically integrates and generates standardized FAQ content in combination with the knowledge base documents, which is published to external channels after manual review.

Images