A question and answer method, device, apparatus and storage medium thereof

Abstract

This application belongs to the field of question-and-answer optimization technology and is applied to intelligent customer service question-and-answer prompt scenarios. It relates to a question-and-answer method, device, equipment and its storage medium, including acquiring target question data; inputting the target question data into a preset intent analysis model; determining the difficulty category of the target question data according to a preset category recognition component; by pre-setting difficulty categories for all question data, subsequent identification only requires simple and scientific labeling, which can achieve the search of a small amount of complex question data by labeling and indexing, while using intent analysis to search for answer data for a large amount of simple question data, which greatly improves the efficiency of dialogue answering and improves the user's dialogue experience.

Description

Technical Field

[0001] This application relates to the field of question-and-answer optimization technology, and is applied to intelligent customer service question-and-answer prompt scenarios. In particular, it relates to a question-and-answer method, device, equipment and its storage medium. Background Technology

[0002] Question-and-answer (CPA) engineering is an intelligent technology that uses robots to automatically analyze a customer's voice or text and provide corresponding answers or suggestions. The financial field involves a wide range of complex knowledge that is difficult for ordinary customer service representatives to accurately grasp. CPA engineering, however, leverages technology to precisely resolve customer questions and quickly provide satisfactory answers.

[0003] Traditional financial customer service requires human staff on duty, which can lead to low service efficiency when there are many customers. By adopting question-and-answer prompt engineering technology, robots can respond quickly to customers, improving consultation and service efficiency and reducing operating costs. However, question-and-answer prompt engineering requires collecting a large number of questions and having experts annotate them. This annotation process consumes a lot of manpower, and often fails to reasonably distinguish the difficulty levels of questions and answers, resulting in insufficiently scientific and simple annotations. Summary of the Invention

[0004] The purpose of this application is to provide a question-and-answer method, apparatus, device and its storage medium to solve the problem that the prior art often cannot reasonably distinguish the difficulty category of questions in the annotation of question data and answer data, resulting in insufficiently scientific and simple annotation.

[0005] To address the aforementioned technical problems, this application provides a question-and-answer method, employing the following technical solution:

[0006] A question-and-answer method includes the following steps:

[0007] Obtain data related to the target problem;

[0008] Input the target question data into a preset intent analysis model, and determine the difficulty category of the target question data according to a preset category recognition component, wherein the difficulty category includes simple question category and complex question category;

[0009] If the target question data belongs to the simple question category, then the solution knowledge content is searched from the preset solution knowledge base based on the intent analysis results, and solution feedback data is generated based on the solution knowledge content;

[0010] If the target question data belongs to the category of complex questions, then the solution knowledge content is searched from the solution knowledge base according to the pre-set annotation index, and solution feedback data is generated based on the solution knowledge content.

[0011] Furthermore, the preset intent analysis model includes a language analysis model based on ChatGLM. Before performing the step of determining the difficulty category of the target question data according to the preset category recognition component, the method further includes:

[0012] All historical question data is obtained through a pre-set question and answer log;

[0013] All the question data are input into the ChatGLM-based language analysis model for intent analysis.

[0014] Based on the intent analysis results, retrieve N search results for the answer knowledge content corresponding to each question data from the answer knowledge base, where N is a positive integer;

[0015] Based on the N search results for the knowledge content of each answer corresponding to all the question data, the difficulty category to which each of the question data belongs is set.

[0016] Furthermore, before performing the step of obtaining N search results for the answer knowledge content corresponding to each of the questions from the answer knowledge base based on the intent analysis results, the method further includes:

[0017] A pre-set strategy for acquiring the knowledge content of the N solutions is provided, wherein the acquisition strategy includes ensuring that the knowledge content of the N solutions is not repeated and that a maximum search time is preset for each search.

[0018] The step of obtaining N search results for answer knowledge content corresponding to each question data from the answer knowledge base based on the intent analysis results specifically includes:

[0019] Step 401: Select the current problem data from all the problem data;

[0020] Step 402: Based on the intent analysis results corresponding to the current question data and the acquisition strategy, obtain N search results for answer knowledge content corresponding to the current question data from the answer knowledge base. The search results for answer knowledge content include the answer knowledge content obtained in each search and the search time in each search.

[0021] Step 403: Take the next question data of the current question data as the latest current question data, and re-execute step 402 until all question data have undergone N search operations for answer knowledge content. Once the search is complete, stop executing step 402.

[0022] Furthermore, the step of obtaining N search results for answer knowledge content corresponding to the current question data from the answer knowledge base based on the intent analysis results corresponding to the current question data and the acquisition strategy specifically includes:

[0023] Based on the intent analysis results corresponding to the current question data and the acquisition strategy, obtain the answer knowledge content of N searches, and identify whether the answer knowledge content of each search is null.

[0024] If the answer to the current search is null, then determine whether the current search is the Nth search corresponding to the current question data.

[0025] If the current search is not the Nth search, then continue to the next search; if the current search is the Nth search, then select the latest current problem data.

[0026] If the answer to the current search is not empty, then determine whether the current search is the Nth search corresponding to the current question data.

[0027] If the current search is the Nth search, then the latest current problem data is selected again;

[0028] If the current search is not the Nth search, the solution knowledge content is marked as unsearchable until the number of searches corresponding to the current question data reaches N. Then, the solution knowledge content is changed to searchable, and the search results for each search are recorded.

[0029] Furthermore, the step of setting the difficulty category of each of the N answer knowledge content search results corresponding to all the question data specifically includes:

[0030] Statistically analyze the search results for the N solutions to each question;

[0031] Based on the statistical results, determine whether there are any problematic data points for which the corresponding answer knowledge content is less than N;

[0032] If all the question data corresponds to N pieces of answer knowledge content, then calculate the average search time and the variance of search time for each of the N searches for all the question data, and sort all the question data based on the average search time and the variance of search time to obtain the first sorting result;

[0033] If there are fewer than N answer knowledge items corresponding to a question, the question data is sorted according to the number of answer knowledge items in ascending order to obtain a second sorting result. For other question data where there are N answer knowledge items, the average search time and the variance of the search time are calculated respectively. The other question data is sorted based on the average search time and the variance of the search time to obtain a third sorting result. The second sorting result and the third sorting result are concatenated to obtain a fourth sorting result.

[0034] Based on the first sorting result or the fourth sorting result, simple problem data and complex problem data are divided.

[0035] Furthermore, the step of sorting all the problem data based on the average search time and the variance of the search time to obtain a first sorting result specifically includes:

[0036] According to the preset comprehensive time consumption formula:

[0037] a i =b i ×ω1+c i ×ω2,

[0038] Calculate the comprehensive time consumption value corresponding to all the problem data, where a i b represents the comprehensive time consumption characteristic value corresponding to problem data i. i ω1 represents the average search time for N searches of problem data i, and c represents the weight value corresponding to the average search time. i ω1 represents the search time variance value corresponding to N searches of problem data i, ω2 represents the weight value corresponding to the search time variance value, and ω1 is greater than ω2.

[0039] The comprehensive time consumption values ​​corresponding to all the problem data are sorted in ascending order to obtain a sequence of comprehensive time consumption values ​​corresponding to all the problem data.

[0040] Based on the comprehensive time consumption characterization value sequence, the corresponding problem data sequence is obtained as the first sorting result;

[0041] The step of sorting the other problem data based on the average search time and the variance of the search time to obtain a third sorting result, and concatenating the second sorting result and the third sorting result to obtain a fourth sorting result, specifically includes:

[0042] The comprehensive time consumption representation formula is used to calculate the comprehensive time consumption representation values ​​corresponding to the other problem data respectively;

[0043] The comprehensive time consumption values ​​corresponding to the other problem data are sorted in ascending order to obtain a sequence of comprehensive time consumption values ​​corresponding to the other problem data.

[0044] Based on the comprehensive time consumption characterization value sequence, the corresponding problem data sequence is obtained as the third sorting result;

[0045] The sequence end position of the second sorting result is concatenated with the sequence end position of the third sorting result;

[0046] The fourth sorting result is obtained by taking the start position of the third sorting result as the start position of the fourth sorting result and taking the start position of the second sorting result as the end position of the fourth sorting result.

[0047] Furthermore, the step of dividing simple problem data and complex problem data according to the first sorting result or the fourth sorting result specifically includes:

[0048] The first sorting result or the fourth sorting result is segmented according to the preset segmentation position;

[0049] The problem data contained from the start position of the sequence of the first sorting result or the fourth sorting result to the splitting position is taken as the simple problem data;

[0050] The problem data contained from the end position of the sequence of the first sorting result or the fourth sorting result to the split position is taken as the complex problem data;

[0051] After performing the step of dividing simple problem data and complex problem data according to the first sorting result or the fourth sorting result, the method further includes:

[0052] All the complex problem data identified are used as recognition knowledge and deployed to the category recognition component.

[0053] The solution knowledge content corresponding to all the complex question data in the solution knowledge base is marked by manual annotation, and the annotation results are recorded as index knowledge of the solution knowledge base in a preset query file.

[0054] To address the aforementioned technical problems, this application also provides a question-and-answer device, which employs the following technical solution:

[0055] A question-and-answer device, comprising:

[0056] The problem data acquisition module is used to acquire target problem data;

[0057] The difficulty category identification module is used to input the target question data into a preset intent analysis model and determine the difficulty category to which the target question data belongs based on a preset category identification component. The difficulty category includes simple question category and complex question category.

[0058] The first answer feedback module is used to search for answer knowledge content from a preset answer knowledge base based on intent analysis results if the target question data belongs to the simple question category, and generate answer feedback data based on the answer knowledge content.

[0059] The second answer feedback module is used to search for answer knowledge content from the answer knowledge base according to the pre-set annotation index if the target question data belongs to the complex question category, and generate answer feedback data based on the answer knowledge content.

[0060] To address the aforementioned technical problems, this application also provides a computer device that employs the following technical solution:

[0061] A computer device includes a memory and a processor, wherein the memory stores computer-readable instructions, and the processor executes the computer-readable instructions to implement the steps of the question-and-answer method described above.

[0062] To address the aforementioned technical problems, this application also provides a computer-readable storage medium, employing the technical solution described below:

[0063] A computer-readable storage medium storing computer-readable instructions that, when executed by a processor, implement the steps of the question-and-answer method as described above.

[0064] Compared with the prior art, the embodiments of this application have the following main advantages:

[0065] The question-answering method described in this application acquires target question data; inputs the target question data into a preset intent analysis model; and determines the difficulty category of the target question data based on a preset category recognition component. By pre-setting difficulty categories for all question data, subsequent identification only requires simple and scientific labeling. This allows for searching a small amount of complex question data by labeling and indexing, while using intent analysis to search for answers to a large amount of simple question data, greatly improving the efficiency of dialogue answering and enhancing the user's dialogue experience. Attached Figure Description

[0066] To more clearly illustrate the solutions in this application, the accompanying drawings used in the description of the embodiments of this application will be briefly introduced below. Obviously, the accompanying drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0067] Figure 1 This is an exemplary system architecture diagram to which this application can be applied;

[0068] Figure 2 This is a flowchart of one embodiment of the question-and-answer method according to this application;

[0069] Figure 3 This is a schematic diagram of a specific embodiment of the question data difficulty category setting method in the question-answering method described in this application;

[0070] Figure 4 yes Figure 3 A flowchart of a specific embodiment of step 303 shown;

[0071] Figure 5 yes Figure 4 A flowchart of a specific embodiment of step 402 shown;

[0072] Figure 6 yes Figure 3 A flowchart of a specific embodiment of step 304 shown;

[0073] Figure 7 yes Figure 6 A flowchart of a specific embodiment of step 605 shown;

[0074] Figure 8 This is a schematic diagram of the structure of one embodiment of the question-and-answer device according to this application;

[0075] Figure 9 This is a schematic diagram of the structure of one embodiment of the computer device according to this application. Detailed Implementation

[0076] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains; the terminology used herein in the specification of the application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application; the terms "comprising" and "having," and any variations thereof, in the specification, claims, and foregoing drawings of this application, are intended to cover non-exclusive inclusion. The terms "first," "second," etc., in the specification, claims, or foregoing drawings of this application are used to distinguish different objects, not to describe a particular order.

[0077] In this document, the term "embodiment" means that a particular feature, structure, or characteristic described in connection with an embodiment may be included in at least one embodiment of this application. The appearance of this phrase in various places throughout the specification does not necessarily refer to the same embodiment, nor is it a separate or alternative embodiment mutually exclusive with other embodiments. It will be explicitly and implicitly understood by those skilled in the art that the embodiments described herein can be combined with other embodiments.

[0078] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0079] like Figure 1 As shown, system architecture 100 may include terminal devices 101, 102, and 103, a network 104, and a server 105. Network 104 serves as the medium for providing communication links between terminal devices 101, 102, and 103 and server 105. Network 104 may include various connection types, such as wired or wireless communication links, or fiber optic cables, etc.

[0080] Users can use terminal devices 101, 102, and 103 to interact with server 105 via network 104 to receive or send messages, etc. Various communication client applications can be installed on terminal devices 101, 102, and 103, such as web browser applications, shopping applications, search applications, instant messaging tools, email clients, social media platform software, etc.

[0081] Terminal devices 101, 102, and 103 can be various electronic devices with displays and support web browsing, including but not limited to smartphones, tablets, e-book readers, MP3 players (Moving Picture Experts Group Audio Layer III), MP4 players (Moving Picture Experts Group Audio Layer IV), laptops, and desktop computers, etc.

[0082] Server 105 can be a server that provides various services, such as a backend server that supports the pages displayed on terminal devices 101, 102, and 103.

[0083] It should be noted that the question-and-answer method provided in this application embodiment is generally executed by a terminal device, and correspondingly, the question-and-answer device is generally set in the terminal device.

[0084] It should be understood that Figure 1The number of terminal devices, networks, and servers shown is merely illustrative. Depending on implementation needs, any number of terminal devices, networks, and servers can be included.

[0085] Continue to refer to Figure 2 A flowchart of an embodiment of the question-answering method according to this application is shown. The question-answering method includes the following steps:

[0086] Step 201: Obtain the target problem data.

[0087] Specifically, the target question data includes, for example, dialogue question data raised by a user through a mobile terminal or question-and-answer terminal, or dialogue question data input by a user through an input terminal. In this embodiment, the question-and-answer method can be used in the field of financial technology, such as in dialogue scenarios between intelligent AI customer service and financial business customers, or in business consultation scenarios between intelligent banking consultation terminals and bank customers. The question-and-answer method can also be used in the field of smart healthcare, such as in dialogue consultation scenarios between medical service consultation terminals and patients. The target question data includes question data in voice form and / or question data in input text form. Of course, it is not excluded that the question-and-answer method can be used in question-and-answer scenarios in other technical fields. The main purpose is to provide users with more efficient answer feedback data through the question-and-answer method, improve the efficiency of answer data feedback, and improve the user's question-and-answer experience. Especially in the field of financial technology, the questions involved are more professional, requiring the answerer to possess corresponding financial knowledge.

[0088] Step 202: Input the target question data into the preset intent analysis model, and determine the difficulty category of the target question data according to the preset category recognition component.

[0089] The difficulty categories include simple problem categories and complex problem categories.

[0090] Specifically, the preset intent analysis model includes a language analysis model based on ChatGLM. In this embodiment, the ChatGLM-based language analysis model is, for example, the ChatGLM large language model. ChatGLM, also known as Zhipu Qingyan, is developed based on the GLM-4 language model jointly trained by Zhipu AI in 2024 and can provide appropriate answers to user questions and requests. Directly using the ChatGLM large language model for language analysis to determine the intent corresponding to the target question data avoids the need to construct a large amount of question-and-answer data and reduces the amount of manual labor required.

[0091] In addition, the preset intent analysis model can also adopt an upgraded version of ChatGLM, such as ChatGLM-9b, which is an open-source dialogue language model that supports bilingual question and answer in Chinese and English and is based on the General Language Model (GLM) architecture.

[0092] In this embodiment, before performing the step of determining the difficulty category of the target problem data according to the preset category identification component, the method further includes a problem data difficulty category setting method.

[0093] For details, please refer to Figure 3 , Figure 3 This is a schematic diagram of a specific embodiment of the question data difficulty category setting method in the question-answering method described in this application, including:

[0094] Step 301: Obtain all historically input question data through a preset question and answer log;

[0095] Specifically, the preset question-and-answer log, for example, is the question-and-answer log between the customer and the customer service representative recorded in the intelligent AI customer service dialogue scenario of financial business. The "all historical input question data" refers to all question data recorded in the question-and-answer log in the historical dialogue scenario. Here, "input" is interpreted in an expanded way, including directly input question data, as well as question data recorded in the question-and-answer log after the question-and-answer speech has been converted.

[0096] In this embodiment, all historical question data is obtained through a preset question and answer log. This allows for direct integration with specific financial business scenarios to obtain all question data that customers have previously consulted, and facilitates the classification of all question data into difficulty categories.

[0097] Step 302: Input all the question data into the ChatGLM-based language analysis model for intent analysis;

[0098] Specifically, all the question data are input into the ChatGLM-based language analysis model for intent analysis. The purpose is to identify the question intent of all the question data and provide corresponding answer data.

[0099] Step 303: Based on the intent analysis results, obtain the N search results for the answer knowledge content corresponding to each question data from the answer knowledge base, where N is a positive integer;

[0100] In this embodiment, before performing the step of obtaining N search results for answer knowledge content corresponding to all question data from the answer knowledge base based on the intent analysis results, the method further includes: pre-setting an acquisition strategy for the N answer knowledge content, wherein the acquisition strategy includes that the N answer knowledge content is not repeated and that a maximum search time is preset for each search.

[0101] The acquisition strategy ensures that the knowledge content of the solutions obtained in the N acquisitions is different. Furthermore, a maximum search time is preset for each search to avoid a deadlock when no results are found, thus enabling the system to operate effectively and stably.

[0102] Continue to refer to Figure 4 , Figure 4 yes Figure 3 A flowchart of a specific embodiment of step 303 shown includes:

[0103] Step 401: Select the current problem data from all the problem data;

[0104] Specifically, all the question data can be pre-sorted or differentiated and numbered. Then, different question data are selected as the current question data according to the search sort sequence or differentiation number result, ensuring that the answer data content is searched N times for all the question data.

[0105] Step 402: Based on the intent analysis results corresponding to the current question data and the acquisition strategy, obtain N search results for answer knowledge content corresponding to the current question data from the answer knowledge base. The search results for answer knowledge content include the answer knowledge content obtained in each search and the search time in each search.

[0106] Specifically, the search results for the answer knowledge content can be composed of the answer knowledge content obtained in each search and the search time in each search. For example, if the answer knowledge content is obtained, it can be output in the form of "answer knowledge content + search time". If the answer knowledge content is not obtained, it can be output in the form of "empty value + maximum search time".

[0107] Continue to refer to Figure 5 , Figure 5 yes Figure 4 A flowchart of a specific embodiment of step 402 shown includes:

[0108] Step 501: Based on the intent analysis results corresponding to the current question data and the acquisition strategy, obtain the answer knowledge content of N searches, and identify whether the answer knowledge content of each search is null.

[0109] Specifically, question data can often be categorized into common question data and non-common question data. Common question data can be understood as questions that are relatively easy to answer, corresponding to the simple question category, and it often has a large number of answers. Non-common question data can be understood as questions that are not easy to answer, corresponding to the complex question category, and it often has fewer answers. Therefore, in a knowledge base with complex answer content, common question data is easy to find answers for, while non-common question data has limitations in terms of search speed or search breadth. Especially when searching N times, the larger the value of N, the more likely it is that there will be no N answers for non-common question data in the knowledge base, resulting in a search null value. Furthermore, even if there are N answers for non-common question data in the knowledge base, due to the maximum search time limit and the complexity of the answer data in the knowledge base, a search null value may occur when the maximum search time ends.

[0110] Step 502: If the answer knowledge content of the current search is empty, then determine whether the current search is the Nth search corresponding to the current question data.

[0111] Step 502a: If the current search is not the Nth search, then continue to execute the next search;

[0112] Step 502b: If the current search is the Nth search, then the latest current problem data is selected again;

[0113] Step 503: If the answer knowledge content of the current search is not empty, then determine whether the current search is the Nth search corresponding to the current question data.

[0114] Step 503a: If the current search is the Nth search, then the latest current problem data is selected again;

[0115] Step 503b: If the current search is not the Nth search, mark the answer knowledge content as unsearchable until the number of searches corresponding to the current question data reaches N, then change the answer knowledge content to searchable and record the search results for each search.

[0116] In this embodiment, when the obtained answer knowledge content is not null in each search, a marking method is used to mark the answer knowledge content as unsearchable in order to ensure that the next search meets the acquisition strategy. This continues until the number of searches corresponding to the current question data reaches N, at which point the answer knowledge content is changed to searchable. Specifically, since there is often a correlation or many-to-many relationship between question data and answer data, changing the answer knowledge content marked as unsearchable to searchable after N searches for the current question data does not affect the search for the next question data, thus ensuring the accuracy and scientific nature of the question data classification.

[0117] Step 403: Take the next question data of the current question data as the latest current question data, and re-execute step 402 until all question data have undergone N search operations for answer knowledge content. Once the search is complete, stop executing step 402.

[0118] Specifically, the next question data of the current question data is taken as the latest current question data. This can be done by combining the search sort sequence or the difference number result set in step 401 to filter out the next question data of the current question data as the latest current question data, thus ensuring that N search results for answer data content are performed for all the question data.

[0119] Step 304: Based on the N search results for the knowledge content of each answer corresponding to all the question data, set the difficulty category to which each of the question data belongs.

[0120] Continue to refer to Figure 6 , Figure 6 yes Figure 3 A flowchart of a specific embodiment of step 304 shown includes:

[0121] Step 601: Calculate the search results for the knowledge content of the N solutions corresponding to all the question data;

[0122] Step 602: Based on the statistical results, determine whether there are any questions where the number of answer knowledge items is less than N.

[0123] Step 603: If the answer knowledge content corresponding to all question data is N, calculate the average search time and the variance of search time for all question data when performing N searches respectively, and sort all question data based on the average search time and the variance of search time to obtain the first sorting result;

[0124] In this embodiment, the step of sorting all the problem data based on the average search time and the variance of the search time to obtain a first sorting result specifically includes: according to a preset comprehensive time representation formula: ai =b i ×ω1+c i ×ω2, calculate the comprehensive time consumption value corresponding to all the problem data respectively, where a i b represents the comprehensive time consumption characteristic value corresponding to problem data i. i ω1 represents the average search time for N searches of problem data i, and c represents the weight value corresponding to the average search time. i Let ω1 represent the search time variance value corresponding to N searches of problem data i, and let ω2 represent the weight value corresponding to the search time variance value. Let ω1 be greater than ω2. Let the comprehensive time representation values ​​corresponding to all problem data be sorted in ascending order to obtain the comprehensive time representation value sequence corresponding to all problem data. Let the corresponding problem data sequence be obtained based on the comprehensive time representation value sequence, which is used as the first sorting result.

[0125] Specifically, if the current problem data is common problem data, it is possible to quickly find different answer data content in N searches. Therefore, it can be understood that the smaller the average search time for N searches, the smaller the search time variance. If the current problem data is non-common problem data, it is possible that the time taken to find different answer data content in N searches is longer, or the next search time is longer than the previous search time, showing a continuous upward trend. In this case, the average search time for N searches will be larger, and the search time variance will also be larger. Therefore, we consider introducing the average search time and the search time variance to form a comprehensive time representation value. Based on the size of the comprehensive time representation value, we process the data in ascending order. At this time, the problem data sequence corresponding to all problem data should also be sorted from the most common problem data to the least common problem data.

[0126] Step 604: If there are fewer than N answer knowledge items corresponding to the problem data, sort the problem data according to the number of answer knowledge items in ascending order to obtain a second sorting result. Calculate the average search time and search time variance for other problem data where the number of answer knowledge items is N. Sort the other problem data based on the average search time and the search time variance to obtain a third sorting result. Combine the second sorting result and the third sorting result to obtain a fourth sorting result.

[0127] In this embodiment, the step of sorting the other problem data based on the average search time and the variance of the search time to obtain a third sorting result, and concatenating the second sorting result and the third sorting result to obtain a fourth sorting result, specifically includes: calculating the comprehensive time representation value corresponding to each of the other problem data using the comprehensive time representation formula; processing the comprehensive time representation values ​​corresponding to each of the other problem data in ascending order to obtain a sequence of comprehensive time representation values ​​corresponding to the other problem data; obtaining the corresponding problem data sequence based on the sequence of comprehensive time representation values, as the third sorting result; concatenating the end position of the sequence of the second sorting result and the end position of the sequence of the third sorting result; using the start position of the sequence of the third sorting result as the start position of the sequence of the fourth sorting result, and using the start position of the sequence of the second sorting result as the end position of the sequence of the fourth sorting result to obtain the fourth sorting result.

[0128] Specifically, if there are fewer than N answer knowledge items corresponding to a problem data point, the problem data is sorted according to the number of answer knowledge items in ascending order to obtain a second sorting result. In this case, the fewer the number of answer knowledge items, the less common the problem data is. For other problem data points with N answer knowledge items, the comprehensive time consumption value mentioned above is used to concatenate the third sorting result. In this case, the problem data at the beginning of the third sorting result sequence is the most common problem data. Therefore, the end position of the second sorting result sequence and the end position of the third sorting result sequence are considered, and the start position of the third sorting result sequence is used as the start position of the fourth sorting result sequence, and the start position of the second sorting result sequence is used as the end position of the fourth sorting result sequence to obtain the fourth sorting result. This achieves sorting from the most common problem data to the least common problem data.

[0129] Step 605: Based on the first sorting result or the fourth sorting result, divide the data into simple problem data and complex problem data.

[0130] Continue to refer to Figure 7 , Figure 7 yes Figure 6 A flowchart of a specific embodiment of step 605 shown includes:

[0131] Step 701: According to the preset splitting position, split the first sorting result or the fourth sorting result.

[0132] Specifically, the preset segmentation position is, for example, dividing the current sequence into a 9:1 ratio starting from the beginning of the sequence and moving along the end of the sequence. The current sequence includes the sequence corresponding to the first sorting result or the sequence corresponding to the fourth sorting result. Furthermore, the preset segmentation position is, for example, selecting M problem data points forward from the beginning of the sequence, starting from the end of the sequence, and using the position between the Mth problem data point and the (M+1)th problem data point as the segmentation position.

[0133] Step 702: The problem data contained from the start position of the sequence of the first sorting result or the fourth sorting result to the splitting position is taken as the simple problem data;

[0134] Step 703: The problem data contained from the end position of the sequence of the first sorting result or the fourth sorting result to the splitting position is taken as the complex problem data.

[0135] In this embodiment, after performing the step of dividing simple problem data and complex problem data according to the first sorting result or the fourth sorting result, the method further includes: deploying all the divided complex problem data as identification knowledge to the category identification component; manually labeling the answer knowledge content corresponding to all the complex problem data in the answer knowledge base, and recording the labeling results as index knowledge of the answer knowledge base in a preset query file.

[0136] By deploying all the segmented complex question data as identification knowledge to the category identification component, when actually using target question data to filter answer data content, the category identification component can prioritize identifying whether the target question data is complex question data. Correspondingly, the answer knowledge content corresponding to each of the complex question data in the answer knowledge base is marked by manual annotation, and the annotation results are recorded as index knowledge of the answer knowledge base in a preset query file. With only simple and scientific annotation, after the target question data is identified as complex question data, it can be directly queried through the index without further intent search, thus improving search efficiency. For simple question data, the intent search method is still used. This achieves the goal of searching a small amount of complex question data by marking and indexing, while using intent analysis to search for answer data for a large amount of simple question data, greatly improving the efficiency of dialogue answering and enhancing the user's dialogue experience.

[0137] Step 203: If the target question data belongs to the simple question category, then search for answer knowledge content from the preset answer knowledge base based on the intent analysis results, and generate answer feedback data based on the answer knowledge content.

[0138] Step 204: If the target question data belongs to the complex question category, then search for answer knowledge content from the answer knowledge base according to the pre-set annotation index, and generate answer feedback data based on the answer knowledge content.

[0139] In this embodiment, the question-answering method further includes using a preset data monitoring component. If the latest updated answer knowledge content in the answer knowledge base reaches a preset first update threshold, the first update threshold is, for example, a preset data volume value, such as 2GB of newly imported answer knowledge content in the answer knowledge base, where 2GB is the first update threshold; or if the latest input question data entry reaches a preset second update threshold, such as a preset number of newly added question data entries, where 200 newly added question data entries, where 200 is the second update threshold, then the method executes the following steps. Figure 3 The difficulty category setting method shown redefines the difficulty category to which all the question data belongs. This enables continuous updating of the question difficulty category as the answer knowledge content in the answer knowledge base gradually increases and / or the number of input question data entries increases. This allows for the redeployment of the recognition knowledge in the category recognition component and the re-labeling of new answer knowledge content in the answer knowledge base, ensuring continuous optimization and updating of the question-answering system and question-answering related knowledge.

[0140] This application utilizes an intent analysis model to categorize all question data into difficulty levels. It enables searching a small number of complex questions by tagging and indexing them, while using intent analysis to search for answers to a large number of simple questions. This significantly improves dialogue efficiency and enhances the user experience. Furthermore, as the answer knowledge base grows and / or the number of input question data entries increases, the difficulty levels are continuously updated. This allows for the redeployment of the identification knowledge in the category recognition component and the re-annotation of new answer knowledge in the answer knowledge base, ensuring continuous optimization and updates to the question-answering system and related knowledge.

[0141] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) refers to the theories, methods, technologies, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to obtain optimal results.

[0142] Foundational technologies for artificial intelligence generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies mainly encompass computer vision, robotics, biometrics, speech processing, natural language processing, and machine learning / deep learning.

[0143] In this embodiment, an intent analysis model is used to categorize all question data into difficulty levels. This enables searching a small number of complex questions by labeling and indexing them, while searching for answers to a large number of simple questions using intent analysis. This significantly improves the efficiency of dialogue answering and enhances the user's dialogue experience. Furthermore, as the answer knowledge base grows and / or the number of input question data entries increases, the difficulty levels are continuously updated. This allows for the redeployment of the identification knowledge in the category identification component and the re-labeling of new answer knowledge in the answer knowledge base, ensuring continuous optimization and updates to the question-answering system and related knowledge.

[0144] Further reference Figure 8 As a response to the above Figure 2 To implement the method shown, this application provides an embodiment of a question-and-answer device, which is similar to... Figure 2 Corresponding to the method embodiments shown, this device can be specifically applied to various electronic devices.

[0145] like Figure 8 As shown, the question-and-answer device 800 described in this embodiment includes: a question data acquisition module 801, a difficulty category identification module 802, a first answer feedback module 803, and a second answer feedback module 804. Wherein:

[0146] Problem data acquisition module 801 is used to acquire target problem data;

[0147] The difficulty category identification module 802 is used to input the target question data into a preset intent analysis model and determine the difficulty category to which the target question data belongs based on a preset category identification component. The difficulty category includes simple question category and complex question category.

[0148] The first answer feedback module 803 is used to search for answer knowledge content from a preset answer knowledge base based on intent analysis results if the target question data belongs to the simple question category, and generate answer feedback data based on the answer knowledge content.

[0149] The second answer feedback module 804 is used to search for answer knowledge content from the answer knowledge base according to the pre-set annotation index if the target question data belongs to the complex question category, and generate answer feedback data according to the answer knowledge content.

[0150] The question-answering device 800 described in this embodiment further includes a difficulty category setting module. This module is used to obtain all historically input question data through a preset question-answering log; to input all the question data into the ChatGLM-based language analysis model for intent analysis; to obtain N search results for each answer knowledge content corresponding to each question data from the answer knowledge base based on the intent analysis results, where N is a positive integer; and to set the difficulty category to which each question data belongs based on the N search results for each answer knowledge content corresponding to each question data.

[0151] The question-and-answer device 800 described in this embodiment also includes a difficulty category update module. The difficulty category update module is used to re-execute the difficulty category setting module and reset the difficulty category to which all the question data belong by means of a preset data monitoring component if the latest updated answer knowledge content in the answer knowledge base reaches a preset first update threshold; or if the latest input question data entry reaches a preset second update threshold.

[0152] The question-answering device proposed in this application belongs to the field of question-answering optimization technology and is applied in intelligent customer service question-answering prompt scenarios. Through an intent analysis model, all question data is categorized into difficulty levels. This enables searching by marking and indexing a small amount of complex question data, while searching for answer data for a large amount of simple question data using intent analysis. This greatly improves the efficiency of dialogue answering and enhances the user's dialogue experience. At the same time, as the answer knowledge base gradually increases and / or the number of input question data entries increases, the question difficulty categories are continuously updated. This allows for the redeployment of the recognition knowledge in the category recognition component and the re-labeling of new answer knowledge content in the answer knowledge base, ensuring continuous optimization and updates to the question-answering system and question-answering related knowledge.

[0153] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by instructing related hardware through computer-readable instructions. These computer-readable instructions can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the methods described above. The aforementioned storage medium can be a non-volatile storage medium such as a magnetic disk, optical disk, or read-only memory (ROM), or random access memory (RAM).

[0154] It should be understood that although the steps in the flowcharts of the accompanying figures are shown sequentially as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the accompanying figures may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily completed at the same time, but can be executed at different times, and their execution order is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the sub-steps or stages of other steps.

[0155] To address the aforementioned technical problems, embodiments of this application also provide a computer device. Please refer to [link / reference needed]. Figure 9 , Figure 9 This is a basic structural block diagram of the computer device in this embodiment.

[0156] The computer device 9 includes a memory 9a, a processor 9b, and a network interface 9c that are interconnected via a system bus. It should be noted that only the computer device 9 with components 9a-9c is shown in the figure; however, it should be understood that it is not required to implement all the shown components, and more or fewer components can be implemented alternatively. Those skilled in the art will understand that the computer device described here is a device capable of automatically performing numerical calculations and / or information processing according to pre-set or stored instructions, and its hardware includes, but is not limited to, microprocessors, application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), digital signal processors (DSPs), embedded devices, etc.

[0157] The computer device can be a desktop computer, laptop, handheld computer, or cloud server, etc. The computer device can interact with the user via a keyboard, mouse, remote control, touchpad, or voice control.

[0158] The memory 9a includes at least one type of readable storage medium, including flash memory, hard disk, multimedia card, card-type memory (e.g., SD or DX memory), random access memory (RAM), static random access memory (SRAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), programmable read-only memory (PROM), magnetic memory, magnetic disk, optical disk, etc. In some embodiments, the memory 9a may be an internal storage unit of the computer device 9, such as the hard disk or memory of the computer device 9. In other embodiments, the memory 9a may also be an external storage device of the computer device 9, such as a plug-in hard disk, smart media card (SMC), secure digital (SD) card, flash card, etc., equipped on the computer device 9. Of course, the memory 9a may include both the internal storage unit and its external storage device of the computer device 9. In this embodiment, the memory 9a is typically used to store the operating system and various application software installed on the computer device 9, such as computer-readable instructions for a question-and-answer method. In addition, the memory 9a can also be used to temporarily store various types of data that have been output or will be output.

[0159] In some embodiments, the processor 9b may be a central processing unit (CPU), controller, microcontroller, microprocessor, or other data processing chip. The processor 9b is typically used to control the overall operation of the computer device 9. In this embodiment, the processor 9b is used to execute computer-readable instructions stored in the memory 9a or to process data, such as executing computer-readable instructions for the question-and-answer method.

[0160] The network interface 9c may include a wireless network interface or a wired network interface, which is typically used to establish communication connections between the computer device 9 and other electronic devices.

[0161] The computer device proposed in this embodiment belongs to the field of question-and-answer optimization technology and is applied in intelligent customer service question-and-answer prompt scenarios. This application uses an intent analysis model to categorize all question data into difficulty levels. It enables searching a small amount of complex question data by tagging and indexing, while searching for answer data for a large amount of simple question data using intent analysis. This greatly improves dialogue answering efficiency and enhances the user's dialogue experience. Furthermore, as the answer knowledge base gradually increases and / or the number of input question data entries grows, the question difficulty categories are continuously updated. This allows for the redeployment of the recognition knowledge in the category recognition component and the re-annotation of new answer knowledge content in the answer knowledge base, ensuring continuous optimization and updates to the question-and-answer system and related knowledge.

[0162] This application also provides another embodiment, namely, providing a computer-readable storage medium storing computer-readable instructions that can be executed by a processor to cause the processor to perform the steps of the question-and-answer method described above.

[0163] The computer-readable storage medium proposed in this embodiment belongs to the field of question-and-answer optimization technology and is applied in intelligent customer service question-and-answer prompt scenarios. This application uses an intent analysis model to categorize all question data into difficulty levels. It enables searching a small amount of complex question data by tagging and indexing, while searching for answer data for a large amount of simple question data using intent analysis. This greatly improves dialogue answering efficiency and enhances the user's dialogue experience. Furthermore, as the answer knowledge base gradually increases and / or the number of input question data entries grows, the question difficulty categories are continuously updated. This allows for the redeployment of the recognition knowledge in the category recognition component and the re-annotation of new answer knowledge content in the answer knowledge base, ensuring continuous optimization and updates to the question-and-answer system and related knowledge.

[0164] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods of the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a storage medium (such as ROM / RAM, magnetic disk, optical disk), and includes several instructions to cause a terminal device (which may be a mobile phone, computer, server, air conditioner, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0165] Obviously, the embodiments described above are only some embodiments of this application, not all embodiments. The accompanying drawings show preferred embodiments of this application, but do not limit the patent scope of this application. This application can be implemented in many different forms; rather, the purpose of providing these embodiments is to provide a more thorough and comprehensive understanding of the disclosure of this application. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing specific embodiments, or make equivalent substitutions for some of the technical features. Any equivalent structures made using the content of this application's specification and drawings, directly or indirectly applied to other related technical fields, are similarly within the scope of patent protection of this application.

Claims

1. A question-and-answer method, characterized in that, Includes the following steps: Obtain data related to the target problem; The method further includes inputting the target question data into a preset intent analysis model, determining the difficulty category of the target question data based on a preset category recognition component, wherein the difficulty category includes simple question category and complex question category, and the preset intent analysis model includes a language analysis model based on ChatGLM. Before performing the step of determining the difficulty category of the target question data based on the preset category recognition component, the method further includes: All historical question data is obtained through a pre-set question and answer log; All the question data are input into the ChatGLM-based language analysis model for intent analysis. Based on the intent analysis results, retrieve N search results for the corresponding answer knowledge content for each question from the answer knowledge base, where N is a positive integer, specifically including: Step 401: Select the current problem data from all the problem data; Step 402: Based on the intent analysis results and acquisition strategy corresponding to the current question data, retrieve N search results for answer knowledge content corresponding to the current question data from the answer knowledge base. The search results for answer knowledge content include the answer knowledge content obtained in each search and the search time in each search. Step 403: Take the next question data of the current question data as the latest current question data, and re-execute step 402 until all question data have undergone N search of answer knowledge content. After the search is completed, stop executing step 402. Based on the N search results for the knowledge content of each answer corresponding to all the question data, the difficulty category to which each of the question data belongs is set; Before performing the step of obtaining N search results for each of the N questions from the answer knowledge base based on the intent analysis results, the method further includes: pre-setting an acquisition strategy for the N answers, wherein the acquisition strategy includes that the N answers are not repeated and that a maximum search time is preset for each search; If the target question data belongs to the simple question category, then the solution knowledge content is searched from the preset solution knowledge base based on the intent analysis results, and solution feedback data is generated based on the solution knowledge content; If the target question data belongs to the category of complex questions, then the solution knowledge content is searched from the solution knowledge base according to the pre-set annotation index, and solution feedback data is generated based on the solution knowledge content.

2. The question-and-answer method according to claim 1, characterized in that, The step of obtaining N search results for answer knowledge content corresponding to the current question data from the answer knowledge base based on the intent analysis result corresponding to the current question data and the acquisition strategy specifically includes: Based on the intent analysis results corresponding to the current question data and the acquisition strategy, obtain the answer knowledge content of N searches, and identify whether the answer knowledge content of each search is null. If the answer to the current search is null, then determine whether the current search is the Nth search corresponding to the current question data. If the current search is not the Nth search, then continue to execute the next search; If the current search is the Nth search, then the latest data for the current problem is selected again; If the answer to the current search is not empty, then determine whether the current search is the Nth search corresponding to the current question data. If the current search is the Nth search, then the latest current search is selected again. Problem data; If the current search is not the Nth search, the solution knowledge content is marked as unsearchable until the number of searches corresponding to the current question data reaches N. Then, the solution knowledge content is changed to searchable, and the search results for each search are recorded.

3. The question-and-answer method according to claim 1, characterized in that, The step of determining the difficulty category of each of the N solutions to the knowledge content search results corresponding to all the question data specifically includes: Statistically analyze the search results for the N solutions to each question; Based on the statistical results, determine whether there are any problematic data points for which the corresponding answer knowledge content is less than N; If all the question data corresponds to N pieces of answer knowledge content, then calculate the average search time and the variance of search time for each of the N searches for all the question data, and sort all the question data based on the average search time and the variance of search time to obtain the first sorting result; If there are fewer than N answer knowledge items corresponding to a question, the question data is sorted according to the number of answer knowledge items in ascending order to obtain a second sorting result. For other question data where there are N answer knowledge items, the average search time and the variance of the search time are calculated respectively. The other question data is sorted based on the average search time and the variance of the search time to obtain a third sorting result. The second sorting result and the third sorting result are concatenated to obtain a fourth sorting result. Based on the first sorting result or the fourth sorting result, simple problem data and complex problem data are divided.

4. The question-and-answer method according to claim 3, characterized in that, The step of sorting all the problem data based on the average search time and the variance of the search time to obtain a first sorting result specifically includes: According to the preset comprehensive time consumption formula: ， Calculate the comprehensive time consumption value corresponding to all the problem data, where, Representing problem data The corresponding comprehensive time consumption value, Representing problem data The average search time for performing N searches. This represents the weight value corresponding to the average search time. Representing problem data The variance of search time for performing N searches. This represents the weight value corresponding to the variance of the search time. Greater than ; The comprehensive time consumption values ​​corresponding to all the problem data are sorted in ascending order to obtain a sequence of comprehensive time consumption values ​​corresponding to all the problem data. Based on the comprehensive time consumption characterization value sequence, the corresponding problem data sequence is obtained as the first sorting result; The step of sorting the other problem data based on the average search time and the variance of the search time to obtain a third sorting result, and then concatenating the second sorting result and the third sorting result to obtain a fourth sorting result, specifically includes: The comprehensive time consumption representation formula is used to calculate the comprehensive time consumption representation values ​​corresponding to the other problem data respectively; The comprehensive time consumption values ​​corresponding to the other problem data are sorted in ascending order to obtain a sequence of comprehensive time consumption values ​​corresponding to the other problem data. Based on the comprehensive time consumption characterization value sequence, the corresponding problem data sequence is obtained as the third sorting result; The sequence end position of the second sorting result is concatenated with the sequence end position of the third sorting result; The fourth sorting result is obtained by taking the start position of the third sorting result as the start position of the fourth sorting result and taking the start position of the second sorting result as the end position of the fourth sorting result.

5. The question-and-answer method according to claim 3, characterized in that, The step of dividing simple problem data and complex problem data according to the first sorting result or the fourth sorting result specifically includes: The first sorting result or the fourth sorting result is segmented according to the preset segmentation position; The problem data contained from the start position of the sequence of the first sorting result or the fourth sorting result to the splitting position is taken as the simple problem data; The problem data contained from the end position of the sequence of the first sorting result or the fourth sorting result to the split position is taken as the complex problem data; After performing the step of dividing simple problem data and complex problem data according to the first sorting result or the fourth sorting result, the method further includes: All the complex problem data identified are used as recognition knowledge and deployed to the category recognition component. The solution knowledge content corresponding to all the complex question data in the solution knowledge base is marked by manual annotation, and the annotation results are recorded as index knowledge of the solution knowledge base in a preset query file.

6. A question-and-answer device, characterized in that, The question-answering device implements the steps of the question-answering method as described in any one of claims 1 to 5, and the question-answering device comprises: The problem data acquisition module is used to acquire target problem data; The difficulty category identification module is used to input the target question data into a preset intent analysis model and determine the difficulty category to which the target question data belongs based on a preset category identification component. The difficulty category includes simple question category and complex question category. The first answer feedback module is used to search for answer knowledge content from a preset answer knowledge base based on intent analysis results if the target question data belongs to the simple question category, and generate answer feedback data based on the answer knowledge content. The second answer feedback module is used to search for answer knowledge content from the answer knowledge base according to the pre-set annotation index if the target question data belongs to the complex question category, and generate answer feedback data based on the answer knowledge content.

7. A computer device, characterized in that, The method includes a memory and a processor, wherein the memory stores computer-readable instructions, and the processor executes the computer-readable instructions to implement the steps of the question-and-answer method as described in any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-readable instructions, which, when executed by a processor, implement the steps of the question-and-answer method as described in any one of claims 1 to 5.

Images