Customer service task response methods, devices, equipment, and media

Abstract

The present application relates to a customer service task response method and its device, equipment, and medium in the field of e-commerce technology, the method comprising: obtaining a user question text provided by a user in a conversation with a customer service representative; when a consultation intention text matching the user question text is determined from a preset conversation knowledge base, and there is a preset task flow corresponding to the consultation intention text, the task flow is driven by the user question text, so that during the execution of the task flow, multiple rounds of corresponding conversations are conducted with the user to obtain all target keywords required for the task goal; the corresponding target response text is determined based on all target keywords, pushed to the user, and the task flow is terminated. The present application can clarify complex user consultation intentions and make correct responses.

Description

Technical Field

[0001] The present application relates to the field of e-commerce technology, and in particular to a customer service task response method and its corresponding device, computer equipment, and computer-readable storage medium. Background Art

[0002] E-commerce customer service is a key component in the e-commerce ecosystem. It not only improves customers' shopping experience and ensures that user problems and concerns are resolved in a timely manner, but also plays a vital role in maintaining brand image, maintaining customer satisfaction and loyalty, and promoting sales conversions.

[0003] In traditional technology, various questions and their answers are usually pre-configured in the customer service system and provided to the customer service robot so that it can automatically find the corresponding answer to the user based on the user's question when the user asks. However, sometimes the dialogue method in which the user provides a question and the customer service provides an answer cannot correctly solve the user's problem, because the user may not be clear about his or her own consulting intention when making the inquiry, which results in the answer found being unable to accurately answer the user's actual consulting intention.

[0004] In view of the shortcomings of traditional practices, the applicant has been engaged in research in related fields for a long time and has taken a different approach to solve industry problems in the field of e-commerce technology. Summary of the invention

[0005] The primary purpose of the present application is to solve at least one of the above problems and to provide a customer service task response method and its corresponding device, computer equipment, and computer program product.

[0006] In order to meet the various objectives of this application, this application adopts the following technical solutions:

[0007] A customer service task response method provided to meet one of the purposes of this application includes the following steps:

[0008] Obtain the text of user questions provided by the user in the conversation with customer service;

[0009] When a consultation intention text matching the user's question text is determined from a preset dialogue knowledge base, and there is a preset task flow corresponding to the consultation intention text, the task flow is driven by the user's question text, so that during the execution of the task flow, multiple rounds of corresponding dialogues are conducted with the user to obtain all target keywords required for the task goal;

[0010] The corresponding target response text is determined based on all target keywords, pushed to the user, and the task process is ended.

[0011] On the other hand, a customer service task response device is provided to meet one of the purposes of the present application, including a text acquisition module, a keyword determination module and a user response module, wherein the text acquisition module is used to acquire a user question text provided by a user in a conversation with a customer service representative; the keyword determination module is used to drive the task process with the user question text when a consultation intention text matching the user question text is determined from a preset conversation knowledge base and there is a preset task process corresponding to the consultation intention text, so that during the execution of the task process, corresponding multiple rounds of conversations are conducted with the user to obtain all target keywords required for the task goal; the user response module is used to determine the corresponding target response text based on all target keywords, push it to the user, and end the task process.

[0012] On the other hand, a computer device provided to meet one of the purposes of the present application includes a central processing unit and a memory, wherein the central processing unit is used to call and run a computer program stored in the memory to execute the steps of the customer service task response method described in the present application.

[0013] On the other hand, a computer program product provided to meet another purpose of the present application includes a computer program / instruction, which, when executed by a processor, implements the steps of the method described in any embodiment of the present application.

[0014] The technical solution of this application has many advantages, including but not limited to the following aspects:

[0015] This application first obtains the user question text in the conversation between the user and the customer service, and then determines the user's consulting intention that matches the user's question text from the preset conversation knowledge base, so that the user's real needs can be understood and identified. After the user's consulting intention is determined, if there is a preset task process corresponding to the intention, it means that the consulting intention is not clear enough. In this regard, the preliminary consulting intention will drive this task process, and the user will have corresponding multiple rounds of dialogue, so as to gradually understand the user's real consulting intention, and correspondingly collect and correctly answer all the target keywords required for the task goal of the real consulting intention. Thus, the workload of manual customer service is reduced, and fast and accurate customer service is provided at the same time. Through targeted multiple rounds of dialogue, the user can be gradually guided to provide additional information corresponding to the preliminary consulting intention, ensuring that the collected information is comprehensive and accurate, and the real consulting intention can be located. In addition, the response speed can be improved, the possible communication delays and errors of manual customer service can be avoided, and users can be ensured to get feedback in time.

[0016] Finally, all target keywords required for the task goal are collected, and the corresponding target response text is determined based on this, and pushed to the user, so as to achieve the task goal and end the task process. This not only improves the service efficiency of customer service, but also improves the customer service experience of users, and can provide more personalized and accurate services. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0018] Figure 1 The network architecture of the e-commerce platform exemplified in this application;

[0019] Figure 2 A flowchart of a typical embodiment of the customer service task response method of the present application;

[0020] Figure 3 This is a principle block diagram of the customer service task response device of this application;

[0021] Figure 4 A schematic diagram of the structure of a computer device used in this application. DETAILED DESCRIPTION

[0022] The embodiments of the present application are described in detail below, and examples of the embodiments are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present application, and cannot be interpreted as limiting the present application.

[0023] It will be understood by those skilled in the art that, unless expressly stated, the singular forms "one", "said", and "the" used herein may also include plural forms. It should be further understood that the term "comprising" used in the specification of the present application refers to the presence of the features, integers, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or groups thereof. It should be understood that when we refer to an element as being "connected" or "coupled" to another element, it may be directly connected or coupled to the other element, or there may be an intermediate element. In addition, the "connection" or "coupling" used herein may include wireless connection or wireless coupling. The term "and / or" used herein includes all or any unit and all combinations of one or more associated listed items.

[0024] It will be understood by those skilled in the art that, unless otherwise defined, all terms (including technical and scientific terms) used herein have the same meaning as those generally understood by those skilled in the art to which this application belongs. It should also be understood that terms such as those defined in general dictionaries should be understood to have meanings consistent with those in the context of the prior art, and will not be interpreted with idealized or overly formal meanings unless specifically defined as here.

[0025] like Figure 1 In the network architecture shown, the e-commerce platform 82 is deployed on the Internet to provide corresponding services to its users. Similarly, the devices 80 of the merchant users of the e-commerce platform 82 and the devices 81 of the consumer users are also connected to the Internet to use the services provided by the e-commerce platform.

[0026] The exemplary e-commerce platform 82 provides supply and demand matching of products and / or services to the general public with the help of Internet infrastructure. In the e-commerce platform 82, products and / or services are provided as commodity information. To simplify the description, the concepts of commodity, product, etc. are used in this application to refer to the products and / or services in the e-commerce platform 82, which may specifically be physical products, digital products, tickets, service subscriptions, other offline services, etc.

[0027] In reality, all entities can access the e-commerce platform 82 as users, use various online services provided by the e-commerce platform 82, and achieve the purpose of participating in the business activities achieved by the e-commerce platform 82. These entities can be natural persons, legal persons or social organizations, etc. Corresponding to the two types of entities, merchants and consumers in business activities, the e-commerce platform 82 has two types of users, merchant users and consumer users. All entities in the product distribution chain in business activities, including manufacturers, sellers, retailers, logistics providers, etc., can use online services in the e-commerce platform 82 as merchant users, while consumers in business activities, including real or potential consumers, can use online services in the e-commerce platform 82 as their corresponding consumer users. In actual business activities, the same entity can act as both a merchant user and a consumer user, and this should be understood flexibly.

[0028] The infrastructure used to deploy the e-commerce platform 82 mainly includes the backend architecture and frontend equipment. The backend architecture runs various online services through the service cluster, including middleware or frontend services for the platform, services for consumers, services for merchants, etc., to enrich and improve its service functions; the frontend equipment mainly covers the terminal devices used by users to access the e-commerce platform 82 as clients, including but not limited to various mobile terminals, personal computers, point-of-sale devices, etc. For example, merchant users can enter product information for their online stores through their terminal devices 80, or generate their product information using the interfaces opened by the e-commerce platform; consumer users can access the webpage of the online store implemented by the e-commerce platform 82 through their terminal devices 81, trigger the shopping process through the shopping buttons provided on the webpage, and call various online services provided by the e-commerce platform 82 during the shopping process, so as to achieve the purpose of shopping orders.

[0029] In some embodiments, the e-commerce platform 82 may be implemented by a processing facility including a processor and a memory, the processing facility storing a set of instructions that, when executed, cause the e-commerce platform 82 to perform the e-commerce and support functions involved in the present application. The processing facility may be part of a server, client, network infrastructure, mobile computing platform, cloud computing platform, fixed computing platform, or other computing platform, and provides electronic components, merchant equipment, payment gateways, application developers, marketing channels, transportation providers, customer equipment, point-of-sale equipment, etc. of the e-commerce platform 82.

[0030] The e-commerce platform 82 can be implemented as online services such as cloud computing services, software as a service (SaaS), infrastructure as a service (IaaS), platform as a service (PaaS), desktop as a service (DaaS), hosted software as a service, mobile backend as a service (MBaaS), information technology management as a service (ITMaaS), etc. In some embodiments, the various functional components of the e-commerce platform 82 can be implemented to be suitable for operation on various platforms and operating systems. For example, for an online store, its administrator users enjoy the same or similar functions regardless of various embodiments such as iOS, Android, HomonyOS, or web pages.

[0031] The e-commerce platform 82 can realize the corresponding independent station for each merchant to run its corresponding online store, and provide the merchant with the corresponding business management engine instance for the merchant to establish, maintain and run one or more online stores in one or more independent stations. The business management engine instance can be used for content management, task automation and data management of one or more online stores, and various specific business processes of the online store can be configured through interfaces or built-in components to support the implementation of business activities. The independent station is the infrastructure of the e-commerce platform 82 with cross-border service functions. Merchants can maintain their online stores more centrally and autonomously based on the independent station. The independent station usually has a domain name and storage space dedicated to the merchant, and different independent stations are relatively independent. The e-commerce platform 82 can provide standardized or personalized technical support for a large number of independent stations, so that merchant users can customize their own business management engine instance and use this business management engine instance to maintain one or more online stores they own.

[0032] The online store can implement backend configuration and maintenance by having the merchant user log in to its business management engine instance as an administrator. With the support of various online services provided by the infrastructure of the e-commerce platform 82, the merchant user can configure various functions in its online store as an administrator, view various data, etc. For example, the merchant user can manage various aspects of its online store, such as viewing the online store's recent activities, updating the online store's product catalog, managing orders, recent visit activities, total order activities, etc.; the merchant user can also view more detailed information about the business and visitors to the merchant's online store by obtaining reports or metrics, such as showing a sales summary of the merchant's overall business, specific sales and participation data of active sales marketing channels, etc.

[0033] The e-commerce platform 82 may provide communication facilities and associated merchant interfaces for providing electronic communications and marketing, such as utilizing electronic message aggregation facilities to collect and analyze communication interactions between merchants, consumers, merchant devices, customer devices, point-of-sale devices, etc., aggregating and analyzing communications, such as for increasing the potential for providing product sales, etc. For example, a consumer may have a question related to a product, which may generate a dialogue between the consumer and the merchant (or an automated processor-based agent on behalf of the merchant), where the communication facility is responsible for interacting and providing the merchant with analysis on how to increase the probability of a sale.

[0034] In some embodiments, an application suitable for installation in a terminal device can be provided to serve the access needs of different users, so that various users can access the e-commerce platform 82 by running the application in the terminal device, such as the merchant backend module of the online store in the e-commerce platform 82, etc. In the process of implementing business activities through these functions, the e-commerce platform 82 can implement various functions related to supporting the implementation of business activities as middleware or online services and open corresponding interfaces, and then implant the toolkit corresponding to the interface access function into the application to implement function expansion and task implementation. The business management engine can include a series of basic functions, and expose these functions to online services and / or application calls through APIs. Online services and applications use corresponding functions by remotely calling corresponding APIs.

[0035] With the support of various components of the business management engine instance, the e-commerce platform 82 can provide online shopping functions, enabling merchants to establish connections with customers in a flexible and transparent manner. Consumer users can select items online, create product orders, provide the delivery address of the goods in the product order, and complete the payment confirmation of the product order. Then, the merchant can review and complete or cancel the order. The review component carried by the business management engine instance can realize the compliance use of business processes to ensure that the order is suitable for fulfillment before actual fulfillment. Orders may sometimes be fraudulent and need to be verified (such as ID card checks). There is a payment method that requires merchants to wait to ensure that funds are received, which can play a role in preventing such risks, etc. Order risks may be generated by fraud detection tools submitted by third parties through order risk APIs, etc. Before fulfillment, merchants may need to obtain payment information or wait to receive payment information in order to mark the order as paid, so that the merchant can prepare to deliver the product. Such situations can be reviewed accordingly. The review process can be implemented by the fulfillment component. Merchants can use the fulfillment component to review, adjust work, and trigger related fulfillment services, such as: manual fulfillment services, which are used when merchants pick and pack products in boxes, purchase shipping labels and enter their tracking numbers, or simply mark items as fulfilled; custom fulfillment services, which can define email notifications; API fulfillment services, which can trigger third-party applications to create fulfillment records in third parties; legacy fulfillment services, which can trigger custom API calls from the business management engine to third parties; gift card fulfillment services. You can provide the generation of numbers and activate gift cards. Merchants can use the order printer application to print shipping orders. The fulfillment process can be executed when the items are packed in boxes and ready for shipping, tracking, delivery, consumer receipt verification, etc.

[0036] It can be seen that the services provided by the e-commerce platform are based on products. The corresponding commodity data is the basic data of the e-commerce platform. The commodity information is provided through the commodity data. The mining and utilization of commodity data is the basis for realizing various technical services, including the use of user transaction data and commodity data in the commodity data of the e-commerce platform to provide basic services for the operation of the data processing system. Therefore, the data processing system can run on any one or more servers of the e-commerce platform's cluster, so as to realize various functions by using the various commodity data provided by the e-commerce platform.

[0037] A customer service task response method of the present application can be programmed as a computer program product and deployed in a client or server for execution. For example, in the exemplary application scenario of the present application, it can be deployed and implemented in a server of an e-commerce customer service platform, thereby accessing an interface opened after the computer program product is running and performing human-computer interaction with the process of the computer program product through a graphical user interface to execute the method.

[0038] See also Figure 2 The customer service task response method of the present application, in its typical embodiment, includes the following steps:

[0039] Step S1100, obtaining a user question text provided by the user in the dialogue with the customer service;

[0040] The user may be a buyer on the e-commerce platform or a buyer in an independent website store built based on the website building service provided by the e-commerce platform, or a merchant of the independent website store.

[0041] The main way that e-commerce platforms and independent website stores provide customer service to users is through online chat tools. These tools can be chat plug-ins integrated into the website, which allow users to communicate directly with customer service in real time.

[0042] The e-commerce platform or the independent online store can provide the buyer with pre-sales customer service and after-sales customer service. The pre-sales customer service provides support before the buyer places an order by answering the buyer's questions, answering product-related questions, providing product information and suggestions, and assisting the buyer in making a purchase strategy; the after-sales service also assists the buyer in handling problems after the order transaction by answering the buyer's questions, including handling returns, exchanges, repairs, complaints, rights protection, etc. When necessary, the buyer can enable the pre-sales customer service and / or after-sales customer service to provide user question texts on demand in the corresponding chat window, and send the user question texts to the corresponding server.

[0043] The e-commerce platform can provide platform customer service to the merchants of the independent website stores. The platform customer service can assist the merchants in setting up and operating their online stores by answering the merchants' questions and answering various questions in the process of setting up and operating online stores. When necessary, the merchants can enable the platform customer service to provide user question texts on demand in the corresponding chat window and send the user question texts to the corresponding server.

[0044] The user provides a user question text, which can be directly entered in the chat window, or entered using the voice-to-text function in the chat window, or a single question text from a plurality of pre-constructed question texts can be selected as the user question text in the chat window as needed.

[0045] It is understandable that the chat window is a carrier for the user to carry out instant messaging dialogue with the customer service. In addition, the server obtains the user's question text by receiving it.

[0046] Step S1200: When a consultation intention text matching the user's question text is determined from a preset dialogue knowledge base, and there is a preset task flow corresponding to the consultation intention text, the task flow is driven by the user's question text, so that during the execution of the task flow, multiple rounds of corresponding dialogues are conducted with the user to obtain all target keywords required for the task goal;

[0047] The dialogue knowledge base stores multiple pre-prepared consultation intention texts and their corresponding standard answer texts or task processes. Each consultation intention text can be a single standard question text, or include a single standard question text and multiple similar question texts.

[0048] The standard question text is a pre-set standardized text that expresses the question, usually expressing a specific consulting intention. The similar question text is a preset question that is semantically similar or related to the corresponding standard question text, usually helping to expand the corresponding standard question so that it can be compatible with and match more varied user question expressions. For ease of understanding, an exemplary example is given: a standard question text: "How to check points", and its multiple similar question texts are: "How many points do I have", "How to check points", and "Where can I check points".

[0049] The standard response text is a standardized answer to the corresponding consultation intention text, which can directly respond to the user's consultation intention and answer the user's doubts. For the sake of understanding, an exemplary example is given, in which a user asks a question based on his consultation intention: "I want to see how many points are left", and the consultation intention text matching the user's question text is: "How to check points", and the consultation intention text corresponds to the standard response text: "Hello, please log in and enter the member page from..., click...".

[0050] It can be understood that if the question corresponding to the consultation intention text is simple and direct and does not require additional information to answer, a standard response text corresponding to the consultation intention text should be configured; on the contrary, if the question corresponding to the consultation intention text is complex and requires a customized answer based on the additional information provided by the user, a task flow corresponding to the consultation intention text should be configured. The task flow usually involves multiple decision points and multiple conversations with the user to ensure that the problem is solved in a customized manner based on the additional information provided by the user.

[0051] In one embodiment, a text representation model is used to extract the deep semantic information of the user's question text, and the deep semantic information is vectorized to obtain the corresponding user question vector. Then, the similarity between the user question vector and the text semantic vector corresponding to each question text included in each consultation intention text in the dialogue knowledge base is calculated, and the highest similarity is screened. When the similarity exceeds a preset threshold, it is confirmed that the consultation intention text corresponding to the similarity matches the user's question text. Further, if the consultation intention text is associated with the corresponding standard answer text in the dialogue knowledge base, the standard answer text is pushed to the user without executing the task flow. It can be understood that the text representation model can be used in advance to determine the text semantic vector corresponding to the question text for each question text included in each consultation intention text in the dialogue knowledge base, and the deep semantic information of the question text can be vectorized with it. The text representation model can use open source pre-trained to convergence Text Transfomer, RoBERTa, XLM-RoBERTa, MPNet, BERT and other models suitable for semantic feature representation in the NLP field. Those skilled in the art can choose one to implement as needed. The similarity calculation can be implemented by any available vector distance algorithm, including but not limited to any of the following: cosine similarity algorithm, vector dot product algorithm, Manhattan distance, Euclidean distance algorithm, Pearson correlation coefficient, etc. The preset threshold can be set by those skilled in the art as needed.

[0052] Various types of sub-processes can be pre-constructed to configure the task processes corresponding to each consultation intention text. The various types of sub-processes are judgment sub-process, inquiry sub-process, waiting sub-process, keyword extraction sub-process, knowledge response sub-process, and request response sub-process. In addition, the task process can achieve one task goal, or multiple task goals. The task goal is to be able to make a direct response to a consultation intention and provide a corresponding target response text. Therefore, achieving multiple task goals is to be able to make a direct response to multiple consultation intentions and provide corresponding target response texts. Secondly, when the task process is executed and any one of its task goals is achieved, the task process is terminated. For the sake of understanding the exemplary example, the consultation intention text is: "Apply for after-sales service", and its task process can achieve a task goal that can be: "Return goods for users", and the corresponding target response text provided can be "Application details link", and the various target keyword types required for the task goal and their corresponding target keywords can be: "Order number: xxx...xxx", "After-sales type: return and refund";

[0053] The inquiry intention text is: "How to bind a domain name", and its task flow can achieve various task goals: the first task goal: "How to bind a domain name, in the trial period", the corresponding target response text can be: "Hello, based on the latest risk control and safety factor regulations, trial period stores do not support binding", and the various target keyword types required for this task goal and their corresponding target keywords can be: "Whether in the trial period: in the trial period";

[0054] The second task objective: "How to bind a domain name, not in the trial period, not purchased a domain name", the corresponding target response text can be: "Hello, if you still have a domain name, you need to purchase a domain name yourself. It is recommended to purchase it at... or..., the domain names provided by these domain name service providers are more stable", the target keywords required for this task objective and their corresponding target keywords can be: "Whether it is in the trial period: Not in the trial period", "Whether the domain name has been purchased: The domain name has not been purchased";

[0055] The third task objective: "How to bind a domain name, not in the trial period, the domain name has been purchased, and the specified DNS record value has not been added in the domain name service provider", the corresponding target response text can be: "Hello, you need to configure... in the DNS... page." The target keywords required for this task objective and their corresponding target keywords can be: "Is it in the trial period: Not in the trial period", "Has the domain name been purchased: The domain name has been purchased", "Has the specified DNS record value been added in the domain name service provider: The specified DNS record value has not been added in the domain name service provider";

[0056] The fourth task objective: "How to bind a domain name, not in the trial period, the domain name has been purchased, and the specified DNS record value has been added in the domain name service provider", the corresponding target response text can be: "Hello, you have bound the domain name", and the target keywords required for this task objective and their corresponding target keywords can be: "Is it in the trial period: Not in the trial period", "Has the domain name been purchased: The domain name has been purchased", "Has the specified DNS record value been added in the domain name service provider: The specified DNS record value has been added in the domain name service provider".

[0057] The judgment sub-process includes parsing the judgment expression, determining the target entity type to which the target keyword to be judged belongs, obtaining the target keyword from the keyword mapping table according to the target entity type, judging whether the target keyword is obtained, and when the target keyword is obtained, performing corresponding conditional judgment on the target keyword, and deciding the next sub-process to be executed according to the judgment result; when the target keyword is not obtained, deciding to obtain the inquiry sub-process corresponding to the target keyword by asking the user as the next sub-process to be executed.

[0058] The inquiry sub-process includes confirming from the keyword mapping table whether the target keyword corresponding to the target entity type specified in this sub-process exists. When the target keyword exists, skipping this sub-process and executing the next sub-process immediately following it; when the target keyword does not exist, querying the user and sending the user a customer service inquiry text corresponding to the target keyword, ending this sub-process and executing the next sub-process immediately following it.

[0059] The waiting sub-process includes, after pushing the customer service inquiry text to the user, waiting until obtaining the user supplementary text provided by the user, and appending the user supplementary text to the user question text of the task process driving this sub-process, ending this sub-process and executing the next sub-process immediately following it.

[0060] The keyword extraction sub-process includes extracting all target keywords required for each task goal that can be achieved by the task process in which this sub-process is located from the user question text, and associating the extracted target keywords with the target entity type to which they belong and storing them in the keyword mapping table.

[0061] The request-response sub-process includes obtaining the target keywords corresponding to the various target implementation types required for the task objectives specified in this sub-process from the keyword mapping table, using them as input parameters for the business request, obtaining the business request, and then calling the business interface to process the business request to obtain the request result returned by the interface, sending it to the user as the target response text, and ending the task process.

[0062] The knowledge response sub-process includes obtaining the target keywords corresponding to each target entity type required for the task objectives specified by this sub-process from the keyword mapping table, splicing all target keywords after the user question text that drives the task process in which this sub-process is located, obtaining the standard response text corresponding to the consultation intention text that matches the spliced ​​user question text from the dialogue knowledge base, sending it to the user as the target response text, and ending the task process.

[0063] The keyword mapping table stores each target keyword and the target entity type to which it belongs.

[0064] Those skilled in the art can flexibly configure the task flow corresponding to the consultation intention text according to the above disclosure, configure each sub-flow therein, and set the execution order between each sub-flow.

[0065] It is not difficult to understand that when a task process is driven by a user question text, the configuration of the task process is loaded, and the sub-processes in the task process are executed according to the pre-set execution order between the sub-processes in the task process, so that during the execution process, each sub-process is executed accordingly based on its previous sub-process until the task process ends.

[0066] Step S1300: determine the corresponding target response text according to all target keywords, push it to the user, and end the task process.

[0067] It is not difficult to understand that when the task process executes to the request response sub-process or the knowledge response sub-process, the sub-process will display all the target keywords required to obtain the task goal, and based on this, the target response text is finally determined and pushed to the user, thereby making a direct response to the user's consultation intention, and should answer the user's doubts, thereby achieving the task goal and ending the task process.

[0068] It can be known from the typical embodiments of the present application that the technical solution of the present application has many advantages, including but not limited to the following aspects:

[0069] This application first obtains the user question text in the conversation between the user and the customer service, and then determines the user's consulting intention that matches the user's question text from the preset conversation knowledge base, so that the user's real needs can be understood and identified. After the user's consulting intention is determined, if there is a preset task process corresponding to the intention, it means that the consulting intention is not clear enough. In this regard, the preliminary consulting intention will drive this task process, and the user will have corresponding multiple rounds of dialogue, so as to gradually understand the user's real consulting intention, and correspondingly collect and correctly answer all the target keywords required for the task goal of the real consulting intention. Thus, the workload of manual customer service is reduced, and fast and accurate customer service is provided at the same time. Through targeted multiple rounds of dialogue, the user can be gradually guided to provide additional information corresponding to the preliminary consulting intention, ensuring that the collected information is comprehensive and accurate, and the real consulting intention can be located. In addition, the response speed can be improved, the possible communication delays and errors of manual customer service can be avoided, and users can be ensured to get feedback in time.

[0070] Finally, all target keywords required for the task goal are collected, and the corresponding target response text is determined based on this, and pushed to the user, so as to achieve the task goal and end the task process. This not only improves the service efficiency of customer service, but also improves the customer service experience of users, and can provide more personalized and accurate services.

[0071] In a further embodiment, step S1200, the execution process of the task flow, includes the following steps:

[0072] Step S1210: Obtain target keywords corresponding to each target entity type required for the task goal from the user question text; when not all target keywords required for the task goal are obtained, execute the corresponding inquiry sub-process according to the unobtained target keywords, obtain the user supplementary text obtained by asking the user, and append it to the user question text;

[0073] In one embodiment, the keyword extraction sub-process in the task process is executed, first by segmenting the user's question text to obtain a segmentation sequence consisting of corresponding word units, the segmentation can be achieved by using an e-commerce dictionary, an open source segmenter, part-of-speech tagging, named entity recognition, etc., then, each word unit in the segmentation sequence is semantically matched with each target entity type required by the task target, when the degree of correlation between the semantics of the word unit and the semantics of the target entity type exceeds a preset threshold, the word unit is determined to be a target keyword belonging to the target entity type; when the degree of correlation between the semantics of the target entity type and the semantics of each word unit does not exceed a preset threshold, it is determined that there is no target keyword corresponding to the target entity type in the user's question text. The preset threshold can be set by a person skilled in the art as needed, and the degree of correlation matching can be flexibly implemented by a person skilled in the art based on a deep learning algorithm in the field of NLP.

[0074] Further, the judgment sub-process immediately following the keyword extraction sub-process is executed, and for each of the task objectives, it is judged whether each target keyword obtained is equivalent to each target keyword required for the task objective corresponding to the same target entity type, and the total number of target keywords obtained is compared with the total number of target keywords required for the task objective. When the total number of target keywords obtained is less than the total number of target keywords required for the task objective, and each target keyword obtained is equivalent to each target keyword required for the task objective corresponding to the same target entity type, it is confirmed that all target keywords required for the task objective have not been obtained at this time, and a decision is made to obtain the inquiry sub-process corresponding to the unobtained target keyword by asking the user as the next sub-process to be executed. It can be understood that, after that, by executing the inquiry sub-process, the user can be asked to send the user a customer service inquiry text for obtaining the target keyword that has not been obtained, and then, by executing the waiting sub-process, the user supplementary text obtained by asking the user can be obtained and appended to the user question text;

[0075] When the total number of target keywords obtained is greater than or equal to the total number of target keywords required, and the target keywords obtained are not equivalent to the target keywords required for the task goal corresponding to the same target entity type, it is confirmed that the task goal cannot be achieved.

[0076] Step S1220: iterate the above process with the appended user question text until all target keywords required for the task objective in the user question text are obtained.

[0077] After the execution of the waiting sub-process, by executing the keyword extraction sub-process again, the target keyword mapping that was not obtained in the appended user question text can be obtained and associated with the target entity type to which it belongs, and stored in the keyword mapping table. After that, by executing the judgment sub-process again, for each of the task objectives, it is judged whether each obtained target keyword is equivalent to each target keyword required for the task objective corresponding to the same target entity type, and the total number of obtained target keywords is compared with the total number of target keywords required for the task objective. When the total number of obtained target keywords is equal to the total number of target keywords required for the task objective, and each obtained target keyword is equivalent to each target keyword required for the task objective corresponding to the same target entity type, it is confirmed that all target keywords required for the task objective in the user question text are obtained, and the iteration is terminated. In addition, after this, by executing the request response sub-process or knowledge response sub-process corresponding to the task objective pre-specified, the target response text can be pushed to the user, and the task objective is achieved and the task process is terminated;

[0078] When the total number of target keywords obtained is less than the total number of target keywords required for the task objective, and each target keyword obtained is equivalent to each target keyword required for the task objective corresponding to the same target entity type, it is confirmed that all target keywords required for the task objective have not been obtained at this time, and the above process is iterated.

[0079] In this embodiment, by asking the user multiple times during the execution of the task flow, the user's supplementary text provided by the user each time is appended to the user question text, thereby realizing corresponding multiple rounds of dialogue with the user, and extracting target keywords that were not obtained previously from the user question text after each appending, so that all target keywords required for the task goal are finally obtained.

[0080] In a further embodiment, step S1210, obtaining target keywords corresponding to each target entity type required for the task goal from the user question text, includes the following steps:

[0081] Step S2201: when repeatedly obtaining target keywords corresponding to the same target entity type required by the task target, determine whether the previously obtained target keywords need to be updated. If an update is required, replace the previously obtained target keywords with the latest obtained target keywords.

[0082] It is understandable that when a user replies to the same customer service inquiry text and provides different user supplementary text, it may lead to subsequent keyword extraction and repeated acquisition of target keywords corresponding to the same target entity type required for the task objective, or the same is true for extracting target keywords corresponding to the same target entity type multiple times during the keyword extraction process at different times.

[0083] For repeated target keywords corresponding to the same target entity type, the decision to update the previously obtained target keywords may depend on multiple factors, including any one or more of the following:

[0084] Depends on the timeliness of the previous and next information: If the latest target keyword contradicts the previously obtained target keyword, or the previous status is updated, then the previously obtained target keyword needs to be updated;

[0085] Depends on the accuracy of previous and next information: If the latest target keyword is more accurate or specific than the previously obtained target keyword, then the previously obtained target keyword needs to be updated;

[0086] Depends on the relevance of previous and next information: If the latest target keyword is more relevant to the current task goal than the previously obtained target keyword, then the previously obtained target keyword needs to be updated.

[0087] The decision not to update the previously obtained target keyword may depend on whether the newly obtained target keyword is the same as the previously obtained target keyword, or is substantially the same and has no more specific additional information, or does not provide any new additional information or information that contradicts the previous information, in which case the previously obtained target keyword does not need to be updated.

[0088] When the previously obtained target keyword needs to be updated, the previously obtained target keyword associated with the target entity type mapping stored in the keyword mapping table is reassigned to the latest target keyword to achieve the update of the previously obtained target keyword.

[0089] Step S2202: When there is no need to update the previously obtained target keyword, the latest obtained target keyword is eliminated.

[0090] At this time, the latest target keyword obtained through keyword extraction can be discarded.

[0091] In this embodiment, it is possible to adapt whether the repeatedly obtained target keywords need to be updated, and correspondingly replace or eliminate the target keywords with the latest obtained target keywords, so as to understand the user's intentions, identify changes in information, and update the information in the keyword mapping table accordingly to ensure that the answers finally provided to the user are accurate and relevant.

[0092] In a further embodiment, step S1210, obtaining target keywords corresponding to each target entity type required for the task goal from the user question text, includes the following steps:

[0093] Step S1211: embed each target entity type required by the task target and the user's question text into a preset prompt template to obtain a corresponding prompt text;

[0094] Execute the keyword extraction sub-process in the task process, embed the user question text and each target entity type associated with all task goals that can be achieved by the task process into a preset prompt template, and obtain the corresponding prompt text,

[0095] The prompt template includes a task description, a user question text to be embedded, and a target entity type to be embedded. Those skilled in the art can refer to the following disclosure to flexibly set the prompt template. An exemplary prompt template is given as follows: Task description in the prompt template: "Please refer to the example. From the following user question text provided by the user in the conversation with the customer service, determine whether the target keyword corresponding to each given target entity type exists. If the target keyword corresponding to the target entity type exists, output the target entity type and its target keyword. Example:

[0096] Sample user question text: "How do I bind a domain name? My store is in the trial period and I haven't purchased a domain name yet";

[0097] Examples of target entity types: "Whether in trial period", "Whether domain name has been purchased", "Whether specified DNS record value has been added in domain name service provider";

[0098] Sample output: The target keywords are: "Is it in the trial period: In the trial period", "Has the domain name been purchased: The domain name has not been purchased." ", the user question text to be embedded: "The provided user question text: {$user_Question$}", and the target entity types to be embedded: "The given target entity types: {$Target_Entity_type$}".

[0099] Step S1212: input the prompt text into the large language model to guide the model to extract the target keyword corresponding to the target entity type in the user question text.

[0100] The question text is input into a large language model to obtain each target entity type and its corresponding target keyword in the output result of the model, and each target entity type is mapped and associated with its corresponding target keyword and stored in a keyword mapping table.

[0101] In this embodiment, by using a large language model to extract target keywords in the user's question text, the reasoning knowledge and accurate understanding ability pre-learned by the large language model can be utilized to ensure efficient and accurate extraction of target keywords.

[0102] In a further embodiment, after step S1100, obtaining the user question text provided by the user in the dialogue with the customer service, and before step S1300, determining the corresponding target answer text according to all target keywords, the following steps are included:

[0103] Step S2200: for multiple task flows, obtain the user question text that drives the task flow, the corresponding multiple rounds of dialogue content with the corresponding user during the execution of the task flow, and all target keywords required for the corresponding task goal, and construct the task execution record of the task flow;

[0104] For each task flow, all the conversation contents between the user and the customer service from the beginning to the end of the conversation in the task flow are collected, including: the user's question text provided at the beginning of the conversation, the content of multiple rounds of conversations (including: the user's supplementary text and the customer service inquiry text in each round of conversation), and all the target keywords required to achieve the task goal corresponding to the end of the task flow. These conversation contents constitute the task execution record of the task flow.

[0105] Step S2210: Collect multiple task execution records corresponding to each task process and add them to the data set;

[0106] For each task process, collect enough task execution records corresponding to the task process, and these task execution records at least include the various task goals that can be achieved by the task process. This allows the subsequent large language model to learn and understand the unique dialogue model and user consultation intentions and their corresponding expressions in each task process, so that when responding to different texts provided by users, it can accurately identify the target keywords required for the corresponding task process and its associated task goals, clarify the user's consultation intentions and provide appropriate follow-up questions and / or responses. Further, all the collected task execution records are added to the dataset.

[0107] Step S2220: construct a training set based on the data set, fine-tune the training of the large language model until it converges, and drive the corresponding task process according to the user question text provided by the user, and conduct corresponding multiple rounds of dialogue with the user during the execution of the task process to obtain all target keywords required for the task goal.

[0108] The training set includes multiple training samples and their supervisory labels, wherein the training sample includes a single text provided by the user in a conversation between the user and the customer service in a single task execution record, and a preset prompt template, and the corresponding prompt text is obtained. The supervisory label of the training sample includes the text corresponding to the single text provided by the user in the task execution record in response to the single text provided by the user, and the target keywords required for the corresponding task goal extracted from the text spliced ​​from the single text provided by the user in the task execution record and all the texts previously provided by the user. Those skilled in the art can flexibly construct the training set based on the data set according to the disclosure herein.

[0109] A single training sample in the training set is called, and the training sample is input into the large language model to obtain the predicted output result of the large language model. A preset cross entropy loss function or mean square error loss function is called, which can be flexibly set by technicians in this field according to prior knowledge or experimental experience. The loss value corresponding to the predicted structured query statement is calculated according to the supervision label of the training sample. When the loss value reaches a preset threshold, it indicates that the large language model has been fine-tuned and trained to a convergence state, so that the training of the large language model can be terminated; when the loss value does not reach the preset threshold, based on LoRA (Low-Rank Adaptation of Large Language Mode ls) is used to fine-tune the large language model. Specifically, the parameters of the large language model are frozen or fixed, and only the dimension reduction matrix A and the dimension increase matrix B are trained, while the input and output dimensions of the large language model remain unchanged. When outputting, the dimension reduction matrix A and the dimension increase matrix B are superimposed with the parameters of the large language model to obtain the corresponding predicted output result. In addition, the initialization processing of the dimension reduction matrix A is implemented by random Gaussian distribution; the initialization processing of the dimension increase matrix B is implemented by 0 matrix, thereby fixing the original parameters of the weight matrix of the model, and performing an implicit low-rank transformation on the weight matrix, determining the weight change parameters of the weight matrix of the model according to the loss value, adjusting the weight matrix accordingly according to the weight change parameters, and then continuing to call other training samples and their supervisory labels to iteratively train the large language model until the model is fine-tuned and trained to a convergence state. The preset threshold can be set as needed by those skilled in the art according to the disclosure herein.

[0110] In this embodiment, the training process of the large language model is disclosed, so that after being trained to a convergent state, it can accurately identify the adapted task flow according to the user's consultation intention, and during the execution of the task flow, it can have multiple rounds of dialogue with the user, and extract all the target keywords required for the task goal by asking the user multiple times accordingly, and finally clarify the user's consultation intention.

[0111] In a further embodiment, step S1300, determining corresponding target response texts according to all target keywords, pushing the texts to the user, and ending the task flow, includes the following steps:

[0112] Step S1310: embed all target keywords into a preset work order template, obtain the corresponding work order as the target response text and push it to the user;

[0113] The work order template can be edited in advance by a technician in this field according to business needs, so that the only information that is not filled in the work order template is the target keywords required for the task target to be embedded, and in order to accurately embed each target keyword in the corresponding position in the work order template, the target entity type to which the target keyword belongs is edited and set in the position. For easy understanding, an exemplary example is given: the work order template for "product after-sales service": "Dear user, your return work order has been generated as follows:

[0114] Order number: {$order number$}

[0115] After-sales type: {$After-sales type$}

[0116] We will process your request as soon as possible and contact you for further assistance. Thank you for your patience and understanding. ”

[0117] It is not difficult to understand that after the work order template is pre-edited, it can be obtained by requesting the business interface based on all the target keywords. After that, all the target keywords are embedded into the preset work order template according to the target entity type to which they belong, and the corresponding work order returned by the interface can be obtained, and then it can be pushed to the user as the target response text.

[0118] Step S1320: When the user confirms the target response text, the target response text is submitted to the work order system.

[0119] It can be understood that after the target response text is pushed to the user, the user can be provided with the opportunity to confirm the target response text. When the user confirms that it is correct, the corresponding target response text will be automatically submitted to the work order system. When the user needs to modify it, an editing window can be provided for the user to edit the target response text. After the user completes the editing and confirms that it is correct, the corresponding target response text will be automatically submitted to the work order system.

[0120] In this embodiment, the corresponding work order is obtained by embedding all the target keywords required for the task objective into the work order template, and after the user confirms it, it is submitted to the work order system. Through dialogue with the user, accurate work orders can be generated efficiently and conveniently, eliminating the need for users to perform complicated work order application and editing operations by themselves, thereby improving the user's customer service experience.

[0121] In a further embodiment, after step S1320, submitting the target response text to the work order system, the following steps are included:

[0122] Step S1330, obtaining the waiting processing time of the target response text, and determining the time difference between the waiting processing time and the estimated processing time corresponding to the target response text;

[0123] A timer initialized to zero can be implemented, and its working process includes: after submitting the target response text to the work order system, starting the timer to start timing, obtaining the corresponding waiting processing time, and then, when the work order system returns the processing result of the target response text, ending the timer. It is not difficult to understand that the waiting processing time is the interval from the work order being submitted for processing to the completion of the processing.

[0124] The estimated processing time is the time required to process the work order corresponding to the work order template when the work order template is formulated, based on historical processing experience and / or actual online processing time.

[0125] The time difference is obtained by subtracting the waiting processing time of the target response text from the estimated processing time corresponding to the target response text, that is, quantifying the estimated remaining processing time.

[0126] Step S1340: When the ratio between the time difference and the estimated processing time reaches a preset threshold, a work order priority modification request is constructed and submitted to the work order system to increase the processing priority of the target response text, and a notification of urged processing is sent to the user.

[0127] The time difference is divided by the expected processing time to obtain the ratio, and the ratio is compared with the preset threshold. It is not difficult to understand that as the waiting processing time continues to increase, the ratio will continue to decrease. When the ratio decreases to be equal to the preset threshold, a work order priority modification request is constructed and submitted to the work order system, and a notification of urging processing is sent to the user, informing the user that his work order has been followed up and has now been expedited. The notification can be sent by email, SMS, or notification in the conversation.

[0128] The work order priority modification request may be to request the work order system to first obtain the processing priority corresponding to the target response text, i.e., the work order, and then to increase the processing priority to a higher level. Thus, the work order system may process the target response text, i.e., the work order with higher priority than before, so that it is processed faster and / or the time required for processing is shorter.

[0129] In this embodiment, the process of expediting work orders is revealed to ensure that the work orders of users who are continuously followed up are processed in a timely manner. Moreover, when it is found that the work order may not be processed in time during the follow-up, the user is proactively notified that it has been urged to be processed, leaving the user with tangible service work results and improving the user's customer service experience.

[0130] See also Figure 3 , a customer service task response device provided to meet one of the purposes of the present application is a functional embodiment of the customer service task response method of the present application. On the other hand, the device is a customer service task response device provided to meet one of the purposes of the present application, including a text acquisition module 1100, a keyword determination module 1200 and a user response module 1300, wherein the text acquisition module 1100 is used to obtain the user question text provided by the user in the dialogue with the customer service; the keyword determination module 1200 is used to drive the task process with the user question text when a consultation intention text matching the user question text is determined from a preset dialogue knowledge base and there is a preset task process corresponding to the consultation intention text, so that during the execution of the task process, corresponding multiple rounds of dialogue are carried out with the user to obtain all target keywords required for the task goal; the user response module 1300 is used to determine the corresponding target response text according to all target keywords, push it to the user, and end the task process.

[0131] In a further embodiment, the keyword determination module 1200 includes: a text appending submodule, which is used to obtain the target keywords corresponding to each target entity type required for the task objective from the user question text; when all the target keywords required for the task objective are not obtained, the corresponding inquiry subprocess is executed according to the unobtained target keywords, and the user supplementary text obtained by asking the user is obtained and appended to the user question text; an iterative acquisition submodule, which is used to iterate the above process with the appended user question text until all the target keywords required for the task objective in the user question text are obtained.

[0132] In a further embodiment, the text appending submodule includes: a keyword updating unit, which is used to determine whether the previously obtained target keyword needs to be updated when the target keyword corresponding to the same target entity type required for the task objective is repeatedly obtained. When updating is required, the previously obtained target keyword is replaced by the latest target keyword; a keyword non-updating unit, which is used to eliminate the latest target keyword when the previously obtained target keyword does not need to be updated.

[0133] In a further embodiment, the text appending submodule includes: a prompt construction unit, which is used to embed each target entity type required for the task objective and the user question text into a preset prompt template to obtain a corresponding prompt text; a prompt input unit, which is used to input the prompt text into the large language model to guide the model to extract the target keywords corresponding to the target entity type in the user question text.

[0134] In a further embodiment, after the text acquisition module 1100 and before the user response module 1300, it includes: a record construction submodule, which is used to obtain the user question text that drives the task process for multiple task processes, and conduct corresponding multi-round dialogues with the corresponding user during the execution of the task process, as well as all target keywords required for the corresponding task objectives, to construct the task execution record of the task process; a data set adding submodule, which is used to collect multiple task execution records corresponding to each task process and add them to the data set; a model training submodule, which is used to construct a training set based on the data set, fine-tune the training of the large language model until it converges, and drive the corresponding task process according to the user question text provided by the user, and conduct corresponding multi-round dialogues with the user during the execution of the task process to obtain all target keywords required for the task objective.

[0135] In a further embodiment, the user response module 1300 includes: a user push sub-module, used to embed all target keywords into a preset work order template, obtain the corresponding work order as a target response text and push it to the user; a system submission sub-module, used to submit the target response text to the work order system when the user confirms the target response text.

[0136] In a further embodiment, after the system submits the sub-module, it includes: a time difference determination sub-module, which is used to obtain the waiting processing time of the target response text, and determine the time difference between the waiting processing time and the expected processing time corresponding to the target response text; an urging processing sub-module, which is used to construct a work order priority modification request and submit it to the work order system when the ratio between the time difference and the expected processing time reaches a preset threshold, so as to increase the processing priority of the target response text, and send a notification of urged processing to the user.

[0137] In order to solve the above technical problems, the present application also provides a computer device. Figure 4 As shown, a schematic diagram of the internal structure of a computer device. The computer device includes a processor, a computer-readable storage medium, a memory, and a network interface connected via a system bus. Among them, the computer-readable storage medium of the computer device stores an operating system, a database, and computer-readable instructions. The database may store a control information sequence. When the computer-readable instructions are executed by the processor, the processor may implement a customer service task response method. The processor of the computer device is used to provide computing and control capabilities to support the operation of the entire computer device. The memory of the computer device may store computer-readable instructions. When the computer-readable instructions are executed by the processor, the processor may execute the customer service task response method of the present application. The network interface of the computer device is used to connect and communicate with a terminal. Those skilled in the art will understand that Figure 4 The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0138] In this embodiment, the processor is used to execute Figure 3 The memory stores the program code and various data required to execute the above modules or submodules. The network interface is used to transmit data between user terminals or servers. The memory in this embodiment stores the program code and data required to execute all modules / submodules in the customer service task response device of this application, and the server can call the program code and data of the server to execute the functions of all submodules.

[0139] The present application also provides a storage medium storing computer-readable instructions. When the computer-readable instructions are executed by one or more processors, the one or more processors execute the steps of the customer service task response method of any embodiment of the present application.

[0140] A person skilled in the art can understand that all or part of the processes in the above-mentioned embodiments of the present application can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the aforementioned storage medium can be a computer-readable storage medium such as a disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0141] In summary, the present application can clarify complex user inquiry intentions and make correct responses.

[0142] It will be understood by those skilled in the art that the various operations, methods, steps, measures, and schemes in the processes discussed in this application may be alternated, changed, combined, or deleted. Furthermore, other steps, measures, and schemes in the various operations, methods, and processes discussed in this application may also be alternated, changed, rearranged, decomposed, combined, or deleted. Furthermore, the steps, measures, and schemes in the various operations, methods, and processes in the prior art that are open source and in this application may also be alternated, changed, rearranged, decomposed, combined, or deleted.

[0143] The above description is only a partial implementation method of the present application. It should be pointed out that for ordinary technicians in this technical field, several improvements and modifications can be made without departing from the principles of the present application. These improvements and modifications should also be regarded as the scope of protection of the present application.

Claims

1. A customer service task response method, characterized in that: The steps include: Obtain the text of user questions provided by the user in the conversation with customer service; When a consultation intention text matching the user's question text is determined from a preset dialogue knowledge base, and there is a preset task flow corresponding to the consultation intention text, the task flow is driven by the user's question text, so that during the execution of the task flow, multiple rounds of corresponding dialogues are conducted with the user to obtain all target keywords required for the task goal; The corresponding target response text is determined based on all target keywords, pushed to the user, and the task process is ended.

2. The customer service task response method according to claim 1, characterized in that: The execution process of the task flow includes the following steps: Obtaining target keywords corresponding to each target entity type required for the task goal from the user question text; when not all target keywords required for the task goal are obtained, executing a corresponding inquiry sub-process according to the unobtained target keywords, obtaining a user supplementary text obtained by asking the user, and appending it to the user question text; The above process is iterated with the appended user question text until all target keywords required for the task objective in the user question text are obtained.

3. The customer service task response method according to claim 1, characterized in that: Obtaining target keywords corresponding to each target entity type required for the task goal from the user question text includes the following steps: When repeatedly obtaining target keywords corresponding to the same target entity type required by the task objective, determine whether the previously obtained target keywords need to be updated, and if so, replace the previously obtained target keywords with the latest obtained target keywords; When there is no need to update the previously obtained target keyword, the latest obtained target keyword is eliminated.

4. The customer service task response method according to claim 1, characterized in that: Obtaining target keywords corresponding to each target entity type required for the task goal from the user question text includes the following steps: Embed each target entity type required by the task goal and the user's question text into a preset prompt template to obtain a corresponding prompt text; The prompt text is input into the large language model to guide the model to extract the target keyword corresponding to the target entity type in the user question text.

5. The customer service task response method according to claim 1, characterized in that: After obtaining the user question text provided by the user in the conversation with the customer service, and before determining the corresponding target answer text according to all target keywords, the following steps are included: For multiple task flows, obtain the user question text that drives the task flow, conduct multiple rounds of dialogue with the corresponding users during the execution of the task flow, and all target keywords required for the corresponding task goals, and construct the task execution record of the task flow; Collect multiple task execution records corresponding to each task process and add them to the data set; A training set is constructed based on the data set, and a large language model is fine-tuned and trained until it converges. The training set is used to drive a corresponding task flow according to a user question text provided by a user, and multiple rounds of dialogues are conducted with the user during the execution of the task flow to obtain all target keywords required for the task goal.

6. The customer service task response method according to claim 1, characterized in that: Determine the corresponding target response text according to all target keywords, push it to the user, and end the task process, including the following steps: Embed all target keywords into a preset work order template, obtain the corresponding work order as the target response text and push it to the user; When the user confirms the target response text, the target response text is submitted to the work order system.

7. The customer service task response method according to claim 6, characterized in that: After the target response text is submitted to the work order system, the following steps are included: Obtaining the waiting processing time of the target response text, and determining the time difference between the waiting processing time and the estimated processing time corresponding to the target response text; When the ratio between the time difference and the estimated processing time reaches a preset threshold, a work order priority modification request is constructed and submitted to the work order system to increase the processing priority of the target response text, and a notification of urged processing is sent to the user.

8. A customer service task response device, characterized in that: include: A text acquisition module is used to acquire the user question text provided by the user in the conversation with the customer service; A keyword determination module is used to drive the task flow with the user's question text when a consultation intention text matching the user's question text is determined from a preset dialogue knowledge base and there is a preset task flow corresponding to the consultation intention text, so that during the execution of the task flow, multiple rounds of dialogues are carried out with the user to obtain all target keywords required for the task goal; The user response module is used to determine the corresponding target response text according to all the target keywords, push it to the user, and end the task process.

9. A computer device comprising a central processing unit and a memory, characterized in that: The central processing unit is used to call and run the computer program stored in the memory to execute the steps of the method according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that: It stores a computer program implemented according to the method described in any one of claims 1 to 7 in the form of computer-readable instructions, and when the computer program is called and executed by a computer, the steps included in the corresponding method are executed.

Images