Question processing method and device based on artificial intelligence, computer device and medium

Abstract

This application belongs to the fields of artificial intelligence and fintech, and relates to an AI-based question processing method, including: converting user-input voice questions into text to obtain a first text; preprocessing the first text to obtain a second text; performing intent recognition on the second text based on a large language model to obtain a target intent; selecting a target thought chain from all thought chains based on the processing complexity of the target intent and the second text; performing execution and reasoning processing according to the target thought chain through the large language model, and outputting answer data corresponding to the target intent; and providing feedback to the user based on the answer data. This application also provides an AI-based question processing device, computer equipment, and storage medium. Furthermore, the answer data can be stored in a blockchain. This application can be applied to dialogue processing scenarios in the financial field, and the use of a large language model with a thought chain library effectively improves the accuracy of answers to voice questions.

Description

Technical Field

[0001] This application relates to the fields of artificial intelligence development technology and financial technology, and in particular to artificial intelligence-based question processing methods, devices, computer equipment and storage media. Background Technology

[0002] With the rapid development of artificial intelligence technology, intelligent question-and-answer systems for human-computer interaction have been widely used in the financial field, such as for communicating insurance plans and home purchase plans.

[0003] Currently, the intelligent question-answering systems used by financial enterprises, such as insurance companies, are typically based on large language models. The process of processing user-input question data in such systems is usually as follows: the intelligent question-answering system uses a large language model to extract abstract features from existing information, generating vector representations, and storing them in a database; then, it uses the large language model to extract features from the user's question data, generating vector representations, and searches the database for content matching the question using vector recall; finally, it uses the large language model to process the question and search results to obtain the corresponding answer. However, this question processing method has the following problems: due to the diverse formats of user-input questions in various usage scenarios, the vector recall accuracy of large language models is relatively low for complex questions, resulting in low accuracy of the intelligent question-answering system's answers to the question data. Summary of the Invention

[0004] The purpose of this application is to propose an artificial intelligence-based question processing method, apparatus, computer device, and storage medium to solve the technical problem that existing question processing methods suffer from low accuracy in answering question data due to the diverse formats of user-input questions in various usage scenarios and the relatively low vector recall accuracy of large language models for complex questions.

[0005] To address the aforementioned technical problems, this application provides an artificial intelligence-based question processing method, employing the following technical solution:

[0006] The received user-input voice question is converted into text to obtain the first text;

[0007] The first text is preprocessed to obtain the second text;

[0008] The intent of the second text is identified based on a pre-set large language model to obtain the target intent corresponding to the second text; wherein, the large language model is equipped with a thought chain library generated based on a pre-set construction method, and the thought chain library includes a certain number of thought chains.

[0009] Determine the processing complexity corresponding to the second text;

[0010] Based on the target intent and the processing complexity, the target thought chain is selected from all thought chains contained in the thought chain library;

[0011] The large language model is used to perform execution and reasoning processing according to the target thought chain, and output answer data corresponding to the target intent.

[0012] Feedback is processed on the user based on the answer data.

[0013] Furthermore, the step of selecting the target thought chain from all thought chains contained in the thought chain library based on the target intent and the processing complexity specifically includes:

[0014] Obtain the intent identification information corresponding to each of the aforementioned thought chains;

[0015] Based on the intent identification information, a first thought chain that matches the target intent is obtained from all the thought chains;

[0016] Determine whether the number of the first thought chain includes multiple chains;

[0017] If so, a second thought chain matching the processing complexity is selected from the first thought chain based on preset filtering rules;

[0018] The second thinking chain is taken as the target thinking chain.

[0019] Furthermore, the step of selecting a second thought chain from the first thought chain that matches the processing complexity based on preset filtering rules specifically includes:

[0020] Determine whether the processing complexity is high.

[0021] If the processing complexity is high, then select the third thinking chain with the longest chain length from all the first thinking chains, and use the third thinking chain as the second thinking chain.

[0022] If the processing complexity is not high, then determine whether the processing complexity is low.

[0023] If the processing complexity is low, then select the fourth thinking chain with the smallest chain length from all the first thinking chains, and use the fourth thinking chain as the second thinking chain.

[0024] If the processing complexity is not low, then the processing complexity is determined to be medium, and the usage frequency value of the fifth thinking chain is obtained; wherein, the fifth thinking chain is all the other thinking chains in the first thinking chain except for the third thinking chain and the fourth thinking chain.

[0025] The sixth thinking chain with the highest usage frequency is selected from the fifth thinking chain, and the sixth thinking chain is used as the second thinking chain.

[0026] Furthermore, the step of determining the processing complexity corresponding to the second text specifically includes:

[0027] The second text is segmented into sentences to obtain corresponding sentences, and the first number of the sentences is counted.

[0028] The second text is segmented to obtain the corresponding words, and the second number of the words is counted.

[0029] Based on the first quantity and the second quantity, the complexity score of the second text is calculated based on a preset calculation strategy;

[0030] Based on the complexity score, a preset mapping data table is queried to retrieve the processing complexity corresponding to the complexity score from the mapping data table.

[0031] Furthermore, the step of calculating the complexity score of the second text based on the first quantity and the second quantity, and based on a preset calculation strategy, specifically includes:

[0032] Determine the preset weight calculation algorithm;

[0033] The first weight corresponding to the number of sentences is determined based on the weighting algorithm described above;

[0034] The second weight corresponding to the number of words is determined based on the weighting algorithm described above;

[0035] Based on the first weight and the second weight, a preset score calculation formula is called to calculate the first quantity and the second quantity to obtain the corresponding calculation result;

[0036] The calculation result is used as the complex score of the second text.

[0037] Furthermore, before the step of performing intent recognition on the second text based on a preset large language model to obtain the target intent corresponding to the second text, the method further includes:

[0038] Determine whether a mind chain library construction request triggered by a specified user has been received;

[0039] If so, display the preset build page;

[0040] Receive the thought chain data input by the specified user on the construction page;

[0041] Construct a corresponding mind chain library based on the aforementioned mind chain data;

[0042] The thought chain library is stored and processed.

[0043] Furthermore, after the step of processing feedback to the user based on the answer data, the method further includes:

[0044] Acquire pre-collected performance test data;

[0045] The performance of the large language model is evaluated based on the performance test data to obtain the corresponding model performance data.

[0046] Obtain the basic model performance data corresponding to the large language model;

[0047] The performance data of the model is compared and analyzed with the performance data of the basic model to obtain the corresponding comparison and analysis results.

[0048] The large language model is adjusted based on the comparison analysis results.

[0049] To address the aforementioned technical problems, this application also provides an artificial intelligence-based question processing device, employing the following technical solution:

[0050] The conversion module is used to convert the received user-input voice question into text to obtain the first text.

[0051] The preprocessing module is used to preprocess the first text to obtain the second text;

[0052] The recognition module is used to perform intent recognition on the second text based on a preset large language model to obtain the target intent corresponding to the second text; wherein, the large language model is equipped with a thought chain library generated based on a preset construction method, and the thought chain library includes a certain number of thought chains.

[0053] A determination module is used to determine the processing complexity corresponding to the second text;

[0054] The filtering module is used to filter out the target thinking chain from all thinking chains contained in the thinking chain library based on the target intent and the processing complexity.

[0055] The first processing module is used to perform execution and reasoning processing according to the target thought chain through the large language model, and output answer data corresponding to the target intent;

[0056] The second processing module is used to provide feedback to the user based on the answer data.

[0057] To address the aforementioned technical problems, this application also provides a computer device that employs the following technical solution:

[0058] The received user-input voice question is converted into text to obtain the first text;

[0059] The first text is preprocessed to obtain the second text;

[0060] The intent of the second text is identified based on a pre-set large language model to obtain the target intent corresponding to the second text; wherein, the large language model is equipped with a thought chain library generated based on a pre-set construction method, and the thought chain library includes a certain number of thought chains.

[0061] Determine the processing complexity corresponding to the second text;

[0062] Based on the target intent and the processing complexity, the target thought chain is selected from all thought chains contained in the thought chain library;

[0063] The large language model is used to perform execution and reasoning processing according to the target thought chain, and output answer data corresponding to the target intent.

[0064] Feedback is processed on the user based on the answer data.

[0065] To address the aforementioned technical problems, this application also provides a computer-readable storage medium, employing the technical solution described below:

[0066] The received user-input voice question is converted into text to obtain the first text;

[0067] The first text is preprocessed to obtain the second text;

[0068] The intent of the second text is identified based on a pre-set large language model to obtain the target intent corresponding to the second text; wherein, the large language model is equipped with a thought chain library generated based on a pre-set construction method, and the thought chain library includes a certain number of thought chains.

[0069] Determine the processing complexity corresponding to the second text;

[0070] Based on the target intent and the processing complexity, the target thought chain is selected from all thought chains contained in the thought chain library;

[0071] The large language model is used to perform execution and reasoning processing according to the target thought chain, and output answer data corresponding to the target intent.

[0072] Feedback is processed on the user based on the answer data.

[0073] Compared with the prior art, the embodiments of this application have the following main advantages:

[0074] This application first performs text conversion processing on the received user-inputted question speech to obtain a first text; then, it preprocesses the first text to obtain a second text; next, it performs intent recognition on the second text based on a preset large language model to obtain the target intent corresponding to the second text; subsequently, it determines the processing complexity corresponding to the second text; further, based on the target intent and the processing complexity, it filters out a target thought chain from all thought chains contained in the thought chain library; and then, through the large language model, it performs execution and reasoning processing according to the target thought chain to output answer data corresponding to the target intent; finally, it provides feedback processing to the user based on the answer data. After performing text conversion and preprocessing on the user-inputted question speech, this application first uses a large language model equipped with a thought chain library generated based on a preset construction method to perform intent recognition on the second text to obtain the target intent. Then, it intelligently filters out a matching target thought chain from all thought chains contained in the thought chain library based on the processing complexity corresponding to the second text and the target intent. Finally, through the large language model, it performs execution and reasoning processing according to the obtained target thought chain to output answer data corresponding to the target intent and provides feedback to the user. This application uses an optimized large language model loaded with a thought chain library generated based on a preset construction method to process user-inputted voice questions. This enables the output of accurate and controllable answer data for user-inputted voice questions, effectively improving the accuracy of answering voice questions. Attached Figure Description

[0075] 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.

[0076] Figure 1 This is an exemplary system architecture diagram to which this application can be applied;

[0077] Figure 2 A flowchart of an embodiment of the AI-based question processing method according to this application;

[0078] Figure 3 This is a schematic diagram of the structure of an embodiment of the AI-based question processing device according to this application;

[0079] Figure 4 This is a schematic diagram of the structure of one embodiment of the computer device according to this application. Detailed Implementation

[0080] 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.

[0081] 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.

[0082] 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.

[0083] 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.

[0084] 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.

[0085] 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.

[0086] 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.

[0087] It should be noted that the AI-based question processing method provided in this application is generally executed by a server / terminal device, and correspondingly, the AI-based question processing device is generally located in the server / terminal device.

[0088] It should be understood that Figure 1 The 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.

[0089] Continue to refer to Figure 2 This document illustrates a flowchart of an embodiment of the AI-based question processing method according to this application. The order of steps in the flowchart can be changed, and some steps can be omitted, depending on different needs. The AI-based question processing method provided in this application can be applied to any scenario requiring dialogue interaction, and thus can be applied to products in these scenarios, such as insurance plan communication, home purchase plan communication, and treatment plan communication in financial insurance scenarios or digital healthcare scenarios. The AI-based question processing method includes the following steps:

[0090] Step S201: Perform text conversion processing on the received user-inputted question voice to obtain the first text.

[0091] In this embodiment, the AI-based question processing method runs on an electronic device (e.g., Figure 1The server / terminal device shown can acquire the question voice via wired or wireless connection. It should be noted that the aforementioned wireless connection methods may include, but are not limited to, 3G / 4G / 5G connections, Wi-Fi connections, Bluetooth connections, WiMAX connections, Zigbee connections, UWB (ultra-wide wireless band) connections, and other currently known or future known wireless connection methods. The question voice refers to voice-formatted question data input by the user to interact with the large language model. The question voice can be in any format, such as questions with different sentence structures or questions containing redundant information; the question voice can also be a complex question. For example, the content of the question voice may include: "I drank with a client and developed a stomach hemorrhage; is this considered a work-related injury?" Furthermore, the question voice can be converted into first text using Automatic Speech Recognition (ASR) technology.

[0092] Step S202: Preprocess the first text to obtain the second text.

[0093] In this embodiment, the preprocessing described above may include text cleaning, which may include removing noise, special characters, HTML tags, and other irrelevant information from the first text.

[0094] Step S203: Based on a preset large language model, perform intent recognition on the second text to obtain the target intent corresponding to the second text; wherein, the large language model is equipped with a thought chain library generated based on a preset construction method, and the thought chain library includes a certain number of thought chains.

[0095] In this embodiment, after the second text is input into the large language model, its deep learning architecture is used to perform deep analysis to understand its meaning and context. Based on the text content, the user's intent, such as querying information, expressing emotion, or requesting services, is then identified. The thought chain library is integrated as an optional module of the large language model to ensure its support. Furthermore, the preset construction method can include manually editing and building the thought chain library, or it can include extracting thought chains from historical cases using the large language model, subjecting these thought chains to manual quality control before building the library. This preset construction method differs from existing methods where the large language model automatically generates thought chains. Additionally, the specific values ​​for the aforementioned number are not limited and can be set according to actual construction needs.

[0096] Step S204: Determine the processing complexity corresponding to the second text.

[0097] In this embodiment, the specific implementation process of determining the processing complexity corresponding to the second text will be further described in detail in subsequent specific embodiments of this application, and will not be elaborated on here.

[0098] Step S205: Based on the target intent and the processing complexity, select the target thought chain from all thought chains contained in the thought chain library.

[0099] In this embodiment, the specific implementation process of selecting the target thought chain from all thought chains contained in the thought chain library based on the target intent and the processing complexity will be further described in detail in subsequent specific embodiments of this application, and will not be elaborated on here.

[0100] Step S206: Using the large language model, perform execution and reasoning processing according to the target thought chain, and output answer data corresponding to the target intent.

[0101] In this embodiment, after selecting the target thought chain that matches both the target intent and the processing complexity from all thought chains contained in the thought chain library, the subsequent large language model will perform execution and reasoning processing according to the target thought chain and output answer data corresponding to the target intent, so as to achieve accurate and controllable answer data for the user's input question voice output, and the use of the thought chain library can ensure that each round of decision-making in the conversation can be answered accurately and controllably.

[0102] Step S207: Feedback processing is performed on the user based on the answer data.

[0103] In this embodiment, after obtaining the aforementioned answer data, further grammar detection and fluency optimization processing can be performed on the answer data to obtain target answer data. Subsequently, the target answer data will be used to provide feedback to the user. The target answer data is in text format. If it is necessary to output the generated target answer data to the user in speech format, text-to-speech (TTS) technology is further used to convert the text into corresponding answer speech data, and this answer speech data is returned to the user, thereby efficiently, accurately, and naturally completing the dialogue interaction processing corresponding to the user's input question speech.

[0104] This application, after performing text conversion and preprocessing on the user-inputted voice question, first uses a large language model equipped with a thought chain library generated based on a preset construction method to perform intent recognition on the second text to obtain the target intent. Then, it intelligently selects the matching target thought chain from all thought chains contained in the thought chain library based on the processing complexity of the second text and the target intent. Finally, the large language model performs execution and reasoning processing according to the obtained target thought chain, outputting answer data corresponding to the target intent and feeding it back to the user. This application, by using an optimized large language model equipped with a thought chain library generated based on a preset construction method to process user-inputted voice questions, can achieve accurate and controllable answer data output for user-inputted voice questions, effectively improving the accuracy of answering voice questions.

[0105] In some alternative implementations, step S205 includes the following steps:

[0106] Obtain the intent identifier information corresponding to each of the aforementioned thought chains.

[0107] In this embodiment, during the construction of the thought chain library, for each thought chain in the library, intent identifier information corresponding to each thought chain is generated based on the problem type information corresponding to each thought chain. Each thought chain may have more than one type of intent tag information.

[0108] Based on the intent identification information, a first thought chain that matches the target intent is obtained from all the thought chains.

[0109] In this embodiment, the matching degree between the target intent and each of the thought chains can be determined by matching the target intent with the intent identifier information of each thought chain. Then, a designated thought chain with a matching degree greater than a preset matching threshold is selected from all thought chains and designated as the first thought chain matching the target intent. A similarity measurement algorithm can be used to calculate the matching degree between the target intent and each thought chain. The selection of the similarity measurement algorithm is not specifically limited and can be set according to actual business needs.

[0110] Determine whether the number of the first thought chain includes multiple chains.

[0111] In this embodiment, the number of the first thought chains may include one or more. If the number of the first thought chains is one, then the first thought chain is directly used as the target thought chain.

[0112] If so, a second thought chain matching the processing complexity is selected from the first thought chain based on preset filtering rules.

[0113] In this embodiment, the specific implementation process of selecting a second thinking chain that matches the processing complexity from the first thinking chain based on the preset filtering rules will be further described in detail in subsequent specific embodiments of this application, and will not be elaborated on here.

[0114] The second thinking chain is taken as the target thinking chain.

[0115] Based on the target intent corresponding to the question speech and the processing complexity of the question speech, this application can adaptively and accurately select the target thought chain for processing the answer to the user-input question speech from all thought chains contained in the thought chain library. This enables more effective use of thought chain technology to improve the performance of the large language model in question feedback processing, thereby improving the processing efficiency of the large language model in processing the user-input question speech and ensuring the accuracy of the generated answer data corresponding to the target intent.

[0116] In some optional implementations of this embodiment, the step of selecting a second thought chain from the first thought chain that matches the processing complexity based on a preset filtering rule includes the following steps:

[0117] Determine whether the processing complexity is high.

[0118] In this embodiment, the processing complexity can include high complexity, medium complexity, and low complexity.

[0119] If the processing complexity is high, then the third thinking chain with the longest chain length is selected from all the first thinking chains, and the third thinking chain is used as the second thinking chain.

[0120] In this embodiment, if the processing complexity is high, the processing task corresponding to the aforementioned question voice is determined to be a high-complexity task. Then, the third thought chain with the longest chain length is intelligently selected from all the first thought chains, and this third thought chain is used as the target thought chain for processing the answer to the user-input question voice. For high-complexity tasks, using the thought chain with the longest chain length provides more reasoning steps and more detailed explanations, thereby effectively improving the accuracy of the answer data generated by the large language model corresponding to the target intent.

[0121] If the processing complexity is not high, then determine whether the processing complexity is low.

[0122] In this embodiment, if the processing complexity is not high, then the processing complexity can be medium or low.

[0123] If the processing complexity is low, then the fourth thinking chain with the smallest chain length is selected from all the first thinking chains, and the fourth thinking chain is used as the second thinking chain.

[0124] In this embodiment, if the processing complexity is low, the processing task corresponding to the aforementioned question speech is determined to be a low-complexity task. Then, the fourth thought chain with the shortest chain length is intelligently selected from all the first thought chains, and this fourth thought chain is used as the target thought chain for processing the answer to the user-input question speech. For low-complexity tasks, by using the thought chain with the shortest chain length, because the user-input question is relatively simple, fewer intermediate steps are needed, thereby effectively improving the performance of the large language model in question feedback processing.

[0125] If the processing complexity is not low, then the processing complexity is determined to be medium, and the usage frequency value of the fifth thought chain is obtained.

[0126] In this embodiment, the fifth thought chain refers to all the thought chains in the first thought chain except for the third and fourth thought chains. The usage frequency value can be obtained by calculating the quotient of the total number of times the fifth thought chain is used in a month and 30.

[0127] The sixth thinking chain with the highest usage frequency is selected from the fifth thinking chain, and the sixth thinking chain is used as the second thinking chain.

[0128] In this embodiment, the highest specified usage frequency value can be determined by comparing the usage frequency values ​​of all the obtained fifth thinking chains, and then the sixth thinking chain corresponding to the specified usage frequency value can be used as the second thinking chain.

[0129] This application, after obtaining a first thought chain matching the target intent from all thought chains based on the intent identifier information, if it detects that there are multiple first thought chains, it intelligently and accurately filters out a second thought chain from the first thought chains for processing the user-inputted question speech based on the processing complexity of the question speech, ensuring the accuracy of the obtained second thought chain. Furthermore, it achieves more effective utilization of thought chain technology to improve the performance of the large language model in question feedback processing, thereby improving the processing efficiency of the large language model in processing user-inputted question speech feedback, and ensuring the accuracy of the generated answer data corresponding to the target intent.

[0130] In some alternative implementations, step S204 includes the following steps:

[0131] The second text is segmented into sentences to obtain corresponding sentences, and the first number of the sentences is counted.

[0132] In this embodiment, the second text can be segmented into sentences by obtaining the speech pause position information corresponding to the user's input question, and based on the speech pause position information, the second text can be segmented into sentences.

[0133] The second text is segmented to obtain the corresponding words, and the second number of the words is counted.

[0134] In this embodiment, the second text can be segmented into words by using a general word segmentation tool, such as ji eba.

[0135] Based on the first quantity and the second quantity, the complexity score of the second text is calculated based on a preset calculation strategy.

[0136] In this embodiment, the specific implementation process of calculating the complex score of the second text based on the first quantity and the second quantity and based on a preset calculation strategy will be further described in detail in subsequent specific embodiments of this application, and will not be elaborated on here.

[0137] Based on the complexity score, a preset mapping data table is queried to retrieve the processing complexity corresponding to the complexity score from the mapping data table.

[0138] In this embodiment, the aforementioned mapping data table is a pre-constructed data table storing a one-to-one correspondence between score ranges and complexities, based on actual complexity statistical analysis requests. Specifically, a specified score range corresponding to the complexity score can be determined first from the mapping data table, and then a specified complexity corresponding to the specified score range can be retrieved from the mapping database, with this specified complexity serving as the processing complexity corresponding to the complexity score. The complexity can include high complexity, medium complexity, and low complexity.

[0139] This application performs sentence segmentation and word segmentation on the second text, and then calculates the complexity score of the second text based on the first number of sentences, the second number of words, and the calculation strategy. Furthermore, based on the use of a mapping data table, it can quickly and accurately query the processing complexity corresponding to the second text, ensuring the accuracy of the obtained processing complexity data.

[0140] In some optional implementations, the step of calculating the complexity score of the second text based on the first quantity and the second quantity, according to a preset calculation strategy, includes the following steps:

[0141] Determine the preset weight calculation algorithm.

[0142] In this embodiment, the selection of the above weight calculation algorithm is not specifically limited. For example, any one of the following algorithms can be used: analytic hierarchy process, entropy method, principal component analysis, independence weight method, factor analysis, information content weight method, etc. Alternatively, two or three algorithms can be randomly selected from multiple algorithms and combined as the above weight selection algorithm.

[0143] The first weight corresponding to the number of sentences is determined based on the weighting algorithm.

[0144] In this embodiment, the number of sentences can be weighted using the weighting algorithm described above to generate a first weight corresponding to the number of sentences.

[0145] The second weight corresponding to the number of words is determined based on the weighting algorithm.

[0146] In this embodiment, the word count can be weighted using the weighting algorithm described above to generate a second weight corresponding to the word count.

[0147] Based on the first weight and the second weight, a preset score calculation formula is called to calculate the first quantity and the second quantity to obtain the corresponding calculation result.

[0148] In this embodiment, the above-mentioned score calculation formula can specifically adopt a weighted summation formula. The calculation result can be obtained by substituting the first weight, the second weight, the first quantity, and the second quantity into the corresponding positions in the score calculation formula.

[0149] The calculation result is used as the complex score of the second text.

[0150] This application effectively ensures the rationality and accuracy of the generated data, namely the first weight corresponding to the number of sentences and the second weight corresponding to the number of words, by using a weighting algorithm. Furthermore, based on the obtained first and second weights, a scoring formula is used to calculate and process the first and second quantities, enabling the rapid and accurate generation of complex scores for the second text. This improves the efficiency of complex score generation and ensures the accuracy of the obtained complex scores.

[0151] In some optional implementations of this embodiment, before step S203, the electronic device may further perform the following steps:

[0152] Determine whether a mind chain library construction request triggered by a specified user has been received.

[0153] In this embodiment, the designated user refers to staff involved in the operation and maintenance of the large language model. The aforementioned thought chain library construction request is a processing request triggered by the designated user and generated based on the actual business need for manually constructing thought chains.

[0154] If so, display the preset build page.

[0155] In this embodiment, the aforementioned construction page is a pre-built, editable page designed to assist relevant users in constructing a mind chain library. A construction reminder box can also be set on the construction page, displaying pre-defined mind chain construction guidance information. This guidance information is pre-written and generated based on actual mind chain construction needs, guiding relevant users on how to perform standardized mind chain construction and generation processes. Furthermore, the construction of the mind chain library requires the generation of multiple mind chains. A mind chain consists of four parts: Prompt, SOP, Knowledge, and Tools. The Prompt is a model hint instruction; the SOP is a standard operating procedure for solving a specific type of intent query; and SOP execution relies on external knowledge and tools (APIs). For the generation of each mind chain, pre-built data input sections corresponding to the Prompt, SOP, Knowledge, and Tools are provided. After a relevant user completes the data input for a mind chain, they can complete the construction of that mind chain by clicking submit. Then, the system receives the relevant data for the next mind chain input from the user and completes the construction of that corresponding mind chain. Finally, all the generated thought chains are integrated to obtain a thought chain library. Furthermore, knowledge retrieval enhancement processing can be performed on the knowledge within the thought chains using QA knowledge, document knowledge, graph knowledge, and table knowledge, thereby improving the knowledge richness of the thought chain library.

[0156] Receive the thought chain data input by the specified user on the construction page.

[0157] In this embodiment, users can input the specific thought chain data corresponding to the thought chain to be generated on the building page according to their actual thought chain generation needs.

[0158] A corresponding thought chain library is constructed based on the thought chain data.

[0159] In this embodiment, a corresponding thought chain can first be constructed based on the thought chain data input by a specified user. Then, all the obtained thought chains are integrated and processed to obtain a corresponding thought chain library. In addition, after the construction of the thought chain library is completed, the logical relationships and reasoning processes in the thought chains within the library can be further checked to see if they are reasonable. The thought chains can also be corrected and optimized based on expert opinions or user feedback.

[0160] The thought chain library is stored and processed.

[0161] In this embodiment, the storage method of the MindChain library is not specifically limited and can be set according to actual usage needs. For example, it can be stored using blockchain, local database, or cloud server.

[0162] This application enables the automatic, rapid, and accurate construction of a mind chain library that meets the user's individual construction needs, based on the construction page and the use of user-inputted mind chain data. This effectively reduces the workload of mind chain library construction, simplifies the process, and improves the intelligence and efficiency of mind chain library construction.

[0163] In some optional implementations of this embodiment, the aforementioned thought chain library is not limited to manual compilation; it can also use thought chains extracted from historical cases by a large language model, and these thought chains undergo manual quality inspection. The specific implementation process may include: 1) First, collecting a large amount of historical case data previously processed by the large language model, including case descriptions, case handling results, and relevant processing basis. Then, preprocessing the historical case data, including text cleaning, word segmentation, and stop word removal, so that the large language model can better understand and process this historical case data. 2) Training the large language model using historical case data, allowing the large language model to learn how to extract key information from case data, understand the logical relationships and reasoning processes of cases. 3) Designing extraction rules for extracting thought chains from text generated by the large language model. Extraction rules can be set based on factors such as language structure and logical reasoning. 4) Evaluating and optimizing the extracted thought chains. Manual evaluation can be used, or automated indicators can be designed to measure the quality of the thought chains. Based on the evaluation results, the large language model is fine-tuned or optimized to improve the accuracy and completeness of thought chain extraction. For each historical case, the large language model generates relevant text descriptions or analyses, and then the thought chains are extracted from the text using the extraction rules mentioned above.

[0164] In some optional implementations of this embodiment, after step S207, the electronic device may further perform the following steps:

[0165] Obtain pre-collected performance test data.

[0166] In this embodiment, the aforementioned performance test data is pre-collected speech data used for performance testing of the large language model. Specifically, the same performance test data is used beforehand to perform performance testing on an initial large language model that does not have the aforementioned thought chain library attached, and the performance data of the basic model corresponding to the initial large language model is collected.

[0167] The performance of the large language model is evaluated based on the performance test data to obtain the corresponding model performance data.

[0168] In this embodiment, the large language model is a model that has been modified from the initial large language model by adding a thought chain library generated based on a preset construction method. By inputting the performance test data into the large language model, model performance data corresponding to preset key performance indicators is collected during the data processing of the performance test data.

[0169] Obtain the basic model performance data corresponding to the large language model.

[0170] In this embodiment, key performance indicators for reflecting the model performance of the large language model are predetermined, such as accuracy, F1 score, recall, etc. Before the large language model is attached to the above-mentioned thought chain library, the large language model at this time is called the initial large language model. The above-mentioned performance test data is used to perform performance tests on the initial large language model, and the baseline values ​​of the key performance indicators are recorded to obtain the corresponding model performance data.

[0171] The performance data of the model is compared and analyzed with the performance data of the basic model to obtain the corresponding comparison and analysis results.

[0172] In this embodiment, the performance data of the model is compared and analyzed with the performance data of the base model to evaluate whether the performance of the large language model is improved after attaching the thought chain library. The comparison and analysis results include: the performance data of the model is better than the performance data of the base model, or the performance data of the model is worse than the performance data of the base model.

[0173] The large language model is adjusted based on the comparison analysis results.

[0174] In this embodiment, if the performance data of the large language model after attaching the thought chain library is improved compared to the performance data of the basic model (i.e., the model performance data is better than the basic model performance data), then attaching the thought chain library is determined to be an effective model optimization strategy for the large language model, and no further processing will be performed on the large language model. Conversely, if the performance data of the large language model after attaching the thought chain library shows a performance decrease or only a minor improvement compared to the basic model performance data (i.e., the model performance data is worse than the basic model performance data), then attaching the thought chain library is determined to be an effective model optimization strategy. However, if attaching the thought chain library is deemed an ineffective model optimization strategy, then subsequent adjustments will be made to the large language model based on an adjustment strategy. Specifically, if the performance improvement of the large language model after attaching the thought chain library is minor, the length or complexity of the thought chains in the thought chain library will be increased; if the performance of the large language model after attaching the thought chain library is decreased, the length or complexity of the chains will be considered.

[0175] After obtaining a large language model with the thought chain library attached, this application will intelligently use the same performance test data to evaluate the performance of the large language model. By comparing and analyzing the obtained model performance data with the basic model performance data, it monitors the performance changes of the large language model before and after attaching the thought chain library. Based on the comparison and analysis results, it will make adaptive adjustments to the large language model, which is conducive to more effectively utilizing thought chain technology to improve the performance of the large language model in dialogue speech processing tasks.

[0176] It should be understood that the sequence number of each step in the above embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0177] It should be emphasized that, to further ensure the privacy and security of the above-mentioned answer data, the answer data can also be stored in a blockchain node.

[0178] The blockchain referred to in this application is a novel application model of computer technologies such as distributed data storage, peer-to-peer transmission, consensus mechanisms, and encryption algorithms. Essentially, a blockchain is a decentralized database, a chain of data blocks linked together using cryptographic methods. Each data block contains information about a batch of network transactions, used to verify the validity of the information (anti-counterfeiting) and generate the next block. A blockchain can include an underlying blockchain platform, a platform product service layer, and an application service layer.

[0179] The embodiments of this application can acquire and process relevant data based on artificial intelligence technology. Artificial intelligence (AI) is the theory, method, technology, and application system that uses 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.

[0180] 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.

[0181] 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 above methods. 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).

[0182] 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.

[0183] Further reference Figure 3 As a response to the above Figure 2 To implement the method shown, this application provides an embodiment of an artificial intelligence-based question processing device, which is similar to... Figure 2 Corresponding to the method embodiments shown, this device can be specifically applied to various electronic devices.

[0184] like Figure 3As shown, the AI-based question processing device 300 described in this embodiment includes: a conversion module 301, a preprocessing module 302, a recognition module 303, a determination module 304, a filtering module 305, a first processing module 306, and a second processing module 307. Wherein:

[0185] The conversion module 301 is used to perform text conversion processing on the received user-inputted voice question to obtain the first text;

[0186] Preprocessing module 302 is used to preprocess the first text to obtain the second text;

[0187] The recognition module 303 is used to perform intent recognition on the second text based on a preset large language model to obtain the target intent corresponding to the second text; wherein, the large language model is equipped with a thought chain library generated based on a preset construction method, and the thought chain library includes a certain number of thought chains.

[0188] The determination module 304 is used to determine the processing complexity corresponding to the second text;

[0189] The filtering module 305 is used to filter out the target thinking chain from all the thinking chains contained in the thinking chain library based on the target intent and the processing complexity.

[0190] The first processing module 306 is used to perform execution and reasoning processing according to the target thought chain through the large language model, and output answer data corresponding to the target intent;

[0191] The second processing module 307 is used to provide feedback to the user based on the answer data.

[0192] In this embodiment, the operations performed by the above modules or units correspond one-to-one with the steps of the AI-based question processing method in the aforementioned implementation method, and will not be repeated here.

[0193] In some optional implementations of this embodiment, the filtering module 305 includes:

[0194] The first acquisition submodule is used to acquire intent identification information corresponding to each of the aforementioned thought chains;

[0195] The second acquisition submodule is used to acquire a first thought chain that matches the target intent from all the thought chains based on the intent identification information;

[0196] The first judgment submodule is used to determine whether the number of the first thought chain includes multiple chains;

[0197] The filtering submodule is used to, if so, filter out a second thinking chain from the first thinking chain that matches the processing complexity based on preset filtering rules;

[0198] The first determining submodule is used to use the second thinking chain as the target thinking chain.

[0199] In some optional implementations of this embodiment, the filtering submodule includes:

[0200] A judgment unit is used to determine whether the processing complexity is high complexity;

[0201] The first filtering unit is used to filter out the third thinking chain with the longest chain length from all the first thinking chains if the processing complexity is high complexity, and use the third thinking chain as the second thinking chain.

[0202] The second judgment unit is used to determine whether the processing complexity is low if the processing complexity is not high.

[0203] The second filtering unit is used to filter out the fourth thinking chain with the smallest chain length from all the first thinking chains if the processing complexity is low, and to use the fourth thinking chain as the second thinking chain.

[0204] The acquisition unit is used to determine that the processing complexity is medium complexity if the processing complexity is not low complexity, and to acquire the usage frequency value of the fifth thinking chain; wherein, the fifth thinking chain is all the other thinking chains in the first thinking chain except for the third thinking chain and the fourth thinking chain.

[0205] The third filtering unit is used to filter out the sixth thinking chain with the highest usage frequency from the fifth thinking chain, and to use the sixth thinking chain as the second thinking chain.

[0206] In some optional implementations of this embodiment, the determining module 304 includes:

[0207] The sentence segmentation module is used to segment the second text into sentences to obtain corresponding sentences, and to count the first number of the sentences.

[0208] The word segmentation submodule is used to segment the second text to obtain the corresponding words and count the second number of the words;

[0209] The calculation submodule is used to calculate the complexity score of the second text based on the first quantity and the second quantity, and based on a preset calculation strategy.

[0210] The query submodule is used to perform query processing on a preset mapping data table based on the complexity score, so as to retrieve the processing complexity corresponding to the complexity score from the mapping data table.

[0211] In some optional implementations of this embodiment, the computation submodule includes:

[0212] The first determining unit is used to determine the preset weight calculation algorithm;

[0213] The second determining unit is used to determine the first weight corresponding to the number of sentences based on the weight calculation algorithm;

[0214] The third determining unit is used to determine the second weight corresponding to the number of words based on the weight calculation algorithm;

[0215] The calculation unit is used to calculate the first quantity and the second quantity based on the first weight and the second weight by calling a preset score calculation formula to obtain the corresponding calculation result;

[0216] The fourth determining unit is used to use the calculation result as the complex score of the second text.

[0217] In some optional implementations of this embodiment, the AI-based question processing device further includes:

[0218] The judgment module is used to determine whether a mind chain library construction request triggered by a specified user has been received;

[0219] The display module is used to display the preset build page if the condition is met.

[0220] A receiving module is used to receive the thought chain data input by the specified user on the construction page;

[0221] The construction module is used to build a corresponding mind chain library based on the mind chain data;

[0222] The storage module is used for storing and processing the thought chain library.

[0223] In some optional implementations of this embodiment, the AI-based question processing device further includes:

[0224] The first acquisition module is used to acquire pre-collected performance test data;

[0225] This module is used to evaluate the performance of the large language model based on the performance test data, and obtain the corresponding model performance data.

[0226] The second acquisition module is used to acquire the basic model performance data corresponding to the large language model;

[0227] The analysis module is used to compare and analyze the model performance data with the basic model performance data to obtain the corresponding comparison and analysis results;

[0228] The adjustment module is used to adjust the large language model based on the comparison analysis results.

[0229] To address the aforementioned technical problems, embodiments of this application also provide a computer device. Please refer to [link / reference needed]. Figure 4 , Figure 4 This is a basic structural block diagram of the computer device in this embodiment.

[0230] The computer device 4 includes a memory 41, a processor 42, and a network interface 43 that are interconnected via a system bus. It should be noted that only the computer device 4 with components 41-43 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. Its hardware includes, but is not limited to, microprocessors, application-specific integrated circuits (ASICs), programmable gate arrays (FPGAs), digital digital processors (DSPs), embedded devices, etc.

[0231] 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.

[0232] The memory 41 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 41 may be an internal storage unit of the computer device 4, such as the hard disk or memory of the computer device 4. In other embodiments, the memory 41 may also be an external storage device of the computer device 4, such as a plug-in hard disk, smart memory card (SMC), secure digital card (SD) card, flash card, etc. of the computer device 4. Of course, the memory 41 may also include both the internal storage unit and the external storage device of the computer device 4. In this embodiment, the memory 41 is typically used to store the operating system and various application software installed on the computer device 4, such as computer-readable instructions for query processing methods based on artificial intelligence. In addition, the memory 41 can also be used to temporarily store various types of data that have been output or will be output.

[0233] In some embodiments, the processor 42 may be a central processing unit (CPU), a controller, a microcontroller, a microprocessor, or other data processing chip. The processor 42 is typically used to control the overall operation of the computer device 4. In this embodiment, the processor 42 is used to execute computer-readable instructions stored in the memory 41 or to process data, for example, to execute computer-readable instructions of the AI-based query processing method.

[0234] The network interface 43 may include a wireless network interface or a wired network interface, which is typically used to establish communication connections between the computer device 4 and other electronic devices.

[0235] This application also provides another embodiment, namely, providing a computer-readable storage medium storing computer-readable instructions that can be executed by at least one processor to cause the at least one processor to perform the steps of the artificial intelligence-based query processing method described above.

[0236] 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.

[0237] 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. An artificial intelligence-based question processing method, characterized by, Includes the following steps: The received user-input voice question is converted into text to obtain the first text; The first text is preprocessed to obtain the second text; The intent of the second text is identified based on a pre-set large language model to obtain the target intent corresponding to the second text; wherein, the large language model is equipped with a thought chain library generated based on a pre-set construction method, and the thought chain library includes a certain number of thought chains. Determine the processing complexity corresponding to the second text; Based on the target intent and the processing complexity, the target thought chain is selected from all thought chains contained in the thought chain library; The large language model is used to perform execution and reasoning processing according to the target thought chain, and output answer data corresponding to the target intent. Feedback is processed on the user based on the answer data; The step of selecting the target thought chain from all thought chains contained in the thought chain library based on the target intent and the processing complexity specifically includes: Obtain the intent identification information corresponding to each of the aforementioned thought chains; Based on the intent identification information, a first thought chain that matches the target intent is obtained from all the thought chains; Determine whether the number of the first thought chain includes multiple chains; If so, a second thought chain matching the processing complexity is selected from the first thought chain based on preset filtering rules; The second thinking chain is taken as the target thinking chain; The step of selecting a second thought chain from the first thought chain that matches the processing complexity based on preset filtering rules specifically includes: Determine whether the processing complexity is high. If the processing complexity is high, then select the third thinking chain with the longest chain length from all the first thinking chains, and use the third thinking chain as the second thinking chain. If the processing complexity is not high, then determine whether the processing complexity is low. If the processing complexity is low, then select the fourth thinking chain with the smallest chain length from all the first thinking chains, and use the fourth thinking chain as the second thinking chain. If the processing complexity is not low, then the processing complexity is determined to be medium, and the usage frequency value of the fifth thinking chain is obtained; wherein, the fifth thinking chain is all the other thinking chains in the first thinking chain except for the third thinking chain and the fourth thinking chain. The sixth thinking chain with the highest usage frequency is selected from the fifth thinking chain, and the sixth thinking chain is used as the second thinking chain. 2.The AI-based question processing method of claim 1, wherein, The step of determining the processing complexity corresponding to the second text specifically includes: The second text is segmented into sentences to obtain corresponding sentences, and the first number of the sentences is counted. The second text is segmented to obtain the corresponding words, and the second number of the words is counted. Based on the first quantity and the second quantity, the complexity score of the second text is calculated based on a preset calculation strategy; Based on the complexity score, a preset mapping data table is queried to retrieve the processing complexity corresponding to the complexity score from the mapping data table. 3.The AI-based question processing method of claim 2, wherein, The step of calculating the complexity score of the second text based on the first quantity and the second quantity, and based on a preset calculation strategy, specifically includes: Determine the preset weight calculation algorithm; The first weight corresponding to the number of sentences is determined based on the weighting algorithm described above; The second weight corresponding to the number of words is determined based on the weighting algorithm described above; Based on the first weight and the second weight, a preset score calculation formula is called to calculate the first quantity and the second quantity to obtain the corresponding calculation result; The calculation result is used as the complex score of the second text. 4.The AI-based question processing method of claim 1, wherein, Before the step of performing intent recognition on the second text based on a preset large language model to obtain the target intent corresponding to the second text, the method further includes: Determine whether a mind chain library construction request triggered by a specified user has been received; If so, display the preset build page; Receive the thought chain data input by the specified user on the construction page; Construct a corresponding mind chain library based on the aforementioned mind chain data; The thought chain library is stored and processed. 5.The AI-based question processing method of claim 1, wherein, After the step of processing feedback to the user based on the answer data, the method further includes: Acquire pre-collected performance test data; The performance of the large language model is evaluated based on the performance test data to obtain the corresponding model performance data. Obtain the basic model performance data corresponding to the large language model; The performance data of the model is compared and analyzed with the performance data of the basic model to obtain the corresponding comparison and analysis results. The large language model is adjusted based on the comparison analysis results.

6. An artificial intelligence-based question processing device, characterized by comprising: include: The conversion module is used to convert the received user-input voice question into text to obtain the first text. The preprocessing module is used to preprocess the first text to obtain the second text; The recognition module is used to perform intent recognition on the second text based on a preset large language model to obtain the target intent corresponding to the second text; wherein, the large language model is equipped with a thought chain library generated based on a preset construction method, and the thought chain library includes a certain number of thought chains. A determination module is used to determine the processing complexity corresponding to the second text; The filtering module is used to filter out the target thinking chain from all thinking chains contained in the thinking chain library based on the target intent and the processing complexity. The first processing module is used to perform execution and reasoning processing according to the target thought chain through the large language model, and output answer data corresponding to the target intent; The second processing module is used to provide feedback to the user based on the answer data; The filtering module includes: The first acquisition submodule is used to acquire intent identifier information corresponding to each of the aforementioned thought chains; The second acquisition submodule is used to acquire a first thought chain that matches the target intent from all the thought chains based on the intent identification information; The first judgment submodule is used to determine whether the number of the first thought chain includes multiple chains; The filtering submodule is used to, if so, filter out a second thinking chain from the first thinking chain that matches the processing complexity based on preset filtering rules; The first determining submodule is used to use the second thinking chain as the target thinking chain; The filtering submodule includes: A judgment unit is used to determine whether the processing complexity is high complexity; The first filtering unit is used to filter out the third thinking chain with the longest chain length from all the first thinking chains if the processing complexity is high complexity, and use the third thinking chain as the second thinking chain. The second judgment unit is used to determine whether the processing complexity is low if the processing complexity is not high. The second filtering unit is used to filter out the fourth thinking chain with the smallest chain length from all the first thinking chains if the processing complexity is low, and to use the fourth thinking chain as the second thinking chain. The acquisition unit is used to determine that the processing complexity is medium complexity if the processing complexity is not low complexity, and to acquire the usage frequency value of the fifth thinking chain; wherein, the fifth thinking chain is all the other thinking chains in the first thinking chain except for the third thinking chain and the fourth thinking chain. The third filtering unit is used to filter out the sixth thinking chain with the highest usage frequency from the fifth thinking chain, and to use the sixth thinking chain as the second thinking chain.

7. A computer device, characterized by The system includes a memory and a processor, wherein the memory stores computer-readable instructions, and the processor, when executing the computer-readable instructions, implements the steps of the AI-based question processing 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 AI-based question processing method as described in any one of claims 1 to 5.

Images