Method, device, equipment, and computer program for generating a question-answering service

Abstract

A method for generating a question answering service, the method being executed by a computer device, includes the step (S101) of displaying a set of function code package components on an edit interaction page of a question answering service editing application, the set of function code package components including a plurality of types of function code package components, the plurality of types of function code package components including at least one trigger component, at least one question answering strategy component, and at least one linguistic question answering model component; and the step (S102) of displaying, in response to a component selection operation on the set of function code package components, the trigger component, the question answering strategy component, and the linguistic question answering model component selected by the component selection operation on a question answering service editing page included in the edit interaction page, the question answering service editing page further including and a step (S103) of generating a question answering service based on the process flow connection relationships among the selected trigger component, question answering strategy component, and linguistic question answering model component in response to a question answering service confirmation operation triggered on the edit interaction page of the question answering service editing application, wherein the trigger component in the question answering service is used to invoke the question answering strategy component in the question answering service, the question answering strategy component in the question answering service is used to perform question answer filtering on the business problem after obtaining the business problem, and the model component in the question answering service is used to output the question answering service result according to the filtered business problem.

Description

[Technical Field]

[0001] This application claims priority to a Chinese patent application bearing application number 2023109149947 and entitled "Method, apparatus, device, and readable storage medium for generating a question-answering service," filed on July 24, 2023, the entire text of which is incorporated herein by reference.

[0002] The present application relates to the field of computers, and more particularly to a method, apparatus, device, and readable storage medium for generating a question-answering service. [Background technology]

[0003] Currently, the technical hurdles for large-scale language model question-answering robots are high, and research and development costs are high. Existing development methods are all manual development performed using code, which has low development efficiency and requires a large amount of time and effort to complete tasks. Implementation hurdles are high, engineers need to have a deep understanding of the algorithms underlying the large-scale language model and prepare huge training datasets, which means high labor costs. Completing tasks requires manual intervention in development, which results in a high implementation error rate and makes it easy for human errors and loopholes to appear, thereby affecting the quality and efficiency of work. Summary of the Invention [Problem to be solved by the invention]

[0004] According to an embodiment of the present application, a method, an apparatus, a device, and a readable storage medium for generating a question-answering service are provided. [Means for solving the problem]

[0005] In one aspect, a method for generating a question answering service is provided, the method being executed by a computing device and comprising: displaying a set of function code package components in an edit interaction page of the question answering service editing application, the set of function code package components including a plurality of types of function code package components, the plurality of types of function code package components including at least one trigger component, at least one question answering strategy component, and at least one linguistic question answering model component; a step of displaying, in response to a component selection operation for the set of function code package components, a trigger component, a question answering strategy component, and a linguistic question answering model component selected by the component selection operation in a question answering service edit page included in the edit interaction page, wherein the question answering service edit page is further used to display a process flow connection relationship between the selected trigger component, the question answering strategy component, and the linguistic question answering model component; and generating a question answering service based on a process flow connection relationship among a selected trigger component, a question answering strategy component, and a linguistic question answering model component in response to a question answering service confirmation operation triggered on an edit interaction page of the question answering service editing application, wherein the trigger component in the question answering service is used to invoke the question answering strategy component in the question answering service, the question answering strategy component in the question answering service is used to perform question answer filtering on the business problem after obtaining the business problem, and the model component in the question answering service is used to output a question answering service result according to the filtered business problem.

[0006] In another aspect, there is provided an apparatus for generating a question-answering service, comprising: a component display module used to display a set of function code package components in an edit interaction page of a question answering service edit application, the set of function code package components including a plurality of types of function code package components, the plurality of types of function code package components including at least one trigger component, at least one question answering strategy component, and at least one linguistic question answering model component; a component editing module used to display, in response to a component selection operation for the set of function code package components, a trigger component, a question answering strategy component, and a linguistic question answering model component selected by the component selection operation in a question answering service editing page included in the editing interaction page, wherein the question answering service editing page is further used to display a process flow connection relationship between the selected trigger component, the question answering strategy component, and the linguistic question answering model component; and a component confirmation module used to generate a question answering service based on a process flow connection relationship among a selected trigger component, a question answering strategy component, and a linguistic question answering model component in response to a question answering service confirmation operation triggered on an edit interaction page of the question answering service editing application, wherein the trigger component in the question answering service is used to invoke the question answering strategy component in the question answering service, the question answering strategy component in the question answering service is used to perform question answer filtering on the business problem after obtaining the business problem, and the model component in the question answering service is used to output the question answering service result according to the filtered business problem.

[0007] In yet another aspect, a computing device is provided, including a processor and a memory, wherein the memory stores computer-readable instructions that, when executed by the processor, cause the processor to implement a method provided by an embodiment of the present application.

[0008] In yet another aspect, a computer-readable storage medium is provided, the computer-readable storage medium storing computer-readable instructions that, when executed by a processor, cause the processor to implement a method provided by an embodiment of the present application.

[0009] In yet another aspect, a computer program product is provided, comprising computer-readable instructions that, when executed by a processor, cause the processor to implement a method provided by an embodiment of the present application.

[0010] The details of one or more embodiments of the present application are set forth in the drawings and description that follow. Other features, objects, and advantages of the present application will become apparent from the description, drawings, and claims.

[0011] In order to more clearly describe the embodiments of the present application or the technical solutions in the prior art, the following briefly introduces the drawings that need to be used in the description of the embodiments or the prior art. It should be apparent that the drawings in the following description are merely embodiments of the present application, and those skilled in the art can obtain other drawings according to the drawings further disclosed without any creative work. [Brief explanation of the drawings]

[0012] [Figure 1] 1 is a schematic diagram of a network architecture provided by an embodiment of the present application; [Figure 2] 1 is a schematic diagram of a scenario for generating a question answering service provided by an embodiment of the present application; [Figure 3]FIG. 2 is a schematic diagram of another scenario for generating a question answering service provided by an embodiment of the present application. [Figure 4] 1 is a flowchart of a method for generating a question and answer service provided by an embodiment of the present application; [Figure 5] 1 is a schematic diagram of a scenario for generating a question answering service provided by an embodiment of the present application; [Figure 6] 1 is a flowchart of accessing a question-answering service in a communication application provided by an embodiment of the present application; [Figure 7] 1 is a flowchart of a method for generating a question and answer service provided by an embodiment of the present application; [Figure 8] 1 is a structural schematic diagram of a question answering service generating device provided by an embodiment of the present application; [Figure 9] 1 is a structural schematic diagram of a computer device provided by an embodiment of the present application; DETAILED DESCRIPTION OF THE INVENTION

[0013] The technical solutions in the embodiments of the present application will be described below clearly and completely in conjunction with the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only some of the embodiments of the present application, not all of the embodiments, and all other embodiments obtained by those skilled in the art based on the embodiments of the present application without any creative work are all within the scope of protection of the present application.

[0014] Artificial intelligence technology is a comprehensive discipline with a wide range of related fields, including both hardware and software technologies. Fundamental AI technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interaction systems, and mechatronics. AI software technology primarily encompasses several major areas, including computer vision, speech processing, natural language processing, and machine learning / deep learning.

[0015] Deep learning (DL) is the learning of specific disciplines and representation levels of sample data, and the information acquired during this learning process is extremely useful for interpreting data such as text, images, and voice. Its ultimate goal is to enable machines to have human-like analytical learning capabilities and to be able to identify data such as text, images, and voice. Deep learning is a complex machine learning algorithm, and its results in terms of voice and image identification far exceed those of conventional related technologies.

[0016] The solution provided by the embodiments of the present application relates to artificial intelligence, and is specifically illustrated by the following embodiments.

[0017] Referring to Fig. 1, Fig. 1 is an example of a network architecture provided by an embodiment of the present application. The schematic diagram of the network architecture may include a work server 100, a terminal device 200, a communication server 300, and a terminal device 400, where a communication connection may exist between the work server 100 and the terminal device 200, a communication connection may exist between the work server 100 and the communication server 300, or a communication connection may exist between the communication server 300 and the terminal device 400. Here, the above communication connection is not limited to a connection method, and may be directly or indirectly connected by a wired communication method, directly or indirectly connected by a wireless communication method, or may be connected by other methods, and the present application is not limited thereto.

[0018] It should be understood that a question-answering service editing application may be installed on the terminal device 200, and when the question-answering service editing application is running on the terminal device 200, it can exchange data with the operation server 100, i.e., the operation server 100 is a background server for the question-answering service editing application. The terminal device 200 can generate a question-answering service through the question-answering service editing application, and the question-answering service can be accessed through the communication application. A communication application may be installed on the terminal device 400, and when the communication application is running on the terminal device 400, it can exchange data with the communication server 300, i.e., the communication server 300 is a background server for the communication application. Here, the question-answering service refers to an intelligent service that can generate answers to questions using a linguistic question-answering model. The question-answering service may be an independent client terminal or an embedded sub-client terminal integrated in the communication application. The operation server 100 and the communication server 300 may be integrated into the same server, and the terminal device 200 and the terminal device 400 may be the same terminal device, but this is not limited thereto.

[0019] Specifically, a question answering service editing application is installed on the terminal device 200, and the question answering service editing application on the terminal device 200 can display a set of function code package components, which set of function code package components includes one or more trigger components, one or more question answering strategy components, and one or more linguistic question answering model components.

[0020] Here, the term "functional code package component" refers to a functional component having pre-orchestrated and pre-defined code functions. The functional code package component may be a component having a predetermined functional function formed by packaging at least one of a pre-written program module or library, and the functional function is used to process a business problem input by a user. The trigger component is a component that triggers the question-answering service workflow and is the starting component for operating the entire question-answering service workflow. The trigger component can be used to detect whether a business problem proposed by a user meets the question-answering trigger condition, and when the problem meets the question-answering trigger condition, it can invoke a question-answering strategy component in the question-answering service. The question-answering strategy component is a component that filters business problems and can filter business problems that cannot be processed by the linguistic question-answering model component. The linguistic question-answering model component can perform semantic identification on the filtered business problems and output the question-answering service results for the business problems.

[0021] The terminal device 200 can trigger a component selection operation in the question-answering service editing application and select a function code package component on a question-answering service editing page of the question-answering service editing application. The selection of the function code package component can be performed by dragging a function code package component displayed on the question-answering service editing page or by establishing a new function code package component. The selected function code package components are displayed on the editing page of the question-answering service editing application, and then a process flow connection relationship between the selected function code package components is established by a line connection. The process flow connection relationship can characterize the flow sequence of at least two function code package components. The line connection can be automatically established based on a preset rule, or can be triggered in response to a connection operation on the function code package components.

[0022] The terminal device 200 can trigger a component setting operation in the question-answering service editing application, display a parameter setting page, and further modify component information of the function code package component on the parameter setting page of the question-answering service editing application. The component information may include a process flow connection relationship, an instance name, and an instance template identifier. The instance name is the name of the customized function code package component, and the instance template identifier is used to determine a function function applied to the function code package component and process the business problem according to the function function. For example, the authority determination subcomponent in the question-answering strategy component can use the function function userlist.find() to search in the user list (userlist) with question authority (problem authority scope) and filter out business problems that do not return results.

[0023] In response to triggering a component confirmation operation in the question answering service editing application, the terminal device 200 can generate a question answering service with process flow connection relationships between the selected trigger component, the question answering strategy component, and the linguistic question answering model component.

[0024] The question-answering service can be accessed through a communication application, which is installed on the terminal device 400. The communication application on the terminal device 400 can display a question-answering interaction page, and a user can input a business problem on the question-answering interaction page. Here, the business problem includes a trigger parameter. When the trigger parameter satisfies a question-answering trigger condition specified by a trigger component in the question-answering service, the terminal device 400 can send the business problem to the communication server 300. The communication server 300 then transfers the business problem to the operation server 100. The operation server 100 processes the business problem using a function code package component selected by the question-answering service to obtain a question-answering service result for the business problem and returns the question-answering service result to the communication server 300. The communication server 300 then transmits the question-answering service result to the terminal device 400, and the communication application on the terminal device 400 can display the question-answering service result for the business problem input by the user on the question-answering interaction page. The question-answering interaction page may be in the form of a window, a floating layer, or other forms, and is not limited thereto.

[0025] The embodiments of the present application provide various types of function code package components that support independent deployment, realize flexible connection of function code package components, and enable question answering services to be generated based on the process flow connection relationships between the function code package components selected by the user. This allows people without specialized programming knowledge to develop question answering services based on visualized function code package components and process flow connection relationships, rather than developing question answering services by writing code, thereby improving the productivity and versatility of question answering service development and reducing development costs and implementation error rates.

[0026] Referring to Figure 2, Figure 2 is a schematic diagram of a scenario for generating a question answering service provided by an embodiment of the present application, in which a question answering service editing application may be installed on a terminal device 200, and in an editing interaction page 210 of the question answering service editing application, the question answering service editing application may display a set of function code package components in a component candidate page 212, and the set of code package components may include trigger components (which may include, for example, an application trigger component and a form trigger component), question answering strategy components (which may include, for example, a problem determination subcomponent and a frequency control subcomponent), and multiple language question answering model components of different algorithm types, etc.

[0027] A user can click on a selection control 211 to trigger a component selection operation in the question answering service editing application, and drag and drop a function code package component selected on a component candidate page 212 onto a question answering service editing page 213 of the question answering service editing application. The component candidate page is a page that displays function code package components for selection, and the selected function code package component can be presented and edited on the question answering service editing page.

[0028] The question answering service editing application in the terminal device 200 can display the selected trigger component, question answering strategy component, and linguistic question answering model component. For example, the terminal device 200 can display the trigger component (including the application trigger component), the question answering strategy component (including the problem determination subcomponent, the frequency control subcomponent, and the authority determination subcomponent), and the linguistic question answering model component on the edit interaction page 220 of the question answering service editing application.

[0029] Here, the application trigger component may be a starting component that starts the entire question answering service workflow. After the editing of the question answering service editing application is completed, the question answering service is formed. When a user accesses the question answering service, the user inputs a business problem on an independent page of the question answering service, and the terminal device 400 can activate a function provided by the question answering service through the application trigger component and call a question answering strategy component in the question answering service.

[0030] The problem determination subcomponent in the question answering strategy component is used to filter business problems that contain prohibited business words or whose file format is an abnormal problem processing format. Prohibited business words refer to words that are not permitted to be used in business problems, and may be words that violate laws and regulations, or words that consist only of symbols or spaces and cannot express actual linguistic meaning. The abnormal problem processing format is specifically set for the selected language question answering model component. For example, if the selected language question answering model component can only process text messages, the abnormal problem processing format can be set to a business problem file format that the language question answering model cannot process, such as pictures and audio. The frequency control subcomponent in the question answering strategy component is used to filter business problems whose frequency of invoking the question answering service by the same user exceeds a filtering threshold specified by the frequency control subcomponent. The unit time range of the frequency of invoking the question answering service may be one hour, one day, or one week, and is not limited here. The authority determination subcomponent in the question answering strategy component is used to filter business problems whose user's question authority does not fall within the problem authority range.

[0031] The edit interaction page 220 of the question-answering service editing application includes a control display area, in which a connection control 221 is displayed. After the user clicks the connection control 221, the terminal device 200 can trigger a component organizing operation in response to the click on the connection control 221. In response to the component organizing operation, the user can drag the connection control 221 from the application trigger component to the problem determination subcomponent to establish a process flow connection between the application trigger component and the problem determination subcomponent, and determine the problem determination subcomponent as the function code package component of the next flow of the application trigger component. Similarly, at the problem determination subcomponent, the user can use the connection control 221 to determine the frequency control subcomponent as the function code package component of the next flow of the problem determination subcomponent, and similarly, determine the authority determination subcomponent as the function code package component of the frequency control subcomponent and the language question answering model component as the function code package component of the next flow of the authority determination subcomponent, thereby completing the establishment of the process flow connection relationships of the selected function code package components.

[0032] A setting control 222 may be further displayed in the control display area of ​​the edit interaction page 220 of the question-answering service editing application. If the user clicks the setting control 222, the terminal device 200 can trigger a component setting operation in response to the click on the setting control 222. After triggering the component setting operation, the terminal device 200 can display a parameter setting page 223 in the question-answering service editing application in response to the click on the problem determination subcomponent, and set the business forbidden words and problem processing abnormality format of the selected problem determination subcomponent. Setting the business forbidden words and problem processing abnormality format may, for example, be setting the business forbidden words as forbidden word group A and determining the picture format and audio format as the problem processing abnormality format.

[0033] A save control 224 may further be displayed in the control display area of ​​the edit interaction page 220 of the question answer service editing application, and if the user clicks on the save control 224, the terminal device 200 can trigger a question answer service confirmation operation in response to the click on the save control 224, and in response to the question answer service confirmation operation, the terminal device 200 can provide a processing flow connection relationship between the trigger component, question answer strategy component, and linguistic question answer model component selected by the user through the question answer service editing application, thereby generating a question answer service.

[0034] It is understood that in the embodiments of the present application, when the above and below embodiments of the present application are applied to specific products or technologies, user permission or consent must be obtained for relevant user data (e.g., business problems, query frequency, query authority, etc.), and the collection, use, and processing of relevant data must comply with relevant national and regional laws, regulations, and standards.

[0035] 3, which is a schematic diagram of a scenario for generating a question-answering service provided by an embodiment of the present application, in which a communication application may be installed in a terminal device 400, and the terminal device 400 can access the question-answering service through the communication application. In a question-answering interaction page 410 of the question-answering service editing application, the terminal device 400 can display the question-answering interaction page in the communication application, and a user can input a business problem A on the question-answering interaction page. The business problem A includes a trigger parameter. If the trigger parameter satisfies a question-answering trigger condition indicated by a trigger component in the question-answering service (which may be to call the question-answering service when a user inputs a business problem on an independent page of the question-answering service), the trigger component will send the business problem A to the communication server. 300, the communication server 300 transfers the business problem A to the work server 100, the work server 100 processes the business problem A using the function code package component selected by the question answering service to obtain the question answering service result for the business problem A and returns the question answering service result to the communication server 300, the communication server 300 sends the question answering service result to the terminal device 400, and as shown on the question answering interaction page 420 of the question answering service editing application, the communication application on the terminal device 400 can display the question answering service result A for the business problem A entered by the user on the question answering interaction page.

[0036] Referring to Fig. 4, Fig. 4 is a flowchart of a method for generating a question-answering service provided by an embodiment of the present application, which may be performed by a computer device, which may be the terminal device 200 as shown in Fig. 1 or the operation server 100. As shown in Fig. 4, the method may include the following steps:

[0037] Step S101: displaying a function code package component set in an edit interaction page of a question answering service editing application, the function code package component set including a plurality of function code package components, the plurality of function code package components including at least one trigger component, at least one question answering strategy component, and at least one linguistic question answering model component; The computing device can display a collection of functional code package components in a question answering service editing application, the collection of functional code package components including one or more trigger components, one or more question answering strategy components, and one or more linguistic question answering model components.

[0038] Specifically, the computer device may display a function code package component set on an edit interaction page of a question-answering service editing application, where the function code package set may include multiple function code package components. The multiple function code package components may include at least one trigger component, at least one question answering strategy component, and at least one linguistic question answering model component. The function code package component set may further include other function code package components, such as a function code package component used for message notification, a function code package component that calls API interfaces of other platforms, a function code package component that stops the flow, and a function code package component that performs no operation. The embodiments of the present application are not limited thereto. A no operation is an operation that does not perform actual processing. A computer program needs to know what operation needs to be performed, and can use no operation to represent when no operation is required.

[0039] Here, a function code package component refers to a component having a function function, predefined actions, and attributes. The attributes may be the internal code logic, name, parameters, return values, etc., of the function code package component. The function function may be implemented in the function code package component by using various prewritten program modules and libraries. The trigger component can be used to detect whether a business problem proposed by a user meets the question answer trigger conditions, and when the problem meets the question answer trigger conditions, it can invoke a question answer strategy component in the question answer service. The question answer strategy component is a component that filters business problems and can filter business problems that cannot be processed by the linguistic question answer model component. The linguistic question answer model component can perform semantic identification on the filtered business problems and output the question answer service results for the business problems.

[0040] The problem trigger conditions indicated by different trigger components may be different. For example, the trigger components may include an application trigger component, a form trigger component, a mention trigger component, a cut-in trigger component, and a periodic trigger component. Here, the problem trigger condition of the application trigger component may be to call the question answering service by a user entering a business problem on a single chat page. The problem trigger condition of the form trigger component may be to call the question answering service by a user submitting a form. The problem trigger condition of the mention trigger component may be to call the question answering service by a user mentioning (which may be an at sign, @) the question answering service (or a specific name that characterizes the question answering service) in a chat. The problem trigger condition of the cut-in trigger component may be to call the question answering service by a user cutting in from a chat page to the single chat page of the question answering service. The problem trigger condition of the periodic trigger component may be to automatically call the question answering service based on time periods, task (event) periods, etc., without the user needing to actively call it. For example, the question answering strategy component in the question answering service may be periodically called every day. The problem trigger conditions of the trigger component can be determined by different functional functions and specific workflows, and the embodiments of the present application are not limited here.

[0041] Different question-answering strategy components filter business questions and answers in different ways. For example, a question-answering strategy component may include a problem determination subcomponent, a frequency control subcomponent, and an authority determination subcomponent. The problem determination subcomponent may filter business questions that contain prohibited business words or whose file format is an abnormal problem processing format. The prohibited business words refer to words that are not permitted to be used in business questions, and may be words that violate laws and regulations, or words that consist only of symbols or spaces and cannot express actual linguistic meanings. The abnormal problem processing format is specifically set for the selected language question-answering model component. For example, if the selected language question-answering model component can only process text messages, the abnormal problem processing format may be set to a business question file format that the language question-answering model cannot process, such as pictures and audio. The frequency control subcomponent is used to filter business questions whose frequency of invoking the question-answering service by the same user exceeds a filtering threshold specified by the frequency control subcomponent. The unit time range of the frequency of invoking the question-answering service may be one hour, one day, or one week, but is not limited thereto. The authority determination subcomponent is used to filter out business problems whose user's question authority does not belong to the problem authority range.

[0042] As can be seen, the question answering strategy component can filter business problems through various logical judgments and send the flow in different directions, for example, business problems in which the user's question authority in the authority judgment subcomponent belongs to the problem authority range can flow to the linguistic question answering model component, and business problems in which the user's question authority does not belong to the problem authority range can flow to the function code package component that performs no operation and the function code package component that notifies the message ``No question authority''.

[0043] The algorithm types of the language question answering models assembled in different language question answering model components may be different, and the algorithm types may refer to the training text sets of the language question answering models from different vertical fields, for example, the vertical fields may be e-commerce customer service, communication chat assistant, and intelligent education question answering, etc. The initial language models adopted may be different, and the training methods may be different, for example, recurrent neural networks (RNNs) and long short-term memory networks (LSTMs), etc., and the embodiments of the present application are not limited thereto.

[0044] Here, the training process of the language question answer model assembled in the language question answer model component may be as follows: obtain a training text set, filter business stop words in the training text set, perform text segmentation on the filtered training text set to obtain M text segments, calculate the occurrence counts of each text segment in the M text segments, and obtain text frequencies corresponding to each of the M text segments, where M is a positive integer; obtain an initial language model, create a text dictionary for the initial language model based on the M text segments and the M text frequencies, where the text dictionary includes word vectors corresponding to each of the M text segments; obtain question answer samples, construct question answer word vectors corresponding to the question answer samples using the text dictionary, and train the initial language model using the question answer word vectors, thereby obtaining a language question answer model component.

[0045] Specifically, the training process of the language question answer model in the language question answer model component may be as follows: obtain a training text set, filter business stop words in the training text set, perform text segmentation on the filtered training text set to obtain M text segments, calculate the occurrence counts of the M text segments, and obtain text frequencies corresponding to each of the M text segments, where M is a positive integer; obtain an initial language model, create a text dictionary for the initial language model based on the M text segments and the M text frequencies, the text dictionary including word vectors corresponding to each of the M text segments; obtain question answer samples, construct question answer word vectors corresponding to the question answer samples according to the text dictionary, and adjust model parameters of the initial language model according to the question answer word vectors; and obtain the language question answer model component when the initial language model satisfies a model convergence condition.

[0046] Specifically, the computer device can obtain a training text set, which may include normal natural expectations (such as web page text, news text, and novel text), and may further include technical corpora for business needs (such as educational text), and the computer device can perform data cleaning on the training text set, re-inspect and verify the training text set, and remove training texts with duplicated, missing values, and invalid values, so as to obtain a training text set with data consistency.

[0047] The computer device can remove business stop words from the training text set after data cleaning. Business stop words refer to words or punctuation marks that are commonly found in natural language text but have no specific meaning in semantic analysis. For example, in Chinese text, business stop words may include "de", "le", "ye", "in", and so on. In English text, business stop words may include "the", "a", "an", and "in", and so on.

[0048] A computer device can perform text segmentation (also known as word segmentation) on a training text set from which business stop words have been removed. The text segmentation can be performed by rule matching or probability matching. Take the example of performing text segmentation on training text C in training text set A using probability matching. For text segmentation of training text C, a word segmentation scheme S will be provided. The task of text segmentation is to find a scheme in word segmentation scheme S that maximizes the probability P(S|C) that a segmented word will appear in training text set A. According to Bayes' formula, the P(S|C) probability can be calculated as shown in formula (1):

number

number

[0049] The computer device counts the number of occurrences of the text segment W after text segmentation in the training text set A of i and can obtain the text frequencies corresponding to each of the M text segments. The text frequency may be the probability of the text segment in the training text set A, that is, P(W i ).

[0050] The computer device can create a text dictionary for the initial language model based on the M text segments and the M text frequencies. Here, the text dictionary may include word vectors corresponding to each of the M text segments. The word vector is to convert text data into a numerical representation form acceptable to the model. The position of the text segment W i in the training text C, the number of the text segment W i and the restored word form of the text segment W i may be further included. The restored word form refers to restoring the text segment to its original form considering the context of the text segment and the part-of-speech. For example, both "吃飯了" and "吃了飯" are restored to "吃飯".

[0051] The initial language model may be a neural network model, such as a recurrent neural network and a long short-term memory network. Here, the initial language model may consist of a plurality of hidden layers and one output layer.

[0052] The computer device can set the size of the hidden layer of the initial language model, the learning rate, etc. Here, the learning rate refers to the gradient descent step size determined according to the model convergence condition and affects the model convergence speed. The gradient refers to adjusting the parameter values ​​of the hidden layer in the initial language model in the gradient descent algorithm.

[0053] The computer device can convert the word vectors in the text dictionary into question-answering training sequences and question-answering target sequences of the initial language model, and the question-answering training sequences and question-answering target sequences are question-answering samples.

[0054] The computer device inputs a question-answering training sequence into an initial language model and trains it according to a predefined loss function, so that the initial language model outputs a question-answering prediction result in an output layer, where the loss function may be a cross-entropy loss function, a mean squared error loss function, etc., but is not limited thereto.

[0055] The computer device can minimize the loss function by adjusting parameter values ​​in the hidden layer of the initial language model according to the calculated gradient, and a language question-answering model component is obtained when a predetermined model convergence condition is satisfied between the question-answering prediction result of the initial language model and the question-answering target sequence. Here, the model convergence condition refers to a performance metric between the question-answering prediction result and the question-answering target sequence being lower than a predetermined value, and the performance metric may include accuracy, precision, recall, etc.

[0056] As can be seen, in the training process of the initial language model, techniques such as dropout (dropout is a type of hidden layer in a neural network model, which randomly sets some neurons to 0 during the training process, thereby reducing the risk of overfitting of the neural network model) and batch normalization (batch normalization is performed on each small batch of input to make the mean value of each feature in the training data closer to 0 and the variance closer to 1, thereby speeding up the training process and improving the stability and generalization ability of the neural network model) can be used to improve the training effect of the model.

[0057] Step S102: In response to a component selection operation for a set of function code package components, a question answering service editing page included in the editing interaction page displays the trigger component, the question answering strategy component, and the linguistic question answering model component selected by the component selection operation, and the question answering service editing page is further used to display the processing flow connection relationships between the selected trigger component, the question answering strategy component, and the linguistic question answering model component.

[0058] Specifically, and also referring to Figure 5, Figure 5 is a schematic diagram of a scenario for generating a question answering service provided by an embodiment of the present application, and as shown in Figure 5, the question answering service editing application may mainly include components such as a stream organization canvas, an application selector, an image rendering engine, an application parameter configurator, a debugger, and a version controller.

[0059] Here, the stream organization canvas can be used to capture the display effects of function code package components. The application selector can position the function code package components on the question-and-answer service editing page when dragging and dropping them, ensuring that the user selects the correct function code package component. The graphics rendering engine can render graphical representations of the function code package components and the line connections (process flow connections) between them. Round modules can be used to represent trigger components, diamond modules mainly represent question-and-answer strategy components, and square modules represent specific application actions, such as sending messages, calling other platform API interfaces, stopping the flow, and no operation. The application parameter configurator can configure graphical representations of function code components and input related configuration information, such as the instance template identifier described below, in the application parameter configurator. The debugger can run the question-and-answer service and generate the execution log table described below, which can then use the log information to statistically record the operation status of the question-and-answer service and quantify service quality. The version controller can store different versions of the question and answer service, realize switching between different versions of the question and answer service through the version controller, and record information of each version change.

[0060] In response to a component selection operation (which may be performed using the application selector shown in FIG. 5) on a set of function code package components of the question and answer service editing application, the computer device may drag and drop the selected function code package component on the question and answer service editing page of the question and answer service editing application, or may first select the required function code package component and then directly establish the selected function code package component at a specified position on the question and answer service editing page of the question and answer service editing application.

[0061] In response to the component organization operation of the question answering service editing application, the computer device can establish a process flow connection relationship between the function code package components by a line connection, for example, establish a process flow connection relationship between a trigger component and a question answering strategy component, connect a line to the question answering strategy component at the trigger component, and determine the question answering strategy component as the function code package component in the next flow of the trigger component.

[0062] In one embodiment, the method for generating a question answering service further includes: establishing a process flow connection relationship between a selected trigger component and a question answering strategy component in response to a first component organization operation on a function code package component selected on a question answering service edit page; determining the selected trigger component as a function code package component to be used for executing a question answering service initiation flow based on the process flow connection relationship between the selected trigger component and the question answering strategy component, and determining the selected question answering strategy component as a function code package component to be used for executing a next flow of the trigger component; establishing a process flow connection relationship between the selected question answering strategy component and a language question answering model component in response to a second component organization operation on the function code package component selected on the question answering service edit page; and determining the selected language question answering model component as a function code package component to be used for executing a next flow of the question answering strategy component based on the process flow connection relationship between the selected question answering strategy component and the language question answering model component.

[0063] Optionally, the computer device may respond with a first component organization operation between a trigger component and a question answering strategy component, and a second component organization operation between a question answering strategy component and a linguistic question answering model component, and the organization process may be as follows: In response to a first component organization operation on a function code package component selected on a question answering service edit page of the question answering service editing application, a process flow connection relationship is established between the selected trigger component and a question answering strategy component, and based on the process flow connection relationship between the selected trigger component and the question answering strategy component, the selected trigger component is determined as the function code package component to be used for executing a question answering service initiation flow and the selected question answering strategy component is determined as the function code package component to be used for executing a next flow of the trigger component; and in response to a second component organization operation on the function code package component selected on the question answering service edit page of the question answering service editing application, a process flow connection relationship is established between the selected question answering strategy component and a language question answering model component, and based on the process flow connection relationship between the selected question answering strategy component and the language question answering model component, the selected language question answering model component is determined as the function code package component to be used for executing a next flow of the question answering strategy component.

[0064] Specifically, the computer device can respond to a first component organization operation between a trigger component and a question answering strategy component selected on a question answering service editing page of a question answering service editing application, connect a line to the trigger component to the question answering strategy component, establish a processing flow connection relationship between the selected trigger component and the question answering strategy component, determine the selected trigger component as a function code package component to be used for executing the question answering service initiation flow, and determine the selected question answering strategy component as a function code package component to be used for executing the next flow of the trigger component.

[0065] Similarly, the computer device may further respond to a second component organization operation between a selected question answering strategy component and a language question answering model component on a question answering service edit page of the question answering service edit application, connect a line to the language question answering model component at the question answering strategy component, establish a process flow connection relationship between the selected question answering strategy component and the language question answering model component, and determine the selected language question answering model component as a function code package component used to execute the next flow of the question answering strategy component, where the question answering strategy component may include a problem determination subcomponent, a frequency control subcomponent, and an authority determination subcomponent.

[0066] In one embodiment, the question answering strategy component includes a problem determination subcomponent, a frequency control subcomponent, and an authority determination subcomponent, and the method for generating a question answering service includes the steps of: in response to a third component organization operation for the problem determination subcomponent and the frequency control subcomponent selected on the question answering service editing page, establishing a process flow connection relationship between the selected problem determination subcomponent and the frequency control subcomponent; and determining the selected problem determination subcomponent as a function code package component used to execute the question answering strategy initiation flow based on the process flow connection relationship between the selected problem determination subcomponent and the frequency control subcomponent; The method further includes determining the frequency control subcomponent as a function code package component used to execute the next flow of the question determination subcomponent, establishing a processing flow connection relationship between the selected frequency control subcomponent and the authority determination subcomponent in response to a fourth component organization operation for the frequency control subcomponent and the authority determination subcomponent selected on the question answering service editing page, and determining the selected authority determination subcomponent as a function code package component used to execute the next flow of the frequency control subcomponent based on the processing flow connection relationship between the selected frequency control subcomponent and the authority determination subcomponent.

[0067] Optionally, the organizing process for the problem determination subcomponent, the frequency control subcomponent, and the authority determination subcomponent of the computer device may be as follows: in response to a third component organizing operation for the problem determination subcomponent and the frequency control subcomponent selected on a question answering service edit page of the question answering service editing application, establish a process flow connection relationship between the selected problem determination subcomponent and the frequency control subcomponent, determine the selected problem determination subcomponent as a function code package component used to execute a question answering strategy initiation flow based on the process flow connection relationship between the selected problem determination subcomponent and the frequency control subcomponent, and determine the selected frequency control subcomponent as a function code package component used to execute a next flow of the problem determination subcomponent; in response to a fourth component organizing operation for the frequency control subcomponent and the authority determination subcomponent selected on the question answering service edit page of the question answering service editing application, establish a process flow connection relationship between the selected frequency control subcomponent and the authority determination subcomponent, and determine the selected authority determination subcomponent as a function code package component used to execute a next flow of the frequency control subcomponent based on the process flow connection relationship between the selected frequency control subcomponent and the authority determination subcomponent.

[0068] Specifically, the computer device can respond to a third component organization operation between a selected problem determination subcomponent and a frequency control subcomponent on the question answering service editing page of the question answering service editing application, connect a line from the problem determination subcomponent to the frequency control subcomponent, establish a processing flow connection relationship between the selected problem determination subcomponent and the frequency control subcomponent, determine the selected problem determination subcomponent as the function code package component used to execute the question answering strategy initiation flow, and determine the selected frequency control subcomponent as the function code package component used to execute the next flow of the problem determination subcomponent.

[0069] Similarly, the computer device can respond to a fourth component organization operation between a frequency control subcomponent and an authority determination subcomponent selected on the question answer service edit page of the question answer service edit application, connect a line to the frequency control subcomponent at the problem determination subcomponent, establish a processing flow connection relationship between the selected frequency control subcomponent and the authority determination subcomponent, and determine the selected authority determination subcomponent as the function code package component to be used in executing the next flow of the frequency control subcomponent.

[0070] The question answering service editing application on the computer device can display the selected trigger components, question answering strategy components, and linguistic question answering model components on a question answering service editing page, and can display process flow connection relationships between the selected trigger components, question answering strategy components, and linguistic question answering model components.

[0071] In one embodiment, the method for generating a question-answering service further includes the steps of: displaying a plurality of instance template identifiers on a parameter setting page in the edit interaction page in response to a component setting operation for a function code package component selected on the question-answering service edit page; determining a target instance template identifier from the plurality of instance template identifiers; and updating the selected function code package component on the question-answering service edit page based on the target instance template identifier, wherein the instance template identifier is used to determine a function function corresponding to the function code package component, and the function function is used to functionally process the business problem.

[0072] Optionally, the computer device may respond with a component setting operation for a function code package component, and the setting process may be as follows: in response to a component setting operation for a function code package component selected on a question answer service editing page of the question answer service editing application, display a plurality of instance template identifiers on a parameter setting page of the question answer service editing application, determine a target instance template identifier from the plurality of instance template identifiers, the plurality of instance template identifiers including a target instance template identifier, and update the selected function code package component on the question answer service editing page of the question answer service editing application based on the target instance template identifier, the instance template identifier is used to determine a function function corresponding to the function code package component, and the function function is used to functionally process the business problem.

[0073] Specifically, in response to a component setting operation for a function code package component (which may be performed using the application parameter configurator shown in FIG. 5 ), the computer device can display multiple instance template identifiers on the parameter setting page of the question-answering service editing application. The instance template identifier can be used to determine a functional function corresponding to the function code package component, and the function code package component processes a business problem using the functional function. The computer device can select a desired instance template identifier from the multiple instance template identifiers and update the function code package component with the selected instance template identifier. For example, for a periodic trigger component, a timer trigger template identifier or an event trigger template identifier can be selected. For example, for a periodic trigger component, if the timer trigger template identifier is selected and the instance parameter is modified to 86,400 seconds, i.e., 1 day, the question-answering service will be triggered periodically every day. If the event trigger template identifier is selected and the instance parameter (event) is modified to indicate that the temperature exceeds 36°C, the question-answering service will be automatically triggered when the temperature exceeds 36°C.

[0074] Optionally, the computer device can further set an instance name of a function code package component, an instance name of a previous flow node, an instance name of a next flow node, and instance parameters, etc., on a parameter setting page of the question answering service editing application, where the instance name and the instance name of the next flow node can jointly characterize a processing flow connection relationship, and the instance parameters may be auxiliary parameters of a function function corresponding to an instance template identifier.

[0075] The computer device can modify the process flow connection relationship between the function code package component and the function code package component of the previous flow node or the function code package component of the next flow node by modifying the instance name of the previous flow node or the instance name of the next flow node. The computer device can modify the instance parameters. For example, in a periodic trigger machine, if a timer trigger template identifier is selected and the instance parameter is modified to 86400 seconds, i.e., 1 day, the question answering service will be triggered periodically every 1 day.

[0076] The computer device can update the function code package component with the selected instance template identifier (or the determined instance name, the instance name of the selected previous flow node, the instance name of the selected next flow node, and the determined instance parameters, etc.) on the question answer service edit page of the question answer service edit application.

[0077] In one embodiment, the method for generating a question-answering service further includes the steps of: setting business prohibited words and a problem processing abnormality format for the selected question-answering subcomponent on a parameter setting page in the edit interaction page in response to a first component setting operation for the selected question-answering subcomponent on the question-answering service edit page; setting a frequency threshold for the selected frequency control subcomponent on the parameter setting page in response to a second component setting operation for the selected frequency control subcomponent on the question-answering service edit page; and setting a problem authority range for the selected authority judgment subcomponent on the parameter setting page in response to a third component setting operation for the authority judgment subcomponent on the question-answering service edit page.

[0078] The computing device may selectively respond with a first component setting operation for a problem determination subcomponent in the question answering strategy component, a second component setting operation for a frequency control subcomponent, and a third component setting operation for an authority determination subcomponent, and the setting process may be as follows: in response to the first component setting operation for a problem determination subcomponent selected on a question answering service editing page of the question answering service editing application, business prohibited words and a problem processing abnormality format for the selected problem determination subcomponent are set on a parameter setting page of the question answering service editing application; in response to the second component setting operation for a frequency control subcomponent selected on the question answering service editing page of the question answering service editing application, a frequency threshold for the selected frequency control subcomponent is set on the parameter setting page; and in response to the third component setting operation for the authority determination subcomponent selected on the question answering service editing page of the question answering service editing application, a problem authority range for the selected authority determination subcomponent is set on the parameter setting page.

[0079] Specifically, in response to a first component setting operation for a problem determination subcomponent selected on a question answer service editing page of a question answer service editing application, the computer device can set business prohibited words and problem processing abnormality formats for the selected problem determination subcomponent on a parameter setting page of the question answer service editing application.

[0080] Since the question answering service requires a relatively high computational cost, the computer device can perform a validity judgment on the business problem entered by the user once through the problem judgment subcomponent, and choose to filter out some typical invalid problems, for example, business problems that contain business prohibited words or whose file format is an abnormal problem processing format, and can introduce the filtered out business problems without any processing and without any operation.

[0081] Here, the problem processing abnormality format is specifically set for the selected language question answering model component. For example, if the selected language question answering model component can only process text messages, the problem processing abnormality format can be set to a business problem file format that the language question answering model cannot process, such as picture and audio. Business prohibited words may refer to words that are not allowed to be used in business problems, may be words that violate laws and regulations, or may be customized words.

[0082] Optionally, the problem determination subcomponent can further filter business problems that only contain trigger parameters (e.g., in the cut-in trigger component, a user cuts in to a single chat page but does not enter a business problem), and other business problems that do not have business problem content (e.g., an empty message where the business problem content is missing during transmission), and business problems that do not have actual natural semantics, such as symbols, emoticons (e.g., / s in emoji), spaces, and special characters (e.g., color values ​​#66ccff, markup language

[0083]

[0084]

[0085]

[0086]

[0087]

[0088]

[0089]

[0090]

[0091]

[0092]

[0093]

[0094]

[0095]

[0096]

[0097]

[0098]

[0099]

[0100]

[0101]

[0102]

[0103]

[0104]

[0105]

[0106]

[0107]

[0108]

[0109]

[0110]

[0111]

[0112]

[0113]

[0114]

[0115]

[0116]

[0117]

[0118]

[0119]

[0120]

[0121]

[0122]

[0123]

[0124]

[0125]

[0126]

[0127]

[0128]

[0129]

[0130]

[0131]

[0132]

[0133]

[0134]

[0135]

[0136]

[0137]

[0138]

[0139]

[0140]

[0141]

[0142]

[0143]

[0144]

[0145]

[0146]

[0147]

[0148]

[0149]

[0150]

[0151]

[0152]

[0153]

[0154]

[0155]

[0156]

[0157]

[0158]

[0159]

[0160]

[0161]

[0162]

[0163]

[0164]

[0165]

[0166]

[0167]

[0168]

[0169]

[0170]

[0171]

[0172]

[0173]

[0174]

[0175]

[0176]

[0177]

[0178]

[0179]

[0180]

[0181]

[0182]

[0183]

[0184]

[0185]

[0186]

[0187]

[0188]

[0189]

[0190]

[0191]

[0192]

[0193]

[0194]

[0195]

[0196]

[0197]

[0198]

[0199]

[0200]

[0201]

[0202]

[0203]

[0204]

[0205]

[0206]

[0207] In response to a second component setting operation for the frequency control subcomponent selected on the question-answering service editing page of the question-answering service editing application, the computer device sets a frequency threshold for the selected frequency control subcomponent on the parameter setting page to control the user's question frequency (frequency of invoking the question-answering service), and filters out business problems whose question frequency by the same user exceeds the filtering threshold set by the frequency control subcomponent. Here, the unit time range for the question frequency may be one hour, one day, or one week, and is not limited here. For example, the unit time range may be set to 86,400 seconds, i.e., one day. The number of questions is set to 100. In other words, the same user is allowed to invoke the question-answering service 100 times each day, and after the user invokes the question-answering service more than 100 times, subsequent business problems are filtered out. In response to a third component setting operation for the authority determination subcomponent selected on the question and answer service editing page of the question and answer service editing application, the computer device can set a question authority range for the selected authority determination subcomponent on the parameter setting page, and filter out business questions whose user's question authority does not belong to the question authority range. The computer device can compare the user who invokes the question and answer service with the built-in user list. If the user who invokes the question and answer service is not in the built-in user list, the computer device filters out the business questions of the user who invokes the question and answer service and returns a reminder that the user does not have the authority to ask a question. Step S103: In response to a question and answer service confirmation operation triggered on the edit interaction page of the question and answer service editing application, a question and answer service is generated based on the process flow connection relationships among the selected trigger component, the question and answer strategy component, and the linguistic question and answer model component, wherein the trigger component in the question and answer service is used to invoke the question and answer strategy component in the question and answer service, and the question and answer strategy component in the question and answer service isAfter obtaining a business problem, the computer device performs question-and-answer filtering on the business problem, and the model component in the question-and-answer service is used to output a question-and-answer service result according to the filtered business problem. Specifically, in response to a question-and-answer service confirmation operation of the question-and-answer service editing application, the computer device can generate a question-and-answer service according to the process flow connection relationship between the selected trigger component, question-and-answer strategy component, and linguistic question-and-answer model component. The question-and-answer service refers to a service that generates a question answer using a linguistic question-and-answer model. The generated question-and-answer service can be directly connected and used in applications supported by the question-and-answer service editing application. For example, if the question-and-answer service editing application supports directly embedding a question-and-answer service in a communication application, the question-and-answer service can be accessed in the communication application. In one embodiment, the method for generating a question-answering service further includes the steps of: establishing an execution log table for the question-answering service based on the selected trigger component, question-answering strategy component, and language question-answering model component; recording log information of function code package components in the question-answering service using the execution log table; calculating statistics of operation states and service quality quantification data for the question-answering service according to the log information; and adjusting at least one of the trigger component, question-answering strategy component, language question-answering model component, or process flow connection relationship in the question-answering service based on the operation state and service quality quantification data of the question-answering service. Optionally, the computer device can display the log information during the operation process of the question-answering service and adjust the question-answering service according to the log information. The step of displaying the log information during the operation process of the question-answering service may be as follows: establishing an execution log table for the question-answering service based on the selected trigger component, question-answering strategy component, and language question-answering model component;The execution log table records log information of the function code package components in the question answering service, displays the log information during the operation process of the question answering service, calculates statistics on the operation status and service quality quantification data for the question answering service according to the log information, and adjusts at least one of the trigger component, the question answering strategy component, the language question answering model component, or the process flow connection relationship in the question answering service based on the operation status and service quality quantification data of the question answering service. Specifically, the computer device can establish an execution log table for the question answering service according to the selected trigger component, the question answering strategy component, and the language question answering model component. The execution log table can record log information of the function code package components in the question answering service, and the execution log table can further include a log grade and a log content classification, and the log grade can be as shown in Table 1. [Table 1] The log content classification can be as shown in Table 2. [Table 2] The computer device can display the log information during the operation process of the question answering service using the execution log table, and calculate statistics on the operation status and service quality quantification data for the question answering service according to the log information. For example, a user may reconfigure a periodic trigger component of a question answering service and update the selected timer trigger template identifier to an event trigger template identifier, but may not set a triggered event (i.e., an instance parameter corresponding to the event trigger template identifier), causing the periodic trigger component to be unable to call the question answering service. In this case, the execution log table may record that the log level of the log information is ERROR, the classification of the log content is a configuration log, and may record the user's correction actions in detail. Here, the service quality quantification data may be, but is not limited to, the occurrence rate of the error log level and the error rate of the question answering service results. The computer device may monitor the operation status of the question answering service,The service quality quantification data can be used to adjust the trigger component, question answering strategy component, linguistic question answering model component, process flow connection relationships between the above components, instance template identifiers, instance parameters, etc. in the question answering service, thereby reducing work errors and manual intervention and improving the accuracy and consistency of the question answering service. For example, if the operating status of the question answering service displays that the load is too high, the frequency threshold of the frequency control subcomponent in the question answering strategy component can be adjusted and improved by increasing the frequency threshold, thereby reducing the operating load of the question answering service and improving the service quality quantification data of the question answering service. As can be understood, the embodiment of the present application provides various types of function code package components that support independent deployment, realizes flexible connection of function code package components, and can generate a question answering service according to the process flow connection relationships between the function code package components selected by the user. A question answering service generally includes three categories: trigger components, question answering strategy components, and linguistic question answering model components. Instead of adopting fixed function code package components and their process flow relationships, a user can flexibly select required function code package components according to business needs and adjust the process flow connection relationships therebetween to generate a question answering service that meets business needs. In the embodiment of the present application, a function code package component set is displayed in a question answering service editing application, and the function code package component set includes at least one trigger component, at least one question answering strategy component, and at least one linguistic question answering model component. In response to a component selection operation for the function code package component set in the question answering service editing application, the selected trigger component, question answering strategy component,and a language question answering model component, and the question answering service editing page is further used to display the process flow connection relationships between the selected trigger component, question answering strategy component, and language question answering model component. In response to a question answer service confirmation operation in the question answering service editing application, a question answering service is generated based on the process flow connection relationships between the selected trigger component, question answering strategy component, and language question answering model component, wherein the trigger component in the question answering service is used to invoke the question answering service, the question answering strategy component in the question answering service is used to perform question answer filtering for the business problem, and the model component in the question answering service is used to output the question answer service result according to the filtered business problem. This provides various types of function code package components that support independent deployment, realizes flexible connection of function code package components, and allows question answering services to be generated based on the process flow connection relationships between function code package components selected by the user. This allows people without specialized programming knowledge to develop question answering services using the visualized function code package components and process flow connection relationships rather than by writing code, thereby improving the productivity and versatility of question answering service development and reducing development costs and implementation error rates. In one embodiment, the process of accessing a question answering service in a communication application may be specifically as follows: displaying a question answering interaction page of the question answering service in the communication application, obtaining a business problem input on the question answering interaction page, the business problem including a trigger parameter, if the trigger parameter satisfies a question answering trigger condition indicated by the trigger component, invoking a question answering strategy component in the question answering service by the trigger component, and filtering the business problem through the question answering strategy component in the question answering service;The filtered business problem is obtained, and question and answer identification is performed for the filtered business problem using the linguistic question and answer model component in the question and answer service, and the question and answer service result for the business problem is output and displayed on the question and answer interaction page. In an embodiment of the present application, a pre-compiled function code package component is used to realize access to the question and answer service in a communication application, which is efficient, convenient, has low development costs, and has a low implementation error rate. In one embodiment, the method for generating a question-answering service further includes a step of acquiring an execution record table of the question-answering service, the execution record table including an instance template identifier. The step of performing question-and-answer identification on the filtered business problem using a language question-and-answer model component in the question-and-answering service and outputting a question-and-answering service result includes a step of determining the filtered business problem as an output data variable of a trigger component and determining the output data variable of the trigger component as an input data variable of a question-and-answering strategy component; a step of determining a functional function corresponding to the question-and-answering strategy component based on the instance template identifier, and processing the input data variable of the question-and-answering strategy component using the functional function corresponding to the question-and-answering strategy component to obtain the output data variable of the question-and-answering strategy component; a step of determining the output data variable of the question-and-answering strategy component as an input data variable of the language question-and-answering model component; a step of determining a functional function corresponding to the language question-and-answering model component based on the instance template identifier; Please refer to FIG. 6, which is a flowchart of accessing a question and answer service in a communication application provided by an embodiment of the present application,Accessing a question-answering service in a communication application is realized by a computer device, which may be the terminal device 400 or the communication server 300 shown in FIG. 1. As shown in FIG. 6, the method may include the following steps: Step S201: Obtain an execution record table for the question-answering service, where the execution record table includes an instance template identifier. Specifically, after completing the editing of the question-answering service, the computer device can establish the execution record table for the question-answering service according to component information in the trigger component, question-answering strategy component, and language question-answering model component selected by the question-answering service, i.e., save the DSL JSON structure represented by the function code package component in the background database. When the trigger component invokes the question-answering strategy component, the computer device can complete the execution of the question-answering service while reading the execution record table. Here, the component information may include the instance name, the instance name of the next flow node, instance parameters, the instance name of the previous flow node, the instance name of the instance template identifier, and the instance name of the next flow node, or the instance name of the previous flow node may characterize a process flow connection relationship. The question answering strategy component may include a problem determination subcomponent, a frequency control subcomponent, and an authority determination subcomponent. DSL Json is a fast and lightweight JSON parsing library. It supports basic JSON types, such as numbers, strings, arrays, and objects. For example, the records for the periodic trigger component in the execution record table may be as shown in Table 3. [Table 3] The computer device can obtain the input data variables and output data variables of the function function corresponding to the function code package component through the instance name in the execution record table, and when calling the question answering service, it can use the execution record table to set the input data variables and output data variables to the corresponding data (e.g.,The input data variable refers to an input in the processing flow of the function code package component, and the output data variable refers to an output in the processing flow of the function code package component. Step S202: Display a question-answering interaction page of a question-answering service in a communication application, and obtain a business problem entered in the question-answering interaction page. The business problem includes a trigger parameter. If the trigger parameter satisfies a question-answering trigger condition specified by the trigger component, invoke a question-answering strategy component in the question-answering service via the trigger component. The computer device displays a question-answering interaction page in a communication application, and can enter a business problem in the question-answering interaction page. The business problem includes a trigger parameter. If the trigger parameter satisfies a question-answering trigger condition specified by the trigger component, invoke a question-answering strategy component in the question-answering service via the trigger component. Specifically, the computer device can display a question-answering interaction page in a communication application, and a user can enter a business problem in the question-answering interaction page. Here, the business problem may include a trigger parameter. The trigger parameter refers to a parameter possessed by the user when submitting the business problem. The business problem may include a prompt word, which may be a word or a sentence designed according to the user's needs and the application scenario. For example, the business problem "What should I eat today?" may include prompt words such as "lunch," "Chinese food," and "takeout." The prompt word is the restored word form (including context information and semantic information) used when training the initial language model in step S101 of the embodiment of FIG. 4. If the trigger parameter satisfies the question-answering trigger condition specified by the trigger component in the question-answering service, the computer device activates a function provided by the question-answering service by the trigger component,and can invoke a question answering strategy component in the question answering service. Here, the question answering service refers to a service that generates a question answer according to a linguistic question answering model. Optionally, the computer device can encrypt a business problem that invokes the question answering strategy component in the question answering service and start the start flow of the question answering service, which process can be as follows: display a question answering interaction page of the question answering service in a communication application, obtain a business problem entered on the question answering interaction page, the business problem includes trigger parameters, and if the trigger parameters satisfy the question answer trigger condition specified by the trigger component, obtain a message certificate through the trigger component and send the business problem and the message certificate to a communication server, whereby the communication server encrypts the business problem according to the message certificate, obtains an encrypted ciphertext, and sends the encrypted ciphertext to a working server, where the message certificate is further stored in the working server, and the working server decrypts the encrypted ciphertext using the stored message certificate to obtain the business problem, which is further used by the working server to output a question answering service result corresponding to the business problem via the question answering service. Specifically, the computer device can encrypt a business problem that invokes a question answering strategy component in the question answering service and start the initiation flow of the question answering service, the process may be as follows: displaying a question answering interaction page in a communication application, inputting a business problem on the question answering interaction page, the business problem including a trigger parameter, if the trigger parameter satisfies the question answering trigger condition indicated by the trigger component, sending the business problem and a message certificate to a communication server corresponding to the communication application by the trigger component, whereby the communication server encrypts the business problem according to the message certificate to obtain an encrypted ciphertext, and sending the encrypted ciphertext to an operation server corresponding to the question answering service editing application, the message certificate is further stored in the operation server, and the operation server decrypts the ciphertext encrypted by the message certificate;The operation server is further used to obtain a business problem through a question-and-answer service and output a question-and-answer service result corresponding to the business problem through the question-and-answer service. Specifically, the operation server packages the question-and-answer service and provides an interface to the outside world to invoke the question-and-answer service, for example, by constructing a network request (HTTP request). The computer device can write the business problem in the request body (Body part) of the network request packet, and the operation server returns the final question-and-answer service result to the computer device by passing a packet (HTTP Response). The computer device can create a question-and-answer service in a communication application and obtain a message evidence of the question-and-answer service, where the message evidence may include the callback address (webhook) of the operation server, an encryption token (Token), and a communication key (Adaptive Communication Environment key, ACE key). The computer device can set the callback address of the question-and-answer service as the callback address of the operation server in the communication application. The computer device can display a question-and-answer interaction page in the communication application, and a user can input a business problem on the question-and-answer interaction page. Here, the business problem may include a trigger parameter, which refers to a parameter possessed by the user when submitting the business problem. If the trigger parameter satisfies the question-answer trigger condition specified by a trigger component in the question-answer service, the computer device can transmit the business problem and a message certificate to a communication server corresponding to the communication application via the trigger component. The communication server can encrypt the accepted business problem with an encryption token and a communication key to obtain an encrypted ciphertext, and the communication server can transmit the encrypted ciphertext to the operation server. The operation server receives the encrypted ciphertext via the trigger component, and stores the encrypted token and the message certificate stored in the operation server.and the communication key to decrypt the encrypted ciphertext to obtain the business problem, and then call the question answering strategy component in the question answering service. The operation server filters the business problem using the question answering strategy component in the question answering service, performs question and answer identification on the filtered business problem using the linguistic question answering model component in the question answering service, outputs the question answering service result, transmits the question answering service result to the computer device, and displays the question answering service result on the question answering interaction page in the communication application. Step S203: Determine the filtered business problem as an output data variable of the trigger component, and determine the output data variable of the trigger component as an input data variable of the question answering strategy component. Step S204: Determine a function function corresponding to the question answering strategy component based on the instance template identifier, and process the input data variable of the question answering strategy component using the function function corresponding to the question answering strategy component to obtain the output data variable of the question answering strategy component, and determine the output data variable of the question answering strategy component as the input data variable of the linguistic question answering model component. The computer device can determine a function function corresponding to the question answering strategy component based on the instance template identifier in the execution record table, process the input data variables of the question answering strategy component using the function function corresponding to the question answering strategy component, and determine the output data variables of the question answering strategy component as input data variables of the linguistic question answering model component. Specifically, the computer device can use the accepted business problem as an input data variable of the question answering strategy component using the trigger component, and the computer device can determine the function function corresponding to the question answering strategy component using the instance template identifier in the execution record table. For example, the authority determination subcomponent in the question answering strategy component searches in the user list uesrlist (question authority range) with question authority using the function function userlist.find(),The computer device can filter out business problems that do not return results. The computer device can process the input data variables of the question answering strategy component using a functional function corresponding to the question answering strategy component to obtain the output data variables of the question answering strategy component (i.e., the business problems after filtering by the question answering strategy component). The computer device can determine the output data variables of the question answering strategy component as the input data variables of the linguistic question answering model component. In one embodiment, the question answering strategy component in the question answering service includes a problem determination subcomponent, a frequency control subcomponent, and an authority determination subcomponent. The step of filtering business problems and obtaining filtered business problems by the question answering strategy component in the question answering service includes the steps of: acquiring question answering statistical information of the user who asked the business problem, the question answering statistical information including question frequency and question authority, where the question frequency is the frequency at which the user calls the question answering service and the question authority is determined based on the user's user authority; acquiring business problems and filtering the business problems according to business prohibited words and problem processing abnormality formats specified by the problem determination subcomponent, and transmitting the business problems filtered by the problem determination subcomponent to the frequency control subcomponent; filtering the business problems filtered by the problem determination subcomponent according to the frequency threshold and question frequency specified by the frequency control subcomponent, and transmitting the business problems filtered by the frequency control subcomponent to the authority determination subcomponent; and filtering the business problems filtered by the frequency control subcomponent according to the problem authority range and question authority specified by the authority determination subcomponent to obtain filtered business problems. The process of the computer device filtering the business questions through the question determination subcomponent, the frequency control subcomponent, and the authority determination subcomponent in the question answering strategy component may be as follows: obtaining question answering statistics information of the user who has issued the business question, the question answering statistics information including question frequency andThe system includes a question authority, where the question frequency is the frequency at which the user calls the question-answering service and the question authority is determined based on the user's user authority, acquires business problems, filters the business problems according to the business prohibited words and problem processing abnormality format specified by the problem determination subcomponent, transmits the business problems filtered by the problem determination subcomponent to a frequency control subcomponent, filters the business problems filtered by the problem determination subcomponent according to the frequency threshold and question frequency specified by the frequency control subcomponent, transmits the business problems filtered by the frequency control subcomponent to an authority determination subcomponent, filters the business problems filtered by the frequency control subcomponent according to the question authority range and question authority specified by the authority determination subcomponent, and transmits the business problems filtered by the authority determination subcomponent to a linguistic question-answering model component in the question-answering service. Specifically, the computer device can acquire question-answering statistical information of the user who posed the business problem. Here, the question-answering statistical information includes a question frequency and a question authority, where the question frequency is the frequency at which the user calls the question-answering service and the question authority is determined based on the user's user authority. The computer device receives the business problem transmitted by the trigger component (i.e., the output data variable of the trigger component), and filters the business problem according to the business prohibited words and problem processing abnormal format instructed by the problem judgment subcomponent, and can filter out business problems that contain business prohibited words or whose file format is in the problem processing abnormal format. Business prohibited words refer to words that are not allowed to be used in business problems, and may be words that violate laws and regulations, or words that consist only of symbols, spaces, etc. and cannot express the meaning of actual language. The problem processing abnormal format is specifically set for the selected language question answering model component. For example, if the selected language question answering model component can only process text messages, the problem processing abnormal format can be set to include pictures,The business problem file format can be set to a format that cannot be processed by a language question-answering model, such as audio. The computer device transmits the business problems filtered by the question determination subcomponent to a frequency control subcomponent, which filters the business problems filtered by the question determination subcomponent according to the frequency threshold and question frequency specified by the frequency control subcomponent. Business problems whose frequency of invoking the question-answering service by the same user exceeds the filtering threshold specified by the frequency control subcomponent can be filtered. The unit time range of the frequency of invoking the question-answering service can be one hour, one day, or one week, and is not limited here. The computer device transmits the business problems filtered by the frequency control subcomponent to an authority determination subcomponent, which filters the business problems filtered by the frequency control subcomponent according to the question authority range and question authority specified by the authority determination subcomponent. Business problems whose question authority does not fall within the question authority range can be filtered. Here, the question authority can be determined by the user authority of the user. For example, user authority refers to whether the user is an object such as an administrator or a member of a chat group in a communication application. The computer device can transmit the business problems filtered by the authority determination subcomponent to the linguistic question answering model component in the question answering service (determine the business problems filtered by the authority determination subcomponent as input data variables of the linguistic question answering model component). In an embodiment of the present application, some business problems are filtered by a question answering strategy component (which may include a problem determination subcomponent, a frequency control subcomponent, an authority determination subcomponent, etc.), thereby reducing the waste of calculation costs of the linguistic question answering model and improving the quality and production efficiency of the question answering service. Step S205: Determine a function function corresponding to the linguistic question answering model component based on the instance template identifier, and use the function function corresponding to the linguistic question answering model component to:The computer device performs a function process on the input data variables of the language question-answering model component to obtain the output data variables of the language question-answering model component. The computer device can determine a function function corresponding to the language question-answering model component based on the instance template identifier in the execution record table, perform a function process on the input data variables of the language question-answering model component using the function function corresponding to the language question-answering model component, and determine the output data variables of the language question-answering model component as the question-answering service result. Specifically, the computer device can determine a function function corresponding to the language question-answering model component based on the instance template identifier in the execution record table, and perform a function process on the input data variables of the language question-answering model component (the business questions after filtering by the authority determination subcomponent) using the function function corresponding to the language question-answering model component. In one embodiment, the step of performing question and answer identification on the filtered business problem by the linguistic question and answer model component in the question and answer service and outputting the question and answer service result includes the steps of: removing business stop words from the filtered business problem to obtain an updated business problem; performing text segmentation on the updated business problem to obtain N text segments, where N is a positive integer; converting each of the N text segments into a text word vector, encoding the text word vector, and obtaining context information and semantic information of the text word vector; and performing semantic identification on the context information and semantic information by the linguistic question and answer model component to obtain the question and answer service result for the business problem. The process of the linguistic question and answer model component may be as follows: first, the authority determination subcomponent performs data preprocessing (including filtering business stop words and text segmentation) on the filtered business problem and converts it into a text word vector, and then performs question and answer identification on the filtered business problem.The process removes business stop words from the filtered business problem, obtains an updated business problem, performs text segmentation on the updated business problem, obtains N text segments, where N is a positive integer, converts each of the N text segments into a text word vector, encodes the text word vector, obtains context information and semantic information of the text word vector, and performs semantic identification on the context information and semantic information based on the linguistic question answering model component to obtain a question answering service result for the business problem. Specifically, the process in which the computer device performs data preprocessing (including filtering business stop words and text segmentation) on the business problem filtered by the authority judgment subcomponent and converts it into a text word vector can refer to the specific description of step S101 in the embodiment of FIG. 4, and will not be repeated here. The computer device encodes the text word vector and obtains the context information and semantic information of the text word vector. Here, the context information and semantic information of the text word vector are the restored word forms of the text segment, and the computer device can perform semantic identification on the context information and semantic information using the linguistic question-answering model component to obtain the question-answering service result for the business problem. After the question-answering service processing is completed, the execution record table will further store the input data variables and output data variables of each function code package component. Step S206: Display the question-answering service result for the business problem on the question-answering interaction page. Specifically, the computer device can display the question-answering service result for the business problem on the question-answering interaction page in the communication application. In an embodiment of the present application, a question-answering interaction page is displayed in the communication application, and a business problem is input on the question-answering interaction page, and the business problem includes trigger parameters. If the trigger parameters satisfy the question-answering trigger conditions specified by the trigger component,A trigger component is used to call a question answering strategy component in the question answering service; an execution record table is established for the question answering service based on the selected trigger component, question answering strategy component, and language question answering model component; component information of the function code package component in the question answering service is recorded in the execution record table; the component information includes a process flow connection relationship, an instance name, and an instance template identifier used to characterize the function code package component in the question answering service; the business problem is determined as an output data variable of the trigger component; the output data variable of the trigger component is determined as an input data variable of the question answering strategy component according to the process flow connection relationship indicated by the execution record table; The method further comprises: determining a function function corresponding to the question answering strategy component based on the instance template identifier; processing the input data variables of the question answering strategy component using the function function corresponding to the question answering strategy component to obtain output data variables of the question answering strategy component; determining the output data variables of the question answering strategy component as input data variables of the language question answering model component; determining a function function corresponding to the language question answering model component based on the instance template identifier in the execution record table; processing the input data variables of the language question answering model component using the function function corresponding to the language question answering model component to obtain output data variables of the language question answering model component; determining the output data variables of the language question answering model component as question answering service results; and displaying the question answering service results for the business problem on the question answering interaction page. This allows the question answering service to repeatedly perform repetitive tasks, reducing labor costs and manual intervention, making the question answering service convenient to use and easy to operate, reducing the implementation error rate, and improving production efficiency, thereby enhancing the competitiveness of the enterprise. Referring to FIG. 7, FIG. 7 is a flowchart of a question answering service generation method provided by an embodiment of the present application, which includes the following steps:The method may be executed by a computer device, which may be the terminal device 200 or the operation server 100 shown in Fig. 1. As shown in Fig. 7, the method may include the following steps: Specifically, the question-answering service editing application in the computer device can graphically render and present the function code package components selected by the user using the graphics rendering engine shown in Fig. 5. For example, a user can select function code package components such as an application trigger component (the application trigger component may be the starting component for operating the entire question-answering service workflow; when a user inputs a business problem into an independent page of the question-answering service, the application trigger component can activate the function provided by the question-answering service and call the question-answering strategy component in the question-answering service). The function code package components may include, for example, a problem determination subcomponent, a problem determination subcomponent, a frequency control subcomponent, an authority determination subcomponent, a linguistic question-answering model component, a reply component (displaying the question-answering service result on the question-answering interaction page), a logging component (recording log information of the function code package components in the question-answering service), a message notification component (displaying a message notification on the question-answering interaction page), a flow stop component (stopping the flow when an error or abnormality occurs in the question-answering service), and a flow termination component (terminating the flow when the question-answering service ends). The establishment of the process flow relationships of the above function code package components is as shown in FIG. 7 , and it is confirmed that the question-answering service is generated according to the process flow relationships of the selected function code package components. For the specific implementation of the function of the application trigger component in the question answering service, please refer to the description of step S202 in the embodiment corresponding to Figure 6, and detailed description will be omitted here.The business problem can be transmitted to the problem determination subcomponent. For specific implementation methods of the problem determination subcomponent, frequency control subcomponent, and authority determination subcomponent in the question answering service, please refer to the description of step S204 in the embodiment corresponding to Figure 6, and detailed description will be omitted here. The computer device can logically determine the flow of a business problem that contains a business prohibited word or whose file format is an abnormal problem processing format as "No" and stop the "No" flow. The computer device can transmit a business problem that does not contain a business prohibited word and whose file format is not an abnormal problem processing format to the frequency control component. The computer device can logically determine the flow of a business problem whose question frequency (frequency of calling the question answering service) of the same user exceeds the filtering threshold specified by the frequency control subcomponent as "No" and stop the "No" flow. Business problems whose question frequency is less than or equal to the filtering threshold are sent to the authority judgment subcomponent. The computer device logically determines the flow of business problems whose user's question authority does not fall within the question authority range as "No," displays a message reminder of "No question authority" on the question-answering interaction page, and stops the "No" flow. Business problems that fall within the question authority range are sent to the language question-answering model component, and the language question-answering model component performs semantic identification on the business problem to obtain a question-answering service result. The computer device displays the question-answering service result for the business problem on the question-answering interaction page using the answering component, records log information of the function code package components in the question-answering service using the logging component, and finally terminates the current question-answering service flow. Embodiments of the present application provide various types of function code package components that support independent deployment, realize flexible connection of function code package components, and can generate a question-answering service according to the process flow connection relationships between the function code package components selected by the user.

[0013] In this embodiment, a question-answering service is developed by a person who does not have specialized programming knowledge, using visualized function code package components and process flow connection relationships, rather than by writing code. This improves the productivity and versatility of question-answering service development, and reduces development costs and implementation error rates.

[0014] Referring to Figure 8, Figure 8 is a structural schematic diagram of a question-answering service generation device provided by an embodiment of the present application. As shown in FIG. 8 , the question and answer service generation device 1 includes a training text acquisition module 2001, a model training module 2002, a component display module 2003, a component editing module 2004, a component confirmation module 2005, a first organization module 2006, a second organization module 2007, a third organization module 2008, a fourth organization module 2009, a first setting module 2010, a second setting module 2011, a third setting module 2012, a fourth setting module 2014, a setting update module 2015, a log table establishment module 2016, a log processing module 2017, a first input module 2018, a transmission processing module 2019, a second input module 2020, a question and answer processing module 2021, and a result display module 2032. Here, the question and answer processing module includes an execution table establishment unit 2022, a first component processing unit 2023, a second component processing unit 2024, a third component processing unit 2025, a statistical information acquisition unit 2026, a first filtering unit 2027, a second filtering unit 2028, a third filtering unit 2029, a pre-processing unit 2030, and a semantics identification unit 2031. The training text acquisition module 2001 acquires a training text set, filters business stop words in the training text set, performs text segmentation on the filtered training text set to obtain M text segments, where M is a positive integer; calculates the number of occurrences of each text segment in the M text segments, and obtains text frequencies corresponding to the M text segments;The model training module 2002 is used for obtaining an initial language model and creating a text dictionary of the initial language model based on M text segments and M text frequencies, where the text dictionary includes word vectors corresponding to the M text segments, respectively; obtaining question-answer samples and constructing question-answer word vectors corresponding to the question-answer samples using the text dictionary; and training the initial language model using the question-answer word vectors to obtain linguistic question-answer model components. The specific functional implementation manners of the training text acquisition module 2001 and the model training module 2002 may refer to the description of step S101 in the embodiment corresponding to FIG. 4, and detailed description thereof will be omitted here. The component display module 2003 is used for displaying a set of function code package components on the editing interaction page of the question-answer service editing application, where the set of function code package components includes multiple types of function code package components, and the multiple types of function code package components include at least one trigger component, at least one question-answering strategy component, and at least one linguistic question-answering model component. For the specific implementation manner of the function of the component display module 2003, please refer to the description of step S101 in the embodiment corresponding to Fig. 4, and detailed description will be omitted here. The component editing module 2004 is used to display the trigger component, question answering strategy component, and linguistic question answering model component selected by the component selection operation on a question answering service editing page included in the editing interaction page in response to a component selection operation on a function code package component set, and the question answering service editing page is further used to display the process flow connection relationship between the selected trigger component, question answering strategy component, and linguistic question answering model component. The specific implementation manner of the function of the component editing module 2004 is as follows:The description of step S102 in the embodiment corresponding to Figure 4 may be referred to, and a detailed description thereof will be omitted here. The component identification module 2005 is used to generate a question answering service based on the process flow connection relationships among the selected trigger component, question answering strategy component, and linguistic question answering model component in response to a question answering service confirmation operation triggered on the edit interaction page of the question answering service editing application, where the trigger component in the question answering service is used to invoke the question answering strategy component in the question answering service, the question answering strategy component in the question answering service is used to perform question answer filtering on the business problem after obtaining the business problem, and the model component in the question answering service is used to output the question answering service result according to the filtered business problem. The specific implementation manner of the component identification module 2005 may be referred to, and a detailed description thereof will be omitted here. The first organization module 2006 is used to establish a process flow connection relationship between the selected trigger component and the question answering strategy component in response to a first component organization operation for the selected function code package component on the question answering service edit page, and to determine the selected trigger component as a function code package component to be used for executing a question answering service initiation flow based on the process flow connection relationship between the selected trigger component and the question answering strategy component, and to determine the selected question answering strategy component as a function code package component to be used for executing a next flow of the trigger component. The second organization module 2007 is used to establish a process flow connection relationship between the selected question answering strategy component and the linguistic question answering model component in response to a second component organization operation for the selected function code package component on the question answering service edit page,and determining, based on the process flow connection relationship between the selected question answering strategy component and the linguistic question answering model component, the selected linguistic question answering model component as a function code package component used to execute the next flow of the question answering strategy component. The third organization module 2008 is used to establish a process flow connection relationship between the selected problem determination subcomponent and the frequency control subcomponent in response to a third component organization operation for the selected problem determination subcomponent and the frequency control subcomponent on the question answering service edit page, and to determine, based on the process flow connection relationship between the selected problem determination subcomponent and the frequency control subcomponent, the selected problem determination subcomponent as a function code package component used to execute the question answering strategy initiation flow and the selected frequency control subcomponent as a function code package component used to execute the next flow of the problem determination subcomponent. The fourth organization module 2009 is used to establish a process flow connection relationship between the selected frequency control subcomponent and the authority determination subcomponent in response to a fourth component organization operation for the frequency control subcomponent and the authority determination subcomponent selected on the question and answer service editing page, and to determine the selected authority determination subcomponent as a function code package component used to execute the next flow of the frequency control subcomponent based on the process flow connection relationship between the selected frequency control subcomponent and the authority determination subcomponent. For specific implementation methods of the first organization module 2006, the second organization module 2007, the third organization module 2008, and the fourth organization module 2009, please refer to the description of step S102 in the embodiment corresponding to Figure 4, and detailed description will be omitted here. The first setting module 2010 is used to establish a process flow connection relationship between the selected frequency control subcomponent and the authority determination subcomponent in response to a fourth component organization operation for the selected problem determination subcomponent on the question and answer service editing page, and to determine the selected authority determination subcomponent as a function code package component used to execute the next flow of the frequency control subcomponent based on the process flow connection relationship between the selected frequency control subcomponent and the authority determination subcomponent.The parameter setting module 2011 is used to set a business prohibited word and a problem processing abnormality format for a selected problem determination subcomponent on the parameter setting page in the edit interaction page. The second setting module 2011 is used to set a frequency threshold for the selected frequency control subcomponent on the parameter setting page in response to a second component setting operation for the frequency control subcomponent selected on the question answering service edit page. The third setting module 2012 is used to set a problem authority range for the selected authority determination subcomponent on the parameter setting page in response to a third component setting operation for the authority determination subcomponent selected on the question answering service edit page. The fourth setting module 2014 is used to display a plurality of instance template identifiers on the parameter setting page in the edit interaction page and determine a target instance template identifier from the plurality of instance template identifiers in response to a component setting operation for a function code package component selected on the question answering service edit page. The setting update module 2015 is used to update the selected function code package component on the question answering service editing page based on the target instance template identifier, the instance template identifier is used to determine the function function corresponding to the function code package component, and the function function is used to process the business problem. For specific implementation methods of the first setting module 2010, the second setting module 2011, the third setting module 2012, the fourth setting module 2014, and the setting update module 2015, please refer to the description of step S102 in the embodiment corresponding to FIG. 4, and detailed description will be omitted here. The log table establishment module 2016 establishes an execution log table for the question answering service based on the selected trigger component, question answering strategy component, and linguistic question answering model component, and records log information of the function code package component in the question answering service in the execution log table.The log processing module 2017 is used to calculate the operating status and service quality quantification data of the question answering service according to the log information, and adjust at least one of the trigger component, the question answering strategy component, the linguistic question answering model component, and the process flow connection relationship in the question answering service based on the operating status and service quality quantification data of the question answering service. For specific functional implementations of the log table establishment module 2016 and the log processing module 2017, please refer to the description of step S103 in the embodiment corresponding to FIG. 4, and detailed description will be omitted here. The first input module 2018 is used to display a question answering interaction page of the question answering service in the communication application and obtain a business problem entered on the question answering interaction page, where the business problem includes a trigger parameter. If the trigger parameters satisfy the question-answer trigger conditions specified by the trigger component, the transmission processing module 2019 obtains a message certificate through the trigger component and sends the business problem and the message certificate to the communication server, which then encrypts the business problem according to the message certificate, obtains the encrypted ciphertext, and sends the encrypted ciphertext to the operation server. The message certificate is then stored in the operation server, which then decrypts the encrypted ciphertext using the stored message certificate to obtain the business problem. The operation server then outputs the question-answer service result corresponding to the business problem through the question-answer service. For specific functional implementations of the first input module 2018 and the transmission processing module 2019, please refer to the description of step S202 in the embodiment corresponding to FIG. 6, and detailed description will be omitted here. The second input module 2020 is used to display a question-answer interaction page of the question-answer service in the communication application and obtain the business problem entered on the question-answer interaction page, where the business problem includes the trigger parameters. The specific functional implementations of the second input module 2020 are as follows:Please refer to the description of step S202 in the embodiment corresponding to FIG. 6, and detailed description will be omitted here. The question answer processing module 2021 is used to call the question answering strategy component in the question answering service through the trigger component if the trigger parameters satisfy the question answering trigger condition specified by the trigger component. The question answer processing module 2021 is further used to filter the business problem through the question answering strategy component in the question answering service and obtain the filtered business problem. In one embodiment, the question answer processing module 2021 is further used to perform question answer identification for the filtered business problem through the linguistic question answering model component in the question answering service and output the question answering service result. In one embodiment, the result display module 2032 is further used to display the question answering service result for the business problem on the question answer interaction page. The execution table establishment unit 2022 is used to obtain an execution record table of the question answering service, where the execution record table includes an instance template identifier. The execution table establishment unit 2022 specifically establishes an execution record table for the question answering service based on the selected trigger component, question answering strategy component, and language question answering model component, and can record component information of the function code package component in the question answering service through the execution record table, where the component information includes the process flow connection relationship, and the instance name and instance template identifier used to characterize the function code package component in the question answering service. The specific function realization manner of the execution table establishment unit 2022 can refer to the description of step S201 in the embodiment corresponding to Figure 6, and detailed description will be omitted here. The first component processing unit 2023 determines the filtered business problem as the output data variable of the trigger component;The first component processing unit 2023 is used to determine the output data variables of the trigger component as the input data variables of the question answering strategy component. Specifically, the first component processing unit 2023 may be used to determine the business problem as the output data variable of the trigger component and determine the output data variables of the trigger component as the input data variables of the question answering strategy component according to the process flow connection relationship indicated by the execution record table. For the specific implementation of the function of the first component processing unit 2023, please refer to the description of step S203 in the embodiment corresponding to FIG. 6, and a detailed description will be omitted here. The second component processing unit 2024 is used to determine a function function corresponding to the question answering strategy component based on the instance template identifier, process the input data variables of the question answering strategy component using the function function corresponding to the question answering strategy component to obtain the output data variables of the question answering strategy component, and determine the output data variables of the question answering strategy component as the input data variables of the linguistic question answering model component. For the specific implementation of the function of the second component processing unit 2024, please refer to the description of step S204 in the embodiment corresponding to FIG. 6, and a detailed description will be omitted here. The third component processing unit 2025 is used for determining a function function corresponding to the language question answering model component based on the instance template identifier, processing input data variables of the language question answering model component using the function function corresponding to the language question answering model component to obtain output data variables of the language question answering model component, and determining the output data variables of the language question answering model component as a question answering service result. Specifically, the third component processing unit 2025 determines a function function corresponding to the language question answering model component based on the instance template identifier in the execution record table, processing input data variables of the language question answering model component using the function function corresponding to the language question answering model component to obtain output data variables of the language question answering model component,The output data variables of the linguistic question answering model component may be used to determine the question answering service result, and the execution record table is further used to store the input data variables and output data variables of the function code package component in the question answering service. The specific functional implementation of the third component processing unit 2025 can refer to the description of step S205 in the embodiment corresponding to FIG. 6, and detailed description will be omitted here. The statistical information acquisition unit 2026 is used to acquire question answering statistical information of users who issue business problems. The question answering statistical information includes question frequency and question authority, where the question frequency is the frequency at which the user calls the question answering service, and the question authority is determined based on the user's user authority. The first filtering unit 2027 is used to acquire business problems, filter them according to the business prohibited words and problem processing abnormality format specified by the problem determination subcomponent, and transmit the business problems filtered by the problem determination subcomponent to the frequency control subcomponent. The second filtering unit 2028 is used to filter the business problems filtered by the problem judgment subcomponent according to the frequency threshold and question frequency specified by the frequency control subcomponent, and transmit the business problems filtered by the frequency control subcomponent to the authority judgment subcomponent. The third filtering unit 2029 is used to filter the business problems filtered by the frequency control subcomponent according to the question authority range and question authority specified by the authority judgment subcomponent, and obtain filtered business problems. For specific implementation methods of the first filtering unit 2027, the second filtering unit 2028, and the third filtering unit 2029, please refer to the description of step S204 in the embodiment corresponding to Figure 6, and detailed description will be omitted here. The pre-processing unit 2030 removes business stop words from the filtered business problems to obtain updated business problems, and performs text segmentation on the updated business problems to obtain N text segments, where N is a positive integer;The semantics identification unit 2031 is used to convert each of the N text segments into a text word vector, encode the text word vector, obtain context information and semantics information of the text word vector, and use a linguistic question answering model component to perform semantics identification on the context information and semantics information to obtain a question answering service result for the business problem. For specific functional implementations of the preprocessing unit 2030 and the semantics identification unit 2031, please refer to the description of step S205 in the embodiment corresponding to FIG. 6, and detailed descriptions will be omitted here. The result display module 2032 is used to display the question answering service result for the business problem on the question answering interaction page. For specific functional implementations of the result display module 2032, please refer to the description of step S206 in the embodiment corresponding to FIG. 6, and detailed descriptions will be omitted here. Please refer to FIG. 9, which is a structural schematic diagram of a computer device provided by an embodiment of the present application. As shown in FIG. 9, the computer device 1000 may include a processor 1001, a network interface 1004, and a memory 1005. In addition, the computer device 1000 may further include a user interface 1003 and at least one communication bus 1002. The communication bus 1002 is used to realize communication between these components. The user interface 1003 may include a display screen and a keyboard, and may optionally further include a standard wired interface and a wireless interface. The network interface 1004 may optionally include a standard wired interface and a wireless interface (e.g., a Wi-Fi interface). The memory 1005 may be a high-speed RAM memory or a non-volatile memory, such as at least one magnetic disk memory. The memory 1005 may optionally include:9, the computer device 1000 may further include at least one storage device located at a location remote from the processor 1001. As shown in Fig. 9, a memory 1005 serving as a computer-readable storage medium may include an operating system, a network communication module, a user interface module, and a device control application program. In the computer device 1000 shown in Fig. 9, the network interface 1004 can provide a network communication network element, the user interface 1003 is mainly used to provide an input interface for a user, and the processor 1001 invokes the device control application program stored in the memory 1005, thereby displaying a set of function code package components on an edit interaction page of the question answering service editing application, the set of function code package components including a plurality of types of function code package components, the plurality of types of function code package components including at least one trigger component, at least one question answering strategy component, and at least one linguistic question answering model component. In response to a component selection operation for the set of components, displaying a trigger component, a question answering strategy component, and a language question answering model component selected by the component selection operation in a question answering service editing page included in the edit interaction page, wherein the question answering service editing page is further used to display a process flow connection relationship between the selected trigger component, the question answering strategy component, and the language question answering model component; and in response to a question answering service confirmation operation triggered in the edit interaction page of the question answering service editing application, generating a question answering service based on the process flow connection relationship between the selected trigger component, the question answering strategy component, and the language question answering model component, wherein the trigger component in the question answering service isThe computer device 1000 described in the embodiments of the present application can execute the description of the question-answering service generation method in the embodiments corresponding to any one of FIGS. 4, 6, and 7 above, and will not be described again here. Also, detailed descriptions of the beneficial effects of adopting the same method will be omitted. It should be noted that the embodiments of the present application further provide a computer-readable storage medium, which stores computer-readable instructions. When the processor executes the computer-readable instructions, the computer device 1000 can execute the description of the question-answering service generation method in the embodiments corresponding to any one of FIGS. 4, 6, and 7 above, and will not be described again here. Furthermore, detailed descriptions of the beneficial effects of the same method will be omitted. For technical details not disclosed in the computer-readable storage medium embodiments of the present application, please refer to the description of the method embodiments of the present application. The computer-readable storage medium may be an internal storage unit of the question-answering service generating device provided by any one of the above-mentioned embodiments or the computer device, such as a hard disk or internal memory of the computer device. The computer-readable storage medium may also be an external storage device of the computer device, such as a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, etc., installed on the computer device. Furthermore, the computer-readable storage medium may be:The computer device may include not only an internal storage unit but also an external storage device. The computer-readable storage medium is used to store the computer-readable instructions and other programs and data required by the computer device. The computer-readable storage medium may also be used to temporarily store data that has been displayed or is to be displayed. It should be noted that the present embodiment also provides a computer program product, which includes computer-readable instructions stored in a computer-readable storage medium. A processor of the computer device reads the computer-readable instructions from the computer-readable storage medium, and executes the computer-readable instructions to cause the computer device to perform the method provided by the embodiment corresponding to any one of Figures 4, 6, and 7 above. The terms "first," "second," etc. in the description, claims, and drawings of the present embodiment are used to distinguish different objects, not to describe a particular order. The term "comprises" and any variations thereof are intended to cover a non-exclusive inclusion. For example, a process, method, apparatus, product, or device that includes a series of steps or units is not limited to the steps or modules already listed, but may optionally further include steps or modules not listed, or may optionally further include other step units inherent to the process, method, apparatus, product, or device. Those skilled in the art will recognize that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be realized by electronic hardware, computer software, or a combination of both, and in order to clearly explain the compatibility of hardware and software, the composition and steps of each example have already been generally described according to the network elements in the above description. Whether these network elements are implemented in hardware or software is up to the discretion of the user.The technical solution is determined by the specific application and design constraints. Professionals may implement the described network elements using different methods for each specific application, but such implementations should not be considered beyond the scope of the present application. The technical features of the above embodiments can be arbitrarily combined. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered within the scope of the present application. The above embodiments represent only some embodiments of the present application, and although the description is relatively specific and detailed, it should not be understood as limiting the scope of the invention patent. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present application, all of which fall within the scope of protection of the present application. Therefore, the scope of protection of this patent application is as follows:The appended claims should be taken as the standard. [Explanation of symbols] 1. Generating device 100. Working server 200. Terminal device 210. Editing interaction page 211. Selection control 212. Component candidate page 213. Question-answering service editing page 220. Editing interaction page 221. Connection control 222. Setting control 223. Parameter setting page 224. Save control 300. Communication server 400. Terminal device 410. Question-answering interaction page 420. Question-answering interaction page 1000. Computer device 1001. Processor 1002. Communication bus 1003. User interface 1004. Network interface 1005. Memory 2001. Training text acquisition module 2002. Model training module 2003. Component display module 2004. Component editing module 2005. Component confirmation module 2006. First compilation module 2007. Second compilation module 2008. Third compilation module 2009. Fourth compilation module 2010. First setting module 2011 a second setting module 2012, a third setting module 2014, a fourth setting module 2015, a setting update module 2016, a log table establishment module 2017, a log processing module 2018, a first input module 2019, a transmission processing module 2020, a second input module 2021, a question and answer processing module 2022, an execution table establishment unit 2023, a first component processing unit 2024, a second component processing unit 2025, a third component processing unit 2026, a statistical information acquisition unit 2027, a first filtering unit 2028, a second filtering unit 2029, a third filtering unit 2030, a pre-processing unit 2031, a semantics identification unit 2032, a result display module,

Claims

1. 1. A method for generating a question-answering service, the method being performed by a computing device, comprising: displaying a set of function code package components in an edit interaction page of a question answering service editing application, the set of function code package components including a plurality of types of function code package components, the plurality of types of function code package components including at least one trigger component, at least one question answering strategy component, and at least one linguistic question answering model component; a step of displaying, in response to a component selection operation for the set of function code package components, a trigger component, a question answering strategy component, and a linguistic question answering model component selected by the component selection operation in a question answering service edit page included in the edit interaction page, wherein the question answering service edit page is further used to display a process flow connection relationship between the selected trigger component, the question answering strategy component, and the linguistic question answering model component; and generating a question and answer service based on a process flow connection relationship among a selected trigger component, a question answering strategy component, and a linguistic question and answer model component in response to a question and answer service confirmation operation triggered on an edit interaction page of the question and answer service editing application, wherein the trigger component in the question and answer service is used to invoke the question and answer strategy component in the question and answer service, the question and answer strategy component in the question and answer service is used to perform question and answer filtering on the business problem after obtaining the business problem, and the model component in the question and answer service is used to output a question and answer service result according to the filtered business problem.

2. The method comprises: displaying a question-answering interaction page of the question-answering service in a communication application, and acquiring a business problem input on the question-answering interaction page, the business problem including a trigger parameter; 2. The method of claim 1, further comprising the steps of: if the trigger parameters satisfy the question-answering trigger condition indicated by the trigger component, obtaining a message certificate by the trigger component and sending the business problem and the message certificate to a communication server, whereby the communication server encrypts the business problem according to the message certificate, obtains an encrypted ciphertext, and sends the encrypted ciphertext to a work server; wherein the message certificate is further stored in the work server, and the work server decrypts the encrypted ciphertext using the stored message certificate to obtain the business problem; and the work server is further used to output a question-answering service result corresponding to the business problem by the question-answering service.

3. The method comprises: displaying a question-answering interaction page of the question-answering service in a communication application, and acquiring a business problem input on the question-answering interaction page, the business problem including a trigger parameter; invoking a question answering strategy component in a question answering service by the trigger component if the trigger parameters satisfy a question answering trigger condition indicated by the trigger component; filtering the business problem through a question answering strategy component in the question answering service to obtain a filtered business problem; performing question and answer identification for the filtered business problem using a linguistic question and answer model component in the question and answer service, and outputting a question and answer service result; The method of claim 1 , further comprising: displaying the question and answer service results for the business problem on the question and answer interaction page.

4. The method further includes the step of obtaining a performance record table of the question-answering service, the performance record table including an instance template identifier; The step of performing question and answer identification for the filtered business problem by a linguistic question and answer model component in the question and answer service and outputting a question and answer service result includes: determining the filtered business problem as an output data variable of the trigger component, and determining the output data variable of the trigger component as an input data variable of the question answering strategy component; determining a function corresponding to the question answering strategy component based on the instance template identifier, and processing input data variables of the question answering strategy component using the function corresponding to the question answering strategy component to obtain output data variables of the question answering strategy component; determining output data variables of the question answering strategy component as input data variables of the linguistic question answering model component; determining a functional function corresponding to the linguistic question answering model component based on the instance template identifier; a step of processing input data variables of the linguistic question answering model component by a function function corresponding to the linguistic question answering model component to obtain output data variables of the linguistic question answering model component; and determining output data variables of the linguistic question answering model component as question answering service results.

5. The step of performing question and answer identification for the filtered business problem by a linguistic question and answer model component in the question and answer service and outputting a question and answer service result includes: removing business stop words in the filtered business problem to obtain an updated business problem; performing text segmentation on the updated business problem to obtain N text segments, where N is a positive integer; converting each of the N text segments into a text word vector, encoding the text word vector, and obtaining context information and semantic information of the text word vector; and performing semantic identification on the context information and the semantic information by the linguistic question answering model component to obtain a question answering service result for the business problem.

6. The question answering strategy component in the question answering service includes a problem determination subcomponent, a frequency control subcomponent, and an authority determination subcomponent, and the step of filtering the business problems by the question answering strategy component in the question answering service and obtaining the filtered business problems includes: a step of acquiring question and answer statistical information of the user who posed the business problem, the question and answer statistical information including a question frequency and a question authority, the question frequency being a frequency at which the user calls the question and answer service, and the question authority being determined based on the user authority; Acquiring the business problems, filtering the business problems according to the business prohibited words and problem processing abnormality format instructed by the problem determination subcomponent, and transmitting the business problems filtered by the problem determination subcomponent to the frequency control subcomponent; filtering the business problems filtered by the problem determination subcomponent according to the frequency threshold indicated by the frequency control subcomponent and the question frequency, and transmitting the business problems filtered by the frequency control subcomponent to the authority determination subcomponent; The method according to any one of claims 3 to 5, further comprising a step of filtering the business problems filtered by the frequency control subcomponent according to the problem authority range and the question authority indicated by the authority determination subcomponent, to obtain filtered business problems.

7. The method comprises: in response to a first component setting operation for a problem determination subcomponent selected on the question and answer service editing page, setting a business prohibited word and a problem processing abnormality format for the selected problem determination subcomponent on a parameter setting page of the editing interaction page; setting a frequency threshold for the selected frequency control subcomponent on the parameter setting page in response to a second component setting operation for the selected frequency control subcomponent on the question answering service editing page; The method of any one of claims 1 to 6, further comprising: setting a problem authority range for the selected authority determination subcomponent on the parameter setting page in response to a third component setting operation for the authority determination subcomponent selected on the question answering service editing page.

8. The method comprises: displaying a plurality of instance template identifiers on a parameter setting page of the edit interaction page in response to a component setting operation for a function code package component selected on the edit interaction page; determining a target instance template identifier from the plurality of instance template identifiers; The method of any one of claims 1 to 7, further comprising: updating, on the question answering service editing page, the selected function code package component based on the target instance template identifier, wherein the instance template identifier is used to determine a function function corresponding to the function code package component, and the function function is used to process the business problem.

9. establishing a process flow connection relationship between a selected trigger component and a question answering strategy component in response to a first component organization operation on the selected function code package component on the question answering service edit page; determining the selected trigger component as a function code package component to be used for executing a question answering service initiation flow based on a process flow connection relationship between the selected trigger component and the question answering strategy component, and determining the selected question answering strategy component as a function code package component to be used for executing a next flow of the trigger component; establishing a process flow connection relationship between the selected question answering strategy component and the linguistic question answering model component in response to a second component organization operation on the selected function code package component on the question answering service edit page; The method according to any one of claims 1 to 8, further comprising: determining the selected linguistic question answering model component as a function code package component to be used in executing a next flow of the question answering strategy component based on a process flow connection relationship between the selected question answering strategy component and the linguistic question answering model component.

10. The question answering strategy component includes a problem determination subcomponent, a frequency control subcomponent, and an authority determination subcomponent, and the method includes: establishing a process flow connection between the selected problem determination subcomponent and the frequency control subcomponent in response to a third component organization operation on the selected problem determination subcomponent and the frequency control subcomponent on the question answering service edit page; Based on the process flow connection relationship between the selected problem determination subcomponent and the frequency control subcomponent, determine the selected problem determination subcomponent as a function code package component used to execute a question answering strategy initiation flow, and determine the selected frequency control subcomponent as a function code package component used to execute a next flow of the problem determination subcomponent; establishing a process flow connection between the selected frequency control subcomponent and the authority determination subcomponent in response to a fourth component organization operation for the selected frequency control subcomponent and the authority determination subcomponent on the question answering service edit page; The method according to any one of claims 1 to 9, further comprising: determining, based on a processing flow connection relationship between the selected frequency control subcomponent and the authority determination subcomponent, the selected authority determination subcomponent as a function code package component to be used in executing a next flow of the frequency control subcomponent.

11. obtaining a training text set, filtering business stop words in the training text set, and performing text segmentation on the filtered training text set to obtain M text segments, where M is a positive integer; Counting the number of occurrences of each text segment in the M text segments to obtain text frequencies corresponding to the M text segments; obtaining an initial language model and creating a text dictionary of the initial language model based on the M text segments and the M text frequencies, wherein the text dictionary includes word vectors corresponding to the M text segments respectively; obtaining a question-answer sample and constructing a question-answer word vector corresponding to the question-answer sample according to the text dictionary; The method of any one of claims 1 to 10, further comprising the step of: training the initial language model with the question answering word vectors to obtain the language question answering model components.

12. The method comprises: establishing an execution log table for the question answering service based on the selected trigger component, question answering strategy component, and linguistic question answering model component; recording log information of the function code package components in the question answering service using the execution log table; collecting statistics of the operation status and service quality quantification data of the question answering service according to the log information; The method according to any one of claims 1 to 11, further comprising: adjusting at least one of a trigger component, a question answering strategy component, a linguistic question answering model component, or the process flow connection relationship in the question answering service based on an operation state of the question answering service and service quality quantification data.

13. A question and answer service generation device, a component display module used to display a set of function code package components in an edit interaction page of a question answering service edit application, the set of function code package components including a plurality of types of function code package components, the plurality of types of function code package components including at least one trigger component, at least one question answering strategy component, and at least one linguistic question answering model component; a component editing module used to display, in response to a component selection operation for the set of function code package components, a trigger component, a question answering strategy component, and a linguistic question answering model component selected by the component selection operation in a question answering service editing page included in the editing interaction page, wherein the question answering service editing page is further used to display a process flow connection relationship between the selected trigger component, the question answering strategy component, and the linguistic question answering model component; a component confirmation module used to generate a question answering service based on a process flow connection relationship between a selected trigger component, a question answering strategy component, and a linguistic question answering model component in response to a question answering service confirmation operation triggered on an edit interaction page of the question answering service editing application, wherein the trigger component in the question answering service is used to invoke the question answering strategy component in the question answering service, the question answering strategy component in the question answering service is used to perform question answer filtering on the business problem after obtaining the business problem, and the model component in the question answering service is used to output a question answering service result according to the filtered business problem.

14. 13. A computing device comprising: a processor; and a memory, wherein the memory stores computer-readable instructions that, when executed by the processor, cause the processor to implement the method of any one of claims 1 to 12.

15. 13. A computer-readable storage medium having stored thereon computer-readable instructions that, when executed by a processor, cause the processor to implement the method of any one of claims 1 to 12.

16. 13. A computer program product comprising computer readable instructions that, when executed by a processor, cause the processor to perform the method of any one of claims 1 to 12.

Images