Identity authentication method

Abstract

The invention relates to an identity authentication method, and the method comprises the steps: receiving an identity authentication request which comprises an identity label of a target user; displaying an identity authentication question, the identity authentication question being a question corresponding to the identity authentication of the target user generated by the questioning agent based on the user information in the combined dynamic knowledge base based on the identity authentication request; receiving a user answer fed back by the target user; the identity authentication result of the target user is displayed, and the identity authentication result is generated after the evaluation agent evaluates the accuracy of the user answer. According to the method and the device, the problem of identity verification security caused when a centralized information source is damaged due to centralized original information storage in related technologies is solved.

Description

Technical Field

[0001] The embodiments of the present application relate to the technical field of identity security authentication, and in particular to an identity authentication method. Background Art

[0002] Identity authentication is an important part of information security. Its main purpose is to protect the authenticity of user identity and thus protect the system from unauthorized access. Common authentication methods include static passwords, two-factor identity authentication, biometric recognition, etc.

[0003] However, the above-mentioned identity authentication schemes all require the collection and storage of original information, and have the centralization of information sources. During identity authentication, they rely on obtaining original information from centralized information sources to perform identity authentication through comparison. This centralized authentication method is not only vulnerable to attacks or data leaks, but once the centralized information source is destroyed, all authentication mechanisms that rely on this information source will face verification security risks. Summary of the invention

[0004] The embodiment of the present application provides an identity authentication method, which can reduce the dependence on the central information source and improve the security of identity authentication. In addition, since the user information in the joint dynamic knowledge base is updated in real time, the timeliness and pertinence of the identity authentication problem can be guaranteed, and the reliability of the identity authentication can be further improved. The technical solution is as follows.

[0005] In one aspect, an identity authentication method is provided, the method comprising: Receiving an identity authentication request, wherein the identity authentication request includes an identity identifier of a target user; Displaying an identity authentication question, wherein the identity authentication question is a question corresponding to the identity authentication of the target user generated by the questioning agent based on the identity authentication request and based on the user information in the joint dynamic knowledge base; Receiving a user answer fed back by the target user; The identity authentication result of the target user is displayed, wherein the identity authentication result is generated after the evaluation agent evaluates the accuracy of the user's answer.

[0006] In another aspect, an identity authentication method is provided, the method comprising: Receiving an identity authentication request sent by a target user device; the identity authentication request includes an identity identifier of the target user; Based on the identity authentication request, generating an identity authentication question corresponding to the target user through a questioning agent, and sending the identity authentication question to the target user device; the questioning agent is used to generate the identity authentication question based on the user information in the joint dynamic knowledge base; When receiving the user answer fed back by the target user device, the user answer is evaluated for accuracy by an evaluation agent to generate an identity authentication result of the target user.

[0007] On the other hand, a computer device is provided, the computer device comprising a processor and a memory, the memory storing at least one computer program, the at least one computer program being loaded and executed by the processor to implement the above-mentioned identity authentication method.

[0008] On the other hand, a computer-readable storage medium is provided, wherein at least one computer program is stored in the computer-readable storage medium, and the computer program is loaded and executed by a processor to implement the above-mentioned identity authentication method.

[0009] On the other hand, a computer program product is provided, which includes a computer program stored on a non-transitory computer-readable storage medium, and the computer program includes program instructions, which, when executed by a computer, cause the computer to execute to implement the identity authentication method provided in the various optional implementations described above.

[0010] The identity authentication method provided by the embodiment of the present application, on the target user device side, after the target user terminal receives the identity authentication request containing the identity identifier of the target user, displays the identity authentication question corresponding to the target user generated and fed back by the questioning agent based on the user information in the joint dynamic knowledge base and the identity authentication request; after receiving the user answer fed back by the target user, displays the identity authentication result generated by the evaluation agent after the accuracy evaluation of the user answer. Through the above method, question-and-answer identity authentication can be performed through the questioning agent, the evaluation agent and the joint dynamic knowledge base, thereby realizing personalized identity authentication corresponding to each user, thereby reducing the dependence on the central information source and improving the security of identity authentication; in addition, because the user information in the joint dynamic knowledge base is updated in real time, the timeliness and pertinence of the identity authentication question can be guaranteed, thereby further improving the reliability of identity authentication.

[0011] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the present application and, together with the description, serve to explain the principles of the present application.

[0013] Figure 1 A flowchart of an identity authentication method provided by an exemplary embodiment of the present application is shown; Figure 2 A flowchart of an identity authentication method provided by another exemplary embodiment of the present application is shown; Figure 3 A schematic diagram of an identity authentication system provided by an exemplary embodiment of the present application is shown; Figure 4 A flowchart of an identity authentication method provided by another exemplary embodiment of the present application is shown; Figure 5 A schematic diagram showing the operation of an identity authentication system provided by an exemplary embodiment of the present application is shown; Figure 6 An interactive schematic diagram of an identity authentication method provided by an exemplary embodiment of the present application is shown; Figure 7 A structural block diagram of a computer device shown in an exemplary embodiment of the present application is shown; Figure 8 A structural block diagram of another computer device shown in an exemplary embodiment of the present application is shown. DETAILED DESCRIPTION

[0014] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, unless otherwise indicated, the same numbers in different drawings represent the same or similar elements. The implementations described in the following exemplary embodiments do not represent all implementations consistent with the present application. Instead, they are merely examples of methods consistent with some aspects of the present application as detailed in the appended claims.

[0015] The present application provides an identity authentication method, which performs identity authentication based on a question-and-answer identity authentication mechanism that integrates a multi-agent system (MAS) and an integrated dynamic knowledge base (IDKB), reduces dependence on a central information source, and improves the security of identity authentication.

[0016] The identity authentication method provided in this application involves the interaction between the user equipment and the identity authentication system. On the target user equipment side, Figure 1 A flowchart of an identity authentication method provided by an exemplary embodiment of the present application is shown. The method can be executed by a target user device, such as a display device. Figure 1 As shown, the method includes the following steps.

[0017] Step 110: receiving an identity authentication request, wherein the identity authentication request includes an identity identifier of a target user.

[0018] The identity authentication request may be initiated by the target user and is used to request identity authentication information; illustratively, the identity authentication request may be sent by the user when identity authentication is triggered based on the target user device, for example, the user may trigger identity authentication when conducting a transaction or information query. After receiving the identity authentication request, the target user device starts the identity authentication process to authenticate the user's identity. The identity identifier is a unique identifier of the target user in the system. Schematically, the identity identifier may be a user name, ID number, mobile phone number, etc., and this application does not limit this.

[0019] Step 120, displaying an identity authentication question, wherein the identity authentication question is based on the identity authentication request and is generated by the questioning agent based on the user information in the joint dynamic knowledge base and corresponds to the identity authentication question of the target user.

[0020] Questioning Agent (QA) is an intelligent device deployed in the identity authentication system that has the ability to intelligently generate questions. It can provide personalized identity authentication questions for the target user based on the identity authentication request received from the target user terminal and the user information in the joint dynamic knowledge base.

[0021] The joint dynamic knowledge base is a database collection that stores various types of user-related information through multiple knowledge bases. Schematically, the user information in the joint dynamic knowledge base may include but is not limited to the user's basic information, user behavior habit information, historical operation information and other dynamically changing information; the user information in the joint dynamic database can be continuously updated as the user interacts with the system to provide more comprehensive and accurate information to support the generation of identity authentication questions.

[0022] After receiving the identity authentication request, the target user device sends the identity authentication request to the questioning agent. Accordingly, the questioning agent receives the identity authentication request and retrieves user information related to the target user from the linked dynamic knowledge base based on the identity identifier carried in the identity authentication request, and generates personalized identity authentication questions corresponding to the target user based on this information; thereafter, the questioning agent sends the identity authentication question to the target user device, so that the target user device displays the identity authentication question to the target user through a display interface so that the user can respond.

[0023] Among them, the identity authentication questions generated by the questioning agent in each round can be one or more, and the number of identity authentication questions generated by the questioning agent in each round can be set based on actual needs, and this application does not impose any restrictions on this.

[0024] Step 130: receiving a user response from the target user.

[0025] After the target user obtains the identity authentication question through the target user device, the target user can give an answer to the identity authentication question. In some possible implementations, the display interface of the target user device can display the identity authentication question and an answer feedback component corresponding to the identity authentication question, such as a text box, or an answer area marked by a specific symbol. The target user can provide feedback on the user's answer through text input based on the answer feedback component, or can also provide feedback on the user's answer through voice input. This application does not limit this.

[0026] Step 140, displaying the identity authentication result of the target user, wherein the identity authentication result is generated after the evaluation agent evaluates the accuracy of the user's answer.

[0027] After receiving the user answer fed back by the target user, the target user device sends the user answer to the evaluation agent, so that the evaluation agent performs accuracy evaluation based on the received user answer, and further determines the identity authentication result of the target user.

[0028] Among them, the Evaluation Agent (EA) is deployed in the identity authentication system. It is an intelligent device that can analyze and judge user answers. It can compare the received user answers with the correct images in the joint dynamic knowledge base, and use pre-trained or set algorithms and logic to evaluate the accuracy of user answers, thereby generating the final identity authentication results.

[0029] After determining the identity authentication result of the target user, the evaluation agent feeds back the identity authentication result to the target user device, so that the target user device displays the identity authentication result to the target user through a display interface; wherein the identity authentication result can be a successful identity authentication or a failed identity authentication.

[0030] In summary, the identity authentication method provided by the embodiment of the present application, on the target user device side, after the target user terminal receives the identity authentication request containing the identity identifier of the target user, displays the identity authentication question corresponding to the target user generated and fed back by the questioning agent based on the user information in the joint dynamic knowledge base and the identity authentication request; after receiving the user answer fed back by the target user, displays the identity authentication result generated by the evaluation agent after the accuracy evaluation of the user answer. Through the above method, question-and-answer identity authentication can be performed through the questioning agent, the evaluation agent and the joint dynamic knowledge base, thereby realizing personalized identity authentication corresponding to each user, thereby reducing the dependence on the central information source and improving the security of identity authentication; in addition, because the user information in the joint dynamic knowledge base is updated in real time, the timeliness and pertinence of the identity authentication question can be guaranteed, thereby further improving the reliability of identity authentication.

[0031] Corresponding to the target terminal side, on the identity authentication system side, Figure 2 A flowchart of an identity authentication method provided by another exemplary embodiment of the present application is shown. The method can be executed by an identity authentication system, such as Figure 2 As shown, the method includes the following steps.

[0032] Step 210: Receive an identity authentication request sent by a target user device; the identity authentication request includes an identity identifier of the target user.

[0033] In an embodiment of the present application, the identity authentication system includes a multi-agent system (MAS) and a joint dynamic knowledge base, wherein the multi-agent system includes a questioning agent and an evaluation agent. When the target user triggers the identity authentication operation on the target user device, the target user device sends an identity authentication request carrying the target user's identity identifier to the identity authentication system; optionally, the identity authentication system can receive the identity authentication request through the questioning agent, and extract the identity identifier in the authentication request to prepare for subsequent operations.

[0034] Step 220, based on the identity authentication request, generates an identity authentication question corresponding to the target user through a questioning agent, and sends the identity authentication question to the target user device; the questioning agent is used to generate the identity authentication question based on the user information in the joint dynamic knowledge base.

[0035] After receiving the authentication request containing the identity identifier, the questioning agent uses the identity identifier as an index to retrieve various information related to the target user in the joint dynamic knowledge base. According to the preset question generation rules and algorithms, combined with the retrieved user information of the target user, a personalized authentication question suitable for the target user is generated. For example, if the joint dynamic knowledge base records the location information of the target user's last login, the questioning agent may generate an authentication question such as "Where was your last login location?" After the authentication question is generated, it is sent back to the target user's device so that the target user can view and answer it.

[0036] In some possible implementations, the identity authentication question generated by the agent may be a short answer question, or, in other possible implementations, the identity authentication question generated by the agent may be a multiple choice question, for example, the identity authentication question is "Which of the following places is your last login location" and provides multiple options for the user to choose from; this application does not limit the question type of the identity authentication question.

[0037] Step 230, upon receiving the user's answer fed back by the target user's device, the user's answer is evaluated for accuracy by an evaluation agent to generate an identity authentication result of the target user.

[0038] After receiving the identity authentication question, the target user device presents it to the target user through a display interface. After the target user provides feedback on the user's answer based on his or her actual situation, the target user device feeds back the user's answer to the evaluation agent in the identity authentication system. Correspondingly, the evaluation agent receives the user's answer. The evaluation agent can use the target user's identity identifier as an index to obtain the correct answer corresponding to the question in the joint dynamic knowledge base, and use a preset evaluation algorithm to compare and analyze the user's answer and the correct answer to determine the accuracy of the user's answer. For example, through technical means such as string matching and semantic analysis, it is determined whether the user's answer is consistent with the correct answer or is within an acceptable error range; the identity authentication result of the target user is generated based on the evaluation result.

[0039] After generating the identity authentication result of the target user, the evaluation agent sends the identity authentication result back to the target user device. Correspondingly, after receiving the identity authentication result, the target user device displays the identity authentication result to the target user on the device interface, such as popping up a prompt box to display "Identity authentication successful" or "Identity authentication failed" and other information to inform the target user of the final result of this identity authentication.

[0040] Furthermore, the identity authentication system can perform corresponding subsequent operations based on the identity authentication result of the target user. For example, if the authentication is successful, the target user is allowed to access specific resources or functions. If the authentication fails, relevant information is recorded, and the number of further attempts can be limited.

[0041] Before the user initiates an authentication request, the authentication system needs to prepare the joint dynamic knowledge base and fine-tune the questioning agent and the evaluation agent. During the fine-tuning training process, the questioning agent can learn from a large number of real-life corpora how to generate highly customized, non-repetitive, and closely related to the user's real-time status authentication questions based on the data in the joint dynamic knowledge base, and dynamically adjust the questioning strategy in different scenarios. The evaluation agent learns how to semantically understand user answers and match them based on keywords through training with a semantic analysis corpus.

[0042] Through the above-mentioned fine-tuning training, questioning agents and evaluation agents can learn to acquire the capabilities of planning, reflection, memory, and tool use in four dimensions. In terms of planning, the planning of questioning agents and evaluation agents is based on their thinking chain capabilities. When generating questions, questioning agents plan the thinking logic of the entire question, including the source of question knowledge, instruction associations, and tool use, etc. Indicatively, the thinking chain for raising questions formulated by questioning agents based on planning capabilities can be: "The questioning agent will split the big question raised to the user into multiple sub-questions: 1. Have you asked the user questions before? How effective are the answers? 2. Which knowledge base should the current question be collected from? 3. How to ask reasonable questions based on the collected knowledge? Should questions be asked from a single knowledge source or after knowledge association? 4. How to use a reasonable tone to ask the user this question?" The questioning agent connects these sub-questions into a complete thinking chain and implements this thinking chain in the process. The system continuously uses its own reflection, memory and tool capabilities, and finally proposes highly customized, non-repetitive and closely related to the user's real-time status authentication questions to the user through the cooperation of three types of tools: knowledge collection, knowledge association and question-posing. It should be noted that the thought chain of the above-mentioned question-posing can be different based on the adjustment of the fine-tuning process and the difference of related configurations, and this application does not limit this. After receiving the user's answer, the evaluation agent also analyzes the user's answer based on the thought chain and plans the thinking logic of the entire evaluation, including question classification, semantic analysis and tool use, etc. In schematic form, the thought chain of the evaluation agent to evaluate the user's answer based on the planning ability can be: The evaluation agent divides the big question of evaluating the user's answer into multiple sub-questions: 1. Which type of knowledge base question does this question belong to? 2. Should the questions from the knowledge base use character verification or semantic similarity verification? 3. Is the semantic similarity verification greater than the threshold value for judging as correct? 4.How to use a reasonable tone to output the evaluation results to the user? "The evaluation agent connects these sub-questions into a complete chain of thinking, and constantly uses its own reflection, memory, and tool capabilities in the implementation of this chain of thinking. Through the mutual cooperation of three types of tools: question classification, semantic similarity verification, and character verification, it finally realizes the correctness of the answer by combining string matching and semantic level understanding and analysis, and then completes identity authentication. Similarly, the above-mentioned thinking chain for evaluating user answers can be different based on the adjustment of the fine-tuning process and the difference in related configurations, and this application does not limit this; in terms of reflection, the questioning agent and the evaluation agent constantly reflect during the authentication process. The questioning agent can optimize subsequent questions based on the effects of previously generated questions to avoid generating repetitive questions and ensure the uniqueness of each authentication; the evaluation agent combines the context and the answer content to analyze whether it is necessary to further ask verification questions or pass the authentication. The reflection function dynamically adjusts the entire authentication process to improve the intelligence of the system; in terms of memory, the questioning agent and the evaluation agent have memory function. Yes, it can store and retrieve relevant information during the authentication process. When asking questions, the questioning agent will remember the previous questions and the user's answers to ensure that subsequent questions will not be repeated or irrelevant; the memory function of the evaluation agent is reflected in the tracking and analysis of user answers. It will not only remember the user's answers in the current session, but also retrieve past authentication records as a reference for analyzing the user's current answers, which helps to reduce the system's reliance on accidental factors and improve the accuracy of authentication; in terms of tool use, the questioning agent and the evaluation agent have the ability to use tools. The tool library of the questioning agent can include three types of tools: knowledge collection, knowledge association, and question asking. These tools can be flexibly called to generate questions. The tool library of the evaluation agent includes three types of tools: question classification, semantic similarity verification, and character verification. These tools can be flexibly called to evaluate the rationality and correctness of the user's answers to ensure the rigor of authentication. The capabilities of the questioning agent and the evaluation agent in the four dimensions of planning, reflection, memory, and tool use are explained through subsequent examples. .

[0043] In summary, the identity authentication method provided by the embodiment of the present application, on the identity authentication system side, after receiving the identity authentication request containing the identity identifier of the target user sent by the target user device, the identity authentication system generates an identity authentication question corresponding to the target user based on the identity authentication request through a questioning agent based on the user information in the joint dynamic knowledge base and the identity authentication request and sends it to the target user device; when receiving the user answer fed back by the target user device, the identity authentication result is generated by the evaluation agent performing an accuracy evaluation on the user answer. Through the above method, question-and-answer identity authentication can be performed through the questioning agent, the evaluation agent and the joint dynamic knowledge base, thereby realizing personalized identity authentication corresponding to each user, thereby reducing dependence on central information sources and improving the security of identity authentication; in addition, since the user information in the joint dynamic knowledge base is updated in real time, the timeliness and pertinence of the identity authentication question can be guaranteed, thereby further improving the reliability of identity authentication.

[0044] Figure 3 A schematic diagram of an identity authentication system provided by an exemplary embodiment of the present application is shown. Figure 3 As shown, the identity authentication system includes a user interface layer 310 , a business logic layer 320 , a data storage layer 330 and a management background layer 340 .

[0045] Among them, the user interface layer may include an authentication request component and an authentication result component. The authentication request component is used to receive the identity authentication request sent by the user device and pass the identity authentication request to the questioning agent of the business logic layer; the authentication result component is used to receive the identity authentication result fed back by the evaluation agent and send the identity authentication result to the user device.

[0046] In some possible implementations, the user equipment may be a part of the identity authentication system, so as to serve as a user interface layer of the identity authentication system to receive identity authentication requests and provide feedback on identity authentication results.

[0047] In some other possible implementations, the user device is independent of the identity authentication system. Schematically, the user device can be set at the user's end (such as home, public places, and other places where the user actually operates), and the identity authentication system is set in a cloud server or data center or other place with powerful computing and storage capabilities.

[0048] The business logic layer corresponds to a multi-agent system, including questioning agents and evaluation agents, which are responsible for generating identity authentication questions and evaluating user answers respectively.

[0049] Among them, the question-asking agent may include a first planning component, a first information retrieval component, a question construction component, a first reflection component and a first memory component; the first planning component is used to plan the generation logic of the question, including the source, type and difficulty of the question; the first information retrieval component is used to retrieve relevant information from the joint dynamic knowledge base to provide data support for question generation; the question construction component is used to generate customized, non-repetitive and user-related identity authentication questions based on the information provided by the indication retrieval component; the first reflection component is used to optimize subsequent questions based on the user's answer effect to the previous one or more questions to avoid poor answer accuracy due to the type or field of the question; the first memory component is used to store the constructed questions to avoid generating repetitive questions and ensure the uniqueness of each authentication process. It can also be used to store historical question and answer information to provide a reference for user personalized preferences for question construction.

[0050] The evaluation agent may include a second planning component, a second information retrieval component, an accuracy judgment component, a second reflection component, and a second memory component; the second planning component is used to plan the evaluation logic of the user's answer, including the selection of the evaluation method and the setting of the evaluation standard; the second information retrieval component is used to retrieve the reference answer of the current question from the joint dynamic knowledge base to provide data support for the evaluation of the user's answer; the accuracy judgment component is used to judge the accuracy of the user's answer based on the determined evaluation method and reference information; the second reflection component is used to judge whether it is necessary to further ask verification questions or pass the authentication. The reflection component can realize the dynamic adjustment of the authentication process for different users and improve the intelligence of the system; the second memory component is used to store and retrieve the user's historical question and answer information to provide a reference for the user's personalized preferences when evaluating the current user's answer. As a reference for evaluating the current user's answer, the accuracy of identity authentication is improved.

[0051] The data storage layer corresponds to a joint dynamic knowledge base, which can include a basic information knowledge base (BIKB), a behavioral knowledge base (BKB), and a supplementary knowledge base (SKB) to provide data support, ensure the personalization and real-time nature of questions, and the accuracy and real-time nature of user answer evaluation. Among them, the static basic information stored in the basic knowledge base can be the user's relatively fixed basic information, such as the registered email address, telephone number, birthday information, etc. The basic knowledge base contains an information update component, and the user can update the static basic information through the information update component provided by the basic knowledge base; the behavioral knowledge base is used to record and analyze the user's activity trajectory and behavioral characteristics and other dynamic behavioral information, including but not limited to the most recently participated projects, recently visited web pages, meeting topics, frequently used contacts, location information, etc. The behavioral knowledge base may include a data analysis component for regularly analyzing the user's behavior model, extracting behavioral characteristics and activity trajectories, so as to enrich and update the background and content of questions; the supplementary knowledge base can accept personalized information actively uploaded by the user, such as favorite books, movie introductions or life experiences, etc. The supplementary knowledge base may include a data management component, and the user can manage and edit personalized information through the data management component.

[0052] The management backend layer provides a management platform for users with management privileges (i.e., administrators), and can receive system management from administrators, including knowledge base management, intelligent agent training, and real-time monitoring, to ensure the correct operation of the system and the accuracy of data; among them, knowledge base management can include configuration management and data auditing. Configuration management means that administrators can configure and manage each knowledge base in the joint dynamic knowledge base to ensure the accuracy and timeliness of information in the knowledge base; data auditing means that administrators can audit the data uploaded by users to ensure the quality and compliance of the data. Intelligent agent training means that administrators can fine-tune the training of each intelligent agent to improve its performance in different scenarios; real-time monitoring means that administrators can try to monitor the operating status of each intelligent agent to ensure its efficient operation and timely discover and solve problems.

[0053] Based on the above identity authentication system, Figure 4 A flowchart of an identity authentication method provided by another exemplary embodiment of the present application is shown. The method can be executed interactively by a target user device and an identity authentication system. Figure 4 As shown, the method includes the following steps.

[0054] Step 401: The target user equipment receives an identity authentication request, where the identity authentication request includes an identity authentication identifier of the target user.

[0055] The target user device receives the identity authentication request initiated by the user.

[0056] Step 402: The target user equipment sends an identity authentication request to the identity authentication system. Correspondingly, the identity authentication system receives the identity authentication request sent by the target user equipment.

[0057] Further, the target user equipment sends the identity authentication request to the questioning agent in the identity authentication system, and the corresponding questioning agent receives the identity authentication request.

[0058] Step 403, the identity authentication system generates identity authentication questions corresponding to the target user through a questioning agent based on the identity authentication request; the questioning agent is used to generate the identity authentication questions based on the user information in the joint dynamic knowledge base.

[0059] The joint dynamic knowledge base is a knowledge base collection composed of multiple knowledge bases, each of which can store user information of users in different dimensions. In some possible implementations, the joint dynamic knowledge base includes at least one of the following: a basic knowledge base, a behavioral knowledge base, and a supplementary knowledge base; the basic knowledge base stores the user's static basic information, the behavioral knowledge base stores the user's dynamic behavioral information, and the supplementary knowledge base stores the personalized user information uploaded by the user.

[0060] In some possible implementations, the process of the identity authentication system generating identity authentication questions corresponding to the target user by asking the intelligent agent can be implemented as follows: Based on the identity identifier of the target user contained in the identity authentication request, the user information of the target user is obtained by searching the information in the joint dynamic knowledge base through questioning the intelligent agent; An identity authentication question corresponding to the target user is constructed based on a question construction element, where the question construction element includes user information of the target user and a first preset prompt word.

[0061] When the questioning agent performs information retrieval in a dynamic knowledge base based on the identity identification of the target user, in some possible implementations, the questioning agent can perform information retrieval in each knowledge base in turn according to a pre-set information retrieval order, and extract user information corresponding to the target user. Schematically, if the information retrieval order is basic knowledge base, behavioral knowledge base to supplementary knowledge base, and the number of identity authentication questions generated by a single authentication is 3, the questioning agent can extract one or more user information from the basic knowledge base, behavioral knowledge base to supplementary knowledge base in turn; or, in other possible implementations, the questioning agent can randomly determine the database for information retrieval to extract user information of the target user from it.

[0062] It should be noted that the questioning agent can obtain the user information set of the target user when retrieving information, and then extract user information from the user information set based on the number of identity authentication questions to construct corresponding identity authentication questions; or, the questioning agent can obtain the corresponding number of user information from the joint dynamic knowledge base according to the number of identity authentication questions when performing information retrieval, so as to construct corresponding identity authentication questions.

[0063] When constructing an identity authentication question based on user information and a first preset prompt word, the question-asking intelligent agent can generate question-asking prompt information based on the acquired user information and the first preset prompt word, and input the question-asking prompt information into the large language model for processing to obtain the identity authentication question; illustratively, the question-asking prompt information can be "Please construct an identity authentication question based on the following user information"; based on different first preset prompt words, different types of identity authentication questions can be generated, and the first preset prompt word can be a keyword or phrase pre-set by relevant personnel to guide the question-asking intelligent agent to construct the identity authentication question.

[0064] In some possible implementations, the questioning agent includes a first short-term memory component; the first short-term memory component records the identity authentication questions that have been constructed during the current authentication process.

[0065] In this case, when the questioning agent performs information retrieval, it can refer to the identity authentication questions that have been constructed in the current authentication process recorded in the first short-term memory component for information retrieval; on the one hand, the question-answering context of the current authentication process can be determined based on the constructed identity authentication questions to retrieve user information that meets the current question-answering context; on the other hand, user information that is different from the constructed identity authentication questions can be screened to ensure the non-repetitiveness of the subsequently constructed identity authentication questions; this process can be implemented as follows: Determine the question-answering context of the current authentication process based on the identity authentication question constructed in the current authentication process recorded in the first short-term memory component; Based on the identity of the target user and the question-answer context of the current authentication process, information retrieval is performed in the joint dynamic knowledge base to obtain the user information of the target user.

[0066] Among them, the question-and-answer context refers to the contextual environment in the current authentication process, which can reflect the logical relationship between the constructed identity authentication questions; the question-asking agent can extract the contextual environment of the current authentication process, that is, the question-asking context, by analyzing factors such as the subject, type, and information field involved in the questions recorded in the first short-term memory component. For example, if the recorded questions involve the user's historical transaction information, then the question-asking intelligence will determine that the current question-and-answer context is related to the user's transaction. When performing information retrieval, it can further extract information related to the transaction. For example, previous questions involve the location and time of the transaction. New identity authentication questions can be constructed based on the name of the transaction product, and so on.

[0067] When constructing the identity authentication question, the questioning agent may also construct the subsequent identity authentication question in combination with the question-answering context and the constructed identity authentication question recorded in the first short-term memory component. That is, the question construction element may also include the question-answering context of the current authentication process, and the identity authentication question does not exist in the first short-term memory component. The process may be implemented as follows: An identity authentication question is constructed based on the user information of the target user, the question-answering context of the current authentication process, and the first preset prompt word; the identity authentication question does not exist in the first short-term memory component.

[0068] The above construction method can ensure that the newly constructed identity authentication question is consistent with the current question and answer context and different from the existing questions in the first short-term memory component. In principle, the question and answer context can be used as an index to obtain the first preset prompt word, thereby screening out the first preset prompt word that meets the current context, and then constructing an identity authentication question that meets the current context.

[0069] In some possible implementations, the questioning agent includes a first long-term memory component; the first long-term memory component stores the user's historical question and answer information.

[0070] In this case, when the questioning agent performs information retrieval, it can refer to the historical answer information stored in the first long-term memory component to determine the user's personalized preferences, so as to filter out user information that meets the user's personalized preferences, thereby constructing personalized questions for the user. This process can be implemented as follows: Determine the personalized preference of the target user based on the historical question and answer information of the target user stored in the first long-term memory component; Based on the identity of the target user and the personalized preferences of the target user, information retrieval is performed in the joint dynamic knowledge base to obtain the user information of the target user.

[0071] Among them, personalized preferences refer to the unique preferences and tendencies of users in answering questions, using systems or processing information, summarized through the analysis of users' historical question and answer information. For example, users are more inclined to answer questions about their own careers, or are better at using specific dates as answers.

[0072] In some possible implementations, when determining the personalized preferences of the target user, the questioning agent may perform a personalized preference assessment based on the length and accuracy of the user's responses to questions in historical question and answer information. For example, by analyzing historical answer information, it may be determined that the user can accurately and quickly answer questions about his or her career experience, but is slow to respond and has a low accuracy rate when answering information such as home address. This indicates that the user tends to prefer questions related to his or her career experience. Therefore, when extracting user information, user information related to the user's career experience may be preferentially extracted.

[0073] When constructing an identity authentication question, the questioning agent can also construct subsequent identity authentication questions based on personalized information, that is, the question construction elements can also include the personalized preferences of the target user; this process can be implemented as follows: The identity authentication question is constructed by asking the intelligent agent based on the user information of the target user, the personalized preferences of the target user and the first preset prompt word.

[0074] The above construction method can ensure that the newly constructed identity authentication questions meet the user's personalized preferences, which is convenient for the user to understand and answer the questions. Indicatively, the user's personalized preferences can be used as an index to obtain the first preset prompt word, so as to screen out the first preset prompt word that meets the user's personalized preferences. For example, if the user's personalized preferences indicate that the user is good at answering professional knowledge, then the first preset prompt word with strong professionalism can be selected, so that the generated identity authentication questions meet the question and answer format of the professional field; if the user's personalized preferences indicate that the user is good at popular daily communication, then the first preset prompt word with strong daily nature can be selected, so that the generated identity authentication questions tend to be easy to understand.

[0075] It should be noted that the above-mentioned method of performing information retrieval and / or question construction based on the information stored in the first short-term memory component by the questioning agent and the method of performing information retrieval and / or question construction based on the information stored in the first long-term memory component can be applied separately or in combination, and the present application does not impose any restrictions on this. That is, the identity authentication question is a question corresponding to the identity authentication of the target user based on the identity authentication request and constructed by the questioning agent based on at least one question construction element; the question construction element includes the user information of the target user obtained from the joint dynamic knowledge base, the question and answer context of the current authentication process, the personalized preferences of the target user, and the first preset prompt word.

[0076] In the process of asking the intelligent agent to construct the identity authentication question, the display interface of the target user terminal can display the question construction progress information, which is used to indicate the construction progress of the identity authentication question constructed by the questioning intelligent agent. In some possible implementations, the question construction progress information can be displayed in the form of question generation process information; the question generation process information is used to indicate the various question generation processes of the question-asking intelligent agent in constructing the identity authentication question; the question generation process includes: a user information retrieval process and a question construction process. Among them, the user information retrieval process information indicating the user information retrieval process in the question generation process information may include a knowledge base for obtaining user information, the type of information obtained, etc. For example, the user information retrieval process information may be "obtaining information from a behavioral knowledge base" or "obtaining information from a supplementary knowledge base"; the question construction process information indicating the question construction process may be "constructing the question", etc. The above information content is only for illustration, and different information content may be displayed based on different actual situations or different pre-set conditions for the question-asking intelligent agent to construct the identity authentication question, and this application does not limit this.

[0077] In some other possible implementations, the question construction progress information can be displayed in at least one of the forms of image progress information and text progress information. For example, the text progress information is displayed as "question generation progress: x%", where the value of x is 0 to 100, and when it reaches 100%, it indicates that the question construction is completed; or, the image information is displayed as a progress bar, and when the progress bar is filled, it indicates that the question construction is completed.

[0078] It should be noted that the display forms of the above-mentioned various question construction progress information can be used in combination or separately, and this application does not impose any restrictions on this.

[0079] Step 404: The identity authentication system sends the identity authentication question to the target user device, and correspondingly, the target user device receives the identity authentication question.

[0080] Step 405: The target user device displays an identity authentication question.

[0081] Step 406: The target user device receives a user answer fed back by the target user.

[0082] Step 407: the target user device sends the user answer to the identity authentication system; correspondingly, the identity authentication system receives the user answer fed back by the target user device.

[0083] Step 408: The identity authentication system evaluates the accuracy of the user's answer through an evaluation agent and generates an identity authentication result for the target user.

[0084] In some possible implementations, the identity authentication system determines the user's identity authentication result through the user's answer to an identity authentication question. In this case, the evaluation result of the user's answer is the identity authentication result of the user. At this time, if the evaluation result of the user's answer is correct, the user's identity authentication result is authentication passed; if the evaluation result of the user's answer is incorrect, the user's identity authentication result is authentication failed.

[0085] In some other possible implementations, the evaluation agent can maintain a corresponding identity authentication score for each user. The identity authentication score has a corresponding initial value, which serves as a benchmark at the beginning of the identity authentication process. The initial value can be set based on actual needs. Based on the accuracy of the user's answer, the evaluation and authentication system will update the identity authentication score (including adding or subtracting points). When the identity authentication score reaches the score threshold corresponding to the authentication pass, or reaches the score threshold corresponding to the authentication failure, it is determined that the identity authentication process is over and the identity authentication result of the target user is obtained. Based on this, the process of the identity authentication system performing accuracy evaluation through the evaluation agent can be implemented as follows: The accuracy of the user's answer is evaluated by the evaluation agent, and the evaluation result corresponding to the user's answer is obtained; Based on the evaluation result corresponding to the user's answer, the identity authentication score is updated; the identity authentication score has a preset initial value; When the identity authentication score is higher than the first score threshold, determining that the identity authentication result of the target user is authentication passed; When the identity authentication score is lower than the second scoring threshold, it is determined that the identity authentication result of the target user is authentication failure, and the second scoring threshold is less than the first scoring threshold.

[0086] Among them, the process of evaluating the accuracy of the user's answer by the evaluation agent and obtaining the evaluation result corresponding to the user's answer can be implemented as follows: The evaluation agent retrieves information in the joint dynamic knowledge base based on the identity authentication problem and the identity identifier of the target user, and obtains the identity reference information of the identity authentication problem; The evaluation agent evaluates the accuracy of the user's answer based on the identity reference information and obtains the evaluation result corresponding to the user's answer.

[0087] In some possible implementations, the evaluation method for the evaluation agent to perform accuracy evaluation may include a character verification method and a semantic similarity verification method. Each evaluation method may have a corresponding relationship with the knowledge base. Schematically, for the basic knowledge base, the user information therein is relatively fixed, and it is suitable for verification by the character verification method. For the behavioral knowledge base, the user information expression therein is relatively flexible, and it is suitable for the semantic similarity verification method. It should be noted that based on different actual needs, the corresponding relationship between different types of knowledge bases and evaluation methods may have different settings. The same type of knowledge base may correspond to one or more verification methods, and this application does not limit this. Based on this, the process of the evaluation agent performing accuracy evaluation can be implemented as follows: Determine an evaluation method by evaluating the type of knowledge base to which the agent belongs based on the identity reference information, the evaluation method including at least one of the following: a character verification method and a semantic similarity verification method; Based on the answer evaluation element, the user's answer is evaluated for accuracy according to the evaluation method to obtain an evaluation result corresponding to the user's answer; the answer evaluation element includes identity reference information and a second preset prompt word.

[0088] The second preset prompt word may be a keyword or phrase pre-set by relevant personnel, used to guide the evaluation agent to perform accuracy evaluation on the user's answer based on the identity reference information in a determined evaluation method; optionally, the evaluation agent may construct evaluation prompt information based on the above content to perform accuracy evaluation based on the evaluation prompt information. For example, the constructed evaluation prompt information may be "Please perform accuracy evaluation on the user's answer based on the question and answer context and the reference identity information in a semantic similarity verification manner", etc. This application does not impose any restrictions on this.

[0089] When evaluating the accuracy of user answers according to the character verification method, the evaluation agent can check the format of the user's answer. If the format is incorrect, the evaluation result is that the user answered incorrectly. If the format is correct, the identity reference information can be compared with the user's answer character by character. When all characters are the same, the evaluation result is determined to be that the user answered correctly. When one character is wrong, the evaluation result is determined to be that the user answered incorrectly. Schematically, when verifying a mobile phone number, first verify whether the user's answer is in an 11-digit format. If the format is incorrect, if the format is correct, further compare the characters of the identity reference information with the characters of the user's answer character by character. When all are consistent, the user's answer is determined to be correct. Otherwise, the answer is determined to be incorrect.

[0090] When evaluating the accuracy of user answers according to the semantic similarity verification method, the evaluation intelligent system can calculate the similarity between the identity reference information and the user answer (such as cosine similarity). When the similarity between the two is greater than a preset similarity threshold, the evaluation intelligent agent determines that the user answer is semantically consistent with the identity reference information and determines that the user answer is correct. When the similarity between the two is less than the preset similarity threshold, the evaluation intelligent agent determines that the user answer is semantically inconsistent with the identity reference information and determines that the user answer is wrong.

[0091] In some possible implementations, the evaluation agent may also maintain a problem difficulty evaluation rule, which may be used to evaluate the relative difficulty of each identity authentication question. Different difficulty levels correspond to different weights, and the relationship between the difficulty level of the question and the weight may be set by relevant personnel, which may be a positive correlation or a negative correlation. In this case, when updating the identity authentication score, the score may be updated with reference to the weight corresponding to the difficulty of the identity authentication question. The basic score of each identity authentication question, in this case, if the evaluation result indicates that the user has answered correctly, the weighted value of the basic score and the weight of the identity authentication question is added to the current identity authentication score. If the evaluation result indicates that the user has failed to answer, the weighted value of the basic score and the weight of the identity authentication question is subtracted from the current identity authentication score. In schematic form, if the basic score of each identity authentication question is 10 points, the current identity authentication score is 50 points, and the weight corresponding to the current identity authentication question is 1, then if the user answers correctly, the updated identity authentication score is 60 points, and if the user answers incorrectly, the updated identity authentication score is 40 points.

[0092] When the updated identity authentication score is higher than the first scoring threshold, it is determined that the user has passed the identity authentication. Schematically, assuming that the first scoring threshold is set to 80 points, if the identity authentication score reaches or exceeds 80 points after the user answers multiple identity authentication questions, it is determined that the user has passed the identity authentication; when the updated identity authentication score is lower than the second scoring threshold, it is determined that the user has failed the identity authentication. Schematically, assuming that the second scoring threshold is set to 30 points, if the identity authentication score reaches or is lower than 30 points after the user answers multiple identity authentication questions, it is determined that the user has failed the identity authentication.

[0093] It should be noted that the settings of the first scoring threshold and the second scoring threshold can be flexibly adjusted according to different scenarios or security policies. For example, in certain high-risk scenarios, higher first scoring thresholds and second scoring thresholds can be set to increase the difficulty of identity authentication; in low-risk scenarios, lower first scoring thresholds and second scoring thresholds can be set to reduce the difficulty of identity authentication and improve authentication efficiency.

[0094] When the authentication score does not reach the corresponding scoring threshold for authentication pass, nor does it reach the corresponding scoring threshold for authentication failure, the evaluation agent cannot determine the authentication result of the current user, and determines that the current authentication process is not over. The authentication process is repeated, including re-constructing the authentication question and evaluating based on the feedback user answer; this process can be implemented as follows: When the identity authentication score is greater than the second score threshold and less than the first score threshold, the identity authentication question corresponding to the target user is constructed again by questioning the intelligent agent.

[0095] Illustratively, when the first scoring threshold is 80 points and the second scoring threshold is 30 points, if the user's identity authentication score is 60 points after answering multiple identity authentication questions, which has reached neither the first scoring threshold nor the second scoring threshold, the evaluation agent cannot determine the identity evaluation result of the current user. The evaluation agent can send a question reconstruction instruction to the questioning agent to instruct the questioning agent to reconstruct the identity authentication question corresponding to the user.

[0096] After entering the re-authentication process, in order to avoid the problem that the accuracy of the user's feedback is low due to the difficulty in understanding the constructed authentication questions, the questioning agent can calculate the correlation value between the historical authentication questions in the current authentication process and the corresponding user answers when re-constructing the authentication questions corresponding to the target user; The questioning agent reconstructs the identity authentication question corresponding to the target user based on the correlation value between the historical identity authentication questions and the corresponding user answers in the current authentication process; the elements of the reconstructed identity authentication question include at least one of the following: the type of updated user information, the knowledge base to which the updated user information belongs, and the updated first preset prompt word.

[0097] In some possible implementations, the questioning agent can calculate the relevance value from multiple dimensions such as text similarity calculation and topic association analysis; wherein, when calculating text similarity, a similarity algorithm can be used to convert the historical identity authentication questions and the corresponding user answers into vector form, based on the word vector, and the similarity value in text semantics is determined by calculating the similarity value between the vectors; when analyzing topic association, the questioning agent can introduce the topic model in natural language processing, and use the model to determine the topic category to which each question and answer belongs, so as to determine the similarity value between the two; wherein, when performing multi-dimensional correlation calculation, the similarity value between the identity authentication question and the corresponding user answer can be comprehensively determined based on the similarity value calculation results of multiple dimensions, for example, the weighted sum of the similarity values ​​of multiple dimensions is determined as the final similarity value, etc. Optionally, the above-mentioned historical identity authentication questions can be identity authentication questions that have been generated in the current identity authentication process, or they can also be historical identity authentication questions within the current time target time period in the current identity authentication process. The acquisition time range of the historical identity authentication questions can be set based on actual needs, and this application does not limit this.

[0098] In some possible implementations, after obtaining the similarity between historical identity authentication questions and corresponding user answers, the questioning agent can select historical identity authentication questions and user answer combinations with higher similarity values ​​for analysis to determine questioning methods, question and answer fields, etc. that are easy for users to understand, so as to construct identity authentication questions corresponding to the questioning methods and / or question and answer fields when constructing identity authentication questions subsequently.

[0099] In some other possible implementations, the process of reconstructing the identity authentication question corresponding to the target user by asking the intelligent agent based on the correlation value between the historical identity authentication question in the current authentication process and the corresponding user answer can be implemented as follows: Determine a comprehensive correlation value based on the correlation values ​​between the historical identity authentication questions and the corresponding user answers in the current authentication process; the comprehensive correlation value is obtained by integrating the various correlation values, and is used to comprehensively reflect the overall degree of correlation between the historical identity authentication questions and the user answers in the authentication process; When the comprehensive relevance value is lower than the relevance threshold, the elements of the identity authentication question are updated and reconstructed; The identity authentication question corresponding to the target user is reconstructed by asking the agent based on the updated elements of the constructed identity authentication question.

[0100] Among them, the comprehensive correlation value can be the average or median of the correlation values ​​between the historical identity authentication questions and the corresponding user answers in the current authentication process, or it can also be a value calculated by other means, such as the value obtained by weighted summation through the target rule; if the comprehensive correlation value is lower than the correlation threshold, it indicates that the effect of the constructed identity authentication question is poor, and the question construction elements need to be updated.

[0101] When updating the elements, at least one of the following elements may be updated: the type of user information, the knowledge base to which the user information belongs, and the first preset prompt word; when updating the type of user information, the type of information with a higher accuracy rate of user answers may be extracted during information retrieval, wherein the accuracy rate of user answers to each type of information may be obtained based on historical data statistics; when updating the knowledge base to which the user information belongs, if the comprehensive relevance of the user answer to the identity authentication question constructed based on the user information extracted from the current knowledge base is low, then switch to another knowledge base for information retrieval; when updating the first preset prompt word, obtain the comprehensive matching value corresponding to each first preset prompt word, if some prompt words frequently cause the user to answer incorrectly or have difficulty in understanding, that is, the comprehensive matching value corresponding to these prompt words is low, then modify or replace these prompt words; for example, if the first preset prompt word "your commonly used contact method" causes ambiguity to the user, that is, the correlation between the user answer and the corresponding identity authentication question is low, then it can be changed to "which mobile phone number do you usually receive important notifications from" and so on.

[0102] After the question construction elements are updated, the question-asking agent re-executes the question construction process based on the updated question construction elements. The process can refer to the relevant content corresponding to step 403 and will not be repeated here.

[0103] When evaluating user answers, the evaluation agent can retrieve information from the joint dynamic knowledge base to obtain identity reference information of the target user corresponding to the current identity authentication question; in some possible implementations, the evaluation agent includes a second short-term memory component; the second short-term memory component records the user answers received during the current authentication process.

[0104] In this case, when evaluating the user's answer, the evaluation agent can refer to the user's answer received in the current authentication process recorded in the second short-term memory component to determine the question-answer context of the current authentication process, so as to determine the identity reference information of the identity authentication question in the current question-answer context, thereby improving the accuracy of the accuracy evaluation of the user's answer based on the reference information. The process can be implemented as follows: Determine the question-answer context of the current authentication process based on the user answers received in the current authentication process and recorded in the second short-term memory component; The identity reference information is obtained by evaluating the intelligent agent to retrieve information in the joint dynamic knowledge base based on the identity authentication question, the identity of the target user and the question-answering context of the current authentication process.

[0105] The evaluation agent can analyze factors such as the subject, type, and information field involved in the user's answers recorded in the second short-term memory component, and extract the contextual environment of the current authentication process, that is, the question and answer context; when obtaining reference information, the user's identity identifier can be used as an index to retrieve user information that conforms to the current question and answer context and matches the identity authentication question in the joint dynamic knowledge base as identity reference information, and the user's answer is evaluated for accuracy in combination with the identity reference information and the second preset prompt word; optionally, the user information that matches the identity authentication question can be user information whose semantic relevance is greater than a semantic relevance threshold.

[0106] Furthermore, when performing accuracy evaluation, the evaluation agent can also perform accuracy evaluation in combination with the current question-answer context, that is, the answer evaluation element can also include the answer context of the current authentication process; this process can be implemented as follows: The evaluation agent evaluates the accuracy of the user's answer based on the identity reference information, the question and answer context of the current authentication process, and the second preset prompt word according to the evaluation method to obtain the evaluation result corresponding to the user's answer.

[0107] The above evaluation method can make it possible to consider the accuracy of the user's answer in the current question-and-answer context during question-and-answer evaluation, thereby improving the evaluation effect. In principle, if the user's answer does not match the topic or scenario in the current question-and-answer context, the evaluation agent will also identify it as a wrong answer.

[0108] In some possible implementations, the evaluation agent includes a second long-term memory component; the second long-term memory component stores the user's historical question and answer information.

[0109] In this case, when evaluating the user's answer, the evaluation agent can refer to the user's historical question and answer information stored in the second long-term memory component to determine the user's personalized preferences, so as to determine the identity reference information corresponding to the identity authentication question that meets the user's personalized preferences, thereby improving the accuracy of the accuracy evaluation of the user's answer based on the reference information. This process can be implemented as follows: Determine the personalized preferences of the target user by evaluating the historical question and answer information of the target user stored in the second long-term memory component of the agent; The identity reference information is obtained by evaluating the intelligent agent to retrieve information in the joint dynamic knowledge base based on the identity authentication problem, the target user's identity and the target user's personalized preferences.

[0110] The process of the evaluation agent determining the personalized preferences of the target user can refer to the process of the questioning agent determining the personalized preferences of the target user, which will not be repeated here.

[0111] Furthermore, when performing accuracy evaluation, the evaluation agent can also perform accuracy evaluation in combination with the user's personalized preferences. That is, the answer evaluation element also includes the personalized preferences of the target user. This process can be implemented as follows: The evaluation agent evaluates the accuracy of the user's answer based on the identity reference information, the personalized preferences of the target user, the second preset prompt word and the evaluation method to obtain the evaluation result.

[0112] Through the above-mentioned evaluation method, the influence of the user's personalized preferences on the user's answers can be considered in the question and answer evaluation, and then the accuracy of the user's answers can be judged to improve the evaluation effect; in principle, if the user's personalized preferences indicate that the user prefers to use "terminal" instead of "mobile phone" when answering questions, then when making accuracy judgments, a corresponding relationship between the two can be established to avoid misjudgment.

[0113] It should be noted that the above-mentioned method of obtaining reference information and evaluating user answers based on the information stored in the second short-term memory component by the evaluation agent can be applied separately or in combination with the method of obtaining reference information and evaluating user answers based on the information stored in the second long-term memory component, and this application does not limit this. In other words, the identity authentication result is generated by the evaluation agent after the accuracy is evaluated based on at least one answer evaluation element, and the answer evaluation element includes the identity reference information of the identity authentication question, the question and answer context of the current authentication process, the personalized preferences of the target user, and the second preset prompt word.

[0114] During the accuracy evaluation process of the evaluation agent, the evaluation progress information can be displayed on the target terminal interface, and the evaluation progress information is used to indicate the generation progress of the identity authentication result generated by the evaluation agent.

[0115] In some possible implementations, the evaluation progress information can be displayed in the form of evaluation process information, which is used to indicate the various evaluation processes of the evaluation agent to generate the identity authentication result; the evaluation process includes: a reference information retrieval process and an accuracy evaluation process. Among them, the identity reference information retrieval process information indicating the reference information retrieval process in the evaluation process information can be "reference information acquisition", and the accuracy evaluation process information indicating the accuracy evaluation process can include evaluation method information, such as "answer semantic analysis is being performed", or, it can also be "answer evaluation is being performed", etc. The above information content is only illustrative, and different information content can be displayed based on the actual situation of the evaluation agent performing accuracy evaluation or different pre-sets, and this application does not limit this; in addition, for the user's answer to the current identity authentication question, after determining the evaluation result of the user's answer, the evaluation agent can feed back the evaluation result of the user's answer to the target user device, so that the target user device displays the evaluation result of the current user's answer, and the evaluation result includes a correct answer or an incorrect answer.

[0116] In some other possible implementations, the evaluation progress information can be displayed in at least one of the forms of image progress information and text progress information. For example, the text progress information is displayed as "Answer evaluation progress: x%", where the value of x is 0 to 100, and when it reaches 100%, it indicates that the answer evaluation is completed; or, the image information is displayed as a progress bar, and when the progress bar is filled, it indicates that the answer evaluation is completed.

[0117] It should be noted that the above-mentioned various display forms of evaluation progress information can be used in combination or separately, and this application does not impose any restrictions on this.

[0118] Figure 5 FIG. 4 shows a schematic diagram of the working of an identity authentication system provided by an exemplary embodiment of the present application. Figure 5As shown, after receiving the identity authentication request, the evaluation agent first starts the questioning agent 510. The questioning agent generates question construction prompt information based on the identity authentication request through the joint dynamic knowledge base and the preset prompt words, and then the large language model processes the question construction prompt information to obtain the identity authentication question; optionally, in the above process, the questioning agent can implement each step by calling tools, such as performing information retrieval in the joint dynamic knowledge base through the knowledge collection tool, generating question construction prompt information through the knowledge association tool, and calling the large language model through the question raising tool, etc. In addition, when generating questions, the questioning agent can use the first short-term memory component to record the question content and tool usage in the current session, so as to provide a question and answer context reference for question construction, ensure the real-time and personalization of each question generation, and avoid repeated questions in the same session; the first long-term memory component is used to store historical question and answer information, so as to provide a user personalized preference reference for question construction. After the questioning agent generates the identity authentication question, it displays the identity authentication question to the user through the display interface of the target user device. The user can return the user answer to the evaluation agent 520 based on the identity authentication question. After receiving the user answer, the evaluation agent can generate evaluation prompt information through the joint dynamic knowledge base and the preset prompt words, and then the large language model processes the evaluation prompt information to obtain the evaluation result corresponding to the user answer. Optionally, in the above process, the evaluation agent can implement each step by calling tools, such as determining the type of identity authentication question through the question classification tool, so as to retrieve the identity reference information in the joint dynamic knowledge base; when the verification method is determined to be semantic similarity verification, the accuracy is verified by the semantic similarity verification tool, and when the verification method is determined to be character verification, the accuracy is verified by the character verification tool. In addition, in the above process, the evaluation agent can use the second short-term memory component to record the user answer and tool usage in the current session to provide a question and answer context reference for the answer evaluation; and can also use the second long-term memory component to record historical question and answer information to provide a user personalized information preference reference for the answer evaluation.

[0119] Step 409: The identity authentication system sends the identity authentication result of the target user to the target user device. Correspondingly, the target user device receives the identity authentication result of the target user.

[0120] When the evaluation agent determines the authentication result based on the authentication score, the user's authentication result can be determined after the user answers multiple authentication questions, and the determined authentication result is sent to the target user device; otherwise, the authentication continues.

[0121] Step 410: The target user device displays the target user's identity authentication result.

[0122] In summary, the identity authentication method provided by the embodiment of the present application, through the interaction between the target user device and the identity authentication system, after the identity authentication system receives the identity authentication request containing the identity identifier of the target user sent by the target user device, based on the identity authentication request, the questioning agent generates an identity authentication question corresponding to the target user based on the user information in the joint dynamic knowledge base and the identity authentication request and sends it to the target user device; when receiving the user answer fed back by the target user device, the evaluation agent performs an accuracy evaluation on the user answer to generate an identity authentication result. Through the above method, question-and-answer identity authentication can be performed through the questioning agent, the evaluation agent and the joint dynamic knowledge base, thereby realizing personalized identity authentication corresponding to each user, thereby reducing dependence on central information sources and improving the security of identity authentication; in addition, since the user information in the joint dynamic knowledge base is updated in real time, the timeliness and pertinence of the identity authentication question can be guaranteed, thereby further improving the reliability of identity authentication.

[0123] In addition, the questioning agent and the evaluation agent can construct identity authentication questions and evaluate the accuracy of user answers based on the question and answer context and / or the user's personalized preferences, so that the formulated identity authentication questions are more in line with user authentication and actual conditions, improve the question construction effect, and enable more comprehensive judgments when evaluating user answers, thereby improving the accuracy of identity authentication.

[0124] Figure 6 FIG. 1 shows an interaction diagram of an identity authentication method provided by an exemplary embodiment of the present application, involving the interaction between a user device 610, a questioning agent 620, an evaluation agent 630, and a joint dynamic knowledge base 640, such as Figure 6 As shown, the user sends an identity authentication request through the display interface of the user device; the user device sends the received identity authentication request to the questioning agent, and the questioning agent obtains the user information of the user from the joint dynamic knowledge base. Correspondingly, the joint dynamic knowledge base returns the user information of the user to the questioning agent; the questioning agent generates and feeds back the identity authentication questions to the user device based on the user information, and the user device displays the identity authentication questions through the display interface.

[0125] The user feeds back the user answer to the user device based on the identity authentication information, and the user device submits the user answer to the evaluation agent. After receiving the user answer, the evaluation agent retrieves the identity reference information corresponding to the identity authentication question in the joint dynamic knowledge base. Accordingly, the joint dynamic knowledge base feeds back the identity authentication reference information to the evaluation agent. The evaluation agent evaluates the user answer based on the identity reference information, and feeds back the evaluation result to the user device. The user device feeds back the evaluation result to the user through a display interface, wherein the evaluation result includes whether the answer is correct or failed. In addition, the evaluation agent determines the identity authentication result of the user based on the evaluation result of the user answer, and after obtaining the identity authentication result of the user, feeds back the identity authentication result to the user device. The user device feeds back the identity authentication result to the user through a display interface, and the identity authentication result indicates whether the identity authentication passed or failed.

[0126] Since the identity authentication method provided in the embodiment of the present application involves interaction between a user terminal and an identity authentication system, in order to ensure the effective implementation of the identity authentication method, when applying it, the application environment may have the following conditions: 1. In terms of network conditions, a stable network connection should be guaranteed to ensure real-time communication between the user terminal and the identity authentication system, between the various intelligent agents in the identity authentication system, and between the various intelligent agents and the joint dynamic knowledge base, so as to ensure the timeliness and accuracy of data transmission during the identity authentication process; in order to ensure the security of user data, data transmission in the network can be encrypted, such as SSL / TLS protocol encryption, etc., to ensure that the data is not eavesdropped or tampered with during transmission. In addition, network security measures such as firewalls and intrusion detection systems can be deployed to prevent unauthorized access or attacks; in order to ensure the continuous operation of the system and avoid single point failures, high-availability architecture designs such as load balancing and redundant backup can be adopted. In addition, a data backup and recovery mechanism can be established to ensure that normal services can be quickly restored in the event of an unexpected situation.

[0127] 2. In terms of technical support, since the operation of the multi-intelligent system and the joint dynamic knowledge base requires a large amount of computing resource support, high-performance servers, sufficient storage space and efficient database management systems can be used when deploying the system to provide sufficient data resources to meet computing needs and reduce the pressure of high concurrent access; in addition, the joint dynamic knowledge base must also be flexible and scalable so that information can be dynamically updated according to the real-time behavior of users. Furthermore, the joint dynamic knowledge base can adopt distributed database technology to support data reading and writing work in colleges and universities, and provide a knowledge base configuration interface to facilitate data management and maintenance; since the embodiments of this application involve the collection and processing of user information, in order to protect user privacy, before collecting and processing information, an authorization request can be sent to the user, and corresponding operations can be performed after receiving user authorization; in addition, the system can also formulate and implement privacy protection strategies to protect the security of user information.

[0128] Figure 7 The block diagram of the structure of a computer device 700 shown in an exemplary embodiment of the present application is shown. The computer device can be implemented as a questioning agent or an evaluation agent in the above-mentioned solution of the present application. The computer device 700 includes a processor (such as a central processing unit (CPU)) 701, a system memory 704 including a random access memory (RAM) 702 and a read-only memory (ROM) 703, and a system bus 705 connecting the system memory 704 and the processor 701.

[0129] The computer device 700 further includes a mass storage device 706 for storing an operating system 709 , application program instances 710 and other program modules 711 . The system memory 704 and the mass storage device 706 may be collectively referred to as memory.

[0130] According to various embodiments of the present application, the computer device 700 can also be connected to a remote computer on the network through a network such as the Internet. That is, the computer device 700 can be connected to the network 708 through the network interface unit 707 connected to the system bus 705, or the network interface unit 707 can be used to connect to other types of networks or remote computer systems (not shown).

[0131] The memory further stores at least one computer program, and the processor 701 implements all or part of the steps in the identity authentication method shown in the above-mentioned embodiments by executing the at least one computer program.

[0132] Figure 8The structural block diagram of another computer device 800 shown in an exemplary embodiment of the present application is shown. The computer device 800 can be implemented as the above-mentioned target user device. For example, the computer device can be an Android terminal device; generally, the computer device 800 includes: a processor 801 and a memory 802. The memory 802 may include one or more computer-readable storage media, and the computer-readable storage medium is used to store at least one instruction, and the at least one instruction is used to be executed by the processor 801 to implement all or part of the steps in the identity authentication method shown in the method embodiment of the present application. In some embodiments, the computer device 800 may also optionally include: a peripheral device interface 803 and at least one peripheral device. The processor 801, the memory 802 and the peripheral device interface 803 can be connected through a bus or a signal line. Each peripheral device can be connected to the peripheral device interface 803 through a bus, a signal line or a circuit board. Specifically, the peripheral device includes: at least one of a radio frequency circuit 804, a display screen 805, a camera assembly 806, an audio circuit 807 and a power supply 808. In some embodiments, the computer device 800 also includes one or more sensors 809. The one or more sensors 809 include but are not limited to: an acceleration sensor 810, a gyroscope sensor 811, a pressure sensor 812, an optical sensor 813, and a proximity sensor 814. Those skilled in the art will appreciate that Figure 8 The structure shown in the figure does not constitute a limitation on the computer device 800, and the computer device 800 may include more or less components than those shown in the figure, or combine some components, or adopt a different arrangement of components.

[0133] In an exemplary embodiment, a computer-readable storage medium is also provided, in which at least one computer program is stored, and the computer program is loaded and executed by a processor to implement all or part of the steps in the above-mentioned identity authentication method. For example, the computer-readable storage medium can be a read-only memory (ROM), a random access memory (RAM), a compact disc (CD-ROM), a magnetic tape, a floppy disk, an optical data storage device, etc.

[0134] In an exemplary embodiment, a computer program product is also provided, the computer program product comprising a computer program stored on a non-transitory computer-readable storage medium, the computer program comprising program instructions, when the program instructions are executed by a computer, the computer is executed to implement the above Figure 1 , Figure 2 or Figure 4 All or part of the steps of the embodiments shown in any embodiment.

[0135] Those skilled in the art will readily appreciate other embodiments of the present application after considering the specification and practicing the invention disclosed herein. The present application is intended to cover any modification, use or adaptation of the present application, which follows the general principles of the present application and includes common knowledge or customary techniques in the art that are not disclosed in the present application. The specification and examples are intended to be exemplary only, and the true scope and spirit of the present application are indicated by the claims.

[0136] It should be understood that the present application is not limited to the precise structures that have been described above and shown in the drawings, and that various modifications and changes may be made without departing from the scope thereof. The scope of the present application is limited only by the appended claims.

Claims

1. An identity authentication method, characterized in that: include: Receiving an identity authentication request, wherein the identity authentication request includes an identity identifier of a target user; Displaying an identity authentication question, wherein the identity authentication question is a question corresponding to the identity authentication of the target user generated by the questioning agent based on the identity authentication request and based on the user information in the joint dynamic knowledge base; Receiving a user answer fed back by the target user; The identity authentication result of the target user is displayed, wherein the identity authentication result is generated after the evaluation agent evaluates the accuracy of the user's answer.

2. The method according to claim 1, characterized in that: The joint dynamic knowledge base includes at least one of the following: a basic knowledge base, a behavioral knowledge base and a supplementary knowledge base; the basic knowledge base stores static basic information of users, the behavioral knowledge base stores dynamic behavioral information of users; the supplementary knowledge base stores personalized user information uploaded by users.

3. The method according to claim 1 or 2, characterized in that: Before displaying the identity authentication question, the method further includes: The question construction progress information is displayed, where the question construction progress information is used to indicate the construction progress of the identity authentication question by the questioning agent.

4. The method according to claim 1 or 2, characterized in that: Before displaying the identity authentication result of the target user, the method further includes: Evaluation progress information is displayed, where the evaluation progress information is used to indicate the generation progress of the identity authentication result generated by the evaluation agent.

5. The method according to claim 1, characterized in that The identity authentication question is a question corresponding to the identity authentication of the target user, which is constructed by the questioning agent based on the identity authentication request and based on at least one question construction element; the question construction element includes the user information of the target user obtained from the joint dynamic knowledge base, the question and answer context of the current authentication process, the personalized preferences of the target user, and a first preset prompt word.

6. The method according to claim 1, characterized in that The identity authentication result is generated by the evaluation agent after an accuracy evaluation is performed based on at least one answer evaluation element, wherein the answer evaluation element includes the identity reference information of the identity authentication question, the question and answer context of the current authentication process, the personalized preferences of the target user, and a second preset prompt word.

7. An identity authentication method, characterized in that: include: Receiving an identity authentication request sent by a target user device; the identity authentication request includes an identity identifier of the target user; Based on the identity authentication request, generate an identity authentication question corresponding to the target user by asking an intelligent agent, and send the identity authentication question to the target user device; The questioning agent is used to generate identity authentication questions based on user information in the joint dynamic knowledge base; When receiving the user answer fed back by the target user device, the user answer is evaluated for accuracy by an evaluation agent to generate an identity authentication result of the target user.

8. The method according to claim 7, characterized in that The step of generating an identity authentication question corresponding to the target user by asking an intelligent agent based on the identity authentication request includes: Based on the identity identifier of the target user included in the identity authentication request, the questioning agent searches the joint dynamic knowledge base to obtain user information of the target user; An identity authentication question corresponding to the target user is constructed based on a question construction element, wherein the question construction element includes user information of the target user and a first preset prompt word.

9. The method according to claim 8, characterized in that The questioning agent includes a first short-term memory component; the first short-term memory component records the identity authentication questions constructed in the current authentication process; The step of obtaining the user information of the target user by performing information retrieval in the joint dynamic knowledge base by the questioning agent includes: Determine the question-and-answer context of the current authentication process based on the identity authentication question constructed in the current authentication process recorded in the first short-term memory component; Based on the identity of the target user and the question-answer context of the current authentication process, information retrieval is performed in the joint dynamic knowledge base to obtain the user information of the target user.

10. The method according to claim 9, characterized in that The question construction element also includes the question-answering context of the current authentication process, and the identity authentication question does not exist in the first short-term memory component.

11. The method according to claim 8, characterized in that The questioning agent includes a first long-term memory component; the first long-term memory component stores the user's historical question and answer information; The step of obtaining the user information of the target user by performing information retrieval in the joint dynamic knowledge base by the questioning agent includes: Determining the personalized preference of the target user based on the historical question and answer information of the target user stored in the first long-term memory component; Information retrieval is performed in the joint dynamic knowledge base based on the identity identifier of the target user and the personalized preference of the target user to obtain the user information of the target user.

12. The method according to claim 11, characterized in that The question construction element also includes the personalized preferences of the target user.

13. The method according to claim 7, characterized in that The step of evaluating the accuracy of the user's answer by an evaluation agent to generate an identity authentication result of the target user includes: The evaluation agent evaluates the accuracy of the user's answer to obtain an evaluation result corresponding to the user's answer; Based on the evaluation result corresponding to the user's answer, updating the identity authentication score; the identity authentication score has a preset initial value; When the identity authentication score is higher than a first score threshold, determining that the identity authentication result of the target user is authentication passed; In the case where the identity authentication score is lower than a second scoring threshold, determining that the identity authentication result of the target user is authentication failure, and the second scoring threshold is less than the first scoring threshold; When the identity authentication score is greater than the second score threshold and less than the first score threshold, the identity authentication question corresponding to the target user is reconstructed by the questioning agent.

14. The method according to claim 13, characterized in that The step of reconstructing the identity authentication question corresponding to the target user through the questioning agent includes: Calculate the correlation value between the historical identity authentication questions and the corresponding user answers in the current authentication process; The questioning agent reconstructs the identity authentication question corresponding to the target user based on the correlation value between the historical identity authentication questions and the corresponding user answers in the current authentication process; the elements of the reconstructed identity authentication question include at least one of the following: the type of updated user information, the knowledge base to which the updated user information belongs, and the updated first preset prompt word.

15. The method according to claim 13, characterized in that The step of evaluating the accuracy of the user's answer by an evaluation agent to obtain an evaluation result corresponding to the user's answer includes: The evaluation agent retrieves information in the joint dynamic knowledge base based on the identity authentication problem and the identity identifier of the target user to obtain identity reference information of the identity authentication problem; The evaluation agent evaluates the accuracy of the user's answer based on the identity reference information to obtain an evaluation result corresponding to the user's answer.

16. The method according to claim 15, characterized in that The step of evaluating the accuracy of the user's answer based on the identity reference information by the evaluation agent to obtain an evaluation result corresponding to the user's answer includes: Determining an evaluation method based on the type of knowledge base to which the identity reference information belongs, the evaluation method comprising at least one of the following: a character verification method and a semantic similarity verification method; Based on the answer evaluation element, the accuracy of the user answer is evaluated according to the evaluation method to obtain an evaluation result corresponding to the user answer; the answer evaluation element includes the identity reference information and the second preset prompt word.

17. The method according to claim 15, characterized in that The evaluation agent includes a second short-term memory component; the second short-term memory component records the user answers received during the current authentication process; The step of performing information retrieval in the joint dynamic knowledge base based on the identity authentication problem and the identity identifier of the target user by the evaluation agent to obtain the identity reference information of the identity authentication problem includes: Determine the question-answer context of the current authentication process based on the user answers received in the current authentication process and recorded in the second short-term memory component; Based on the identity authentication question, the identity identifier of the target user and the question-answer context of the current authentication process, information retrieval is performed in the joint dynamic knowledge base to obtain the identity reference information.

18. The method according to claim 17, characterized in that The answer assessment element also includes the question and answer context of the current authentication process.

19. The method according to claim 15, characterized in that The evaluation agent includes a second long-term memory component; the second long-term memory component stores the user's historical question and answer information; The step of performing information retrieval in the joint dynamic knowledge base based on the identity authentication problem and the identity identifier of the target user by the evaluation agent to obtain the identity reference information of the identity authentication problem includes: Determining the personalized preference of the target user based on the historical question and answer information of the target user stored in the second long-term memory component by the evaluation agent; The identity reference information is obtained by the evaluation agent performing information retrieval in the joint dynamic knowledge base based on the identity authentication problem, the target user identity identifier and the target user's personalized preference.

20. The method according to claim 19, characterized in that The answer evaluation element also includes the personalized preference of the target user.

Images