Identity authentication system and method based on natural language processing

Abstract

The invention relates to an identity authentication system and method based on natural language processing, and relates to the technical field of user identity authentication. The method comprises the following steps: constructing a user portrait according to user information of an enterprise user; generating a verification question bank according to the logged-in user account; adjusting the weight coefficient of the skill problem according to the difficulty level fed back by the client; answering information fed back by the client is collected, and the correct coefficient of each group of conventional questions and the correct coefficient of each group of skill questions are obtained; generating an authentication passing coefficient according to the correct coefficient of each group of conventional problems, the correct coefficient of each group of skill problems and the proportion weight of the skill problems; and analyzing whether authentication is passed according to the authentication passing coefficient. According to the method, the weight of the skill problem is adaptively adjusted according to the difficulty rating of the skill problem by the user, so that the authentication passing coefficient calculated according to the correct coefficient of each group of problems is more reasonable, and the user identity authentication accuracy of an enterprise management platform is effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of user identity authentication, and particularly to an identity authentication system and method based on natural language processing. Background Art

[0002] With the improvement of network security awareness, multi-factor authentication has become a trend, and question verification is one of the effective means to achieve this goal.

[0003] For example, when an enterprise user accesses an enterprise management platform, they need to first enter the login page of the authentication center and enter their username and password in the authentication center for login verification. To reduce the risk of misoperation or illegal login, after the verification is passed, the authentication center will generate question verification and return it to the user.

[0004] Existing natural language generation algorithms have achieved good results in generating questions based on given text information. These algorithms can understand and process natural language, thus generating questions that conform to the context and grammar. However, due to the differences between user portraits and text information, when generating a verification question bank based on user portraits, there may be a situation where the difficulty of verification questions does not match the actual situation of employees. Summary of the Invention

[0005] Based on this, in view of the above problems, it is necessary to provide an identity authentication system and method based on natural language processing that can adaptively adjust the weight of verification questions according to user feedback.

[0006] An identity authentication method based on natural language processing provided by the present invention includes:

[0007] Collecting user information of enterprise users;

[0008] Constructing a user portrait according to the user information of enterprise users;

[0009] Generating a verification question bank according to the user portrait corresponding to the logged-in user account, where the verification question bank includes regular questions, skill questions, and corresponding correct answers, and the skill questions include skill questions and corresponding difficulty level self-evaluation options;

[0010] Adjusting the weight coefficient of the skill questions according to the difficulty level feedback by the client, where the value of the weight coefficient ranges from 0 to 1;

[0011] Collecting the response information feedback by the client to obtain the correct coefficients of each group of regular questions and the correct coefficients of each group of skill questions;

[0012] Generating an authentication pass coefficient according to the correct coefficients of each group of regular questions, the correct coefficients of each group of skill questions, and the proportion weight of the skill questions;

[0013] Analyze whether the authentication is passed according to the authentication passing coefficient.

[0014] In one embodiment, the user information includes the user's basic information, work resume, and behavior data.

[0015] In one embodiment, the steps of constructing a user portrait based on the user information of an enterprise user include:

[0016] Use natural language processing to parse the user's work resume and extract key features, where the key features include years of work experience, position, number of projects, project types, and skill tags;

[0017] Integrate the key features with the basic information and behavior data to obtain a complete user portrait.

[0018] In one embodiment, the steps of generating a verification question bank based on the user portrait corresponding to the logged-in user account include:

[0019] Generate general questions using a language generation model based on the basic information, years of work experience, position, number of projects, and project types;

[0020] Retrieve the correct answers from the key information and relevant relationships based on the general questions;

[0021] Generate relevant distractors based on the correct answers.

[0022] In one embodiment, the steps of generating a verification question bank based on the user portrait corresponding to the logged-in user account further include:

[0023] Determine the type of technical knowledge question bank to be retrieved according to the skill tags;

[0024] Retrieve the corresponding skill question bank from external question bank resources according to the type of technical knowledge question bank;

[0025] Match a difficulty level self-assessment option for each skill question.

[0026] In one embodiment, the steps of adjusting the weight coefficient of skill questions according to the difficulty level feedback from the client include:

[0027] According to the rule that the lower the difficulty level, the higher the weight coefficient, set the weight coefficient according to the preset difficulty level, and the value of the weight coefficient ranges from 0 to 1;

[0028] Determine the corresponding weight coefficient according to the difficulty level corresponding to the skill question.

[0029] In one embodiment, the steps of comparing the answer information with the correct answers to obtain the correct coefficients of each group of general questions and the correct coefficients of each group of skill questions include:

[0030] Compare the response information with the correct answers belonging to the same question. If the response information is consistent with the correct answers, the correct coefficient takes a value of 1; if the response information is inconsistent with the correct answers, the correct coefficient takes a value of 0.

[0031] In one embodiment, the authentication pass coefficient is obtained by calculation, where C is the authentication pass coefficient, X is the correct coefficient of regular questions, i is the number of regular questions, Y is the correct coefficient of skill questions, k is the weight coefficient of skill questions, and n is the number of skill questions.

[0032] In one embodiment, the steps of analyzing whether authentication is passed according to the authentication pass coefficient include:

[0033] If the authentication pass coefficient ≥ 0.8, display that the authentication is passed;

[0034] If the authentication pass coefficient < 0.8, display that the authentication fails and send a warning message to the system management terminal.

[0035] The present invention also provides an identity authentication system based on natural language processing, and the system includes:

[0036] A collection module for collecting user information of enterprise users;

[0037] A construction module for constructing a user portrait according to the user information of enterprise users;

[0038] A generation module for generating a verification question bank according to the user portrait corresponding to the logged-in user account, where the verification question bank includes regular questions, skill questions and corresponding correct answers, and the skill questions include skill questions and corresponding difficulty level self-evaluation options;

[0039] A matching module for adjusting the weight coefficient of skill questions according to the difficulty level feedback by the client, and the value of the weight coefficient is between 0 and 1;

[0040] An analysis module for collecting the response information feedback by the client to obtain the correct coefficients of each group of regular questions and the correct coefficients of each group of skill questions;

[0041] An operation module for generating an authentication pass coefficient according to the correct coefficients of each group of regular questions, the correct coefficients of each group of skill questions and the proportion weight of skill questions;

[0042] A judgment module for analyzing whether authentication is passed according to the authentication pass coefficient.

[0043] The above identity authentication system and method based on natural language processing distinguish the difficulty levels of regular questions and skill questions, and adaptively adjust the weights of skill questions according to the difficulty ratings of the skill questions by users, so as to make the authentication pass coefficient calculated based on the correct coefficients of each group of questions more reasonable, effectively improving the accuracy of user identity authentication on the enterprise management platform. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0045] Figure 1 It is a flowchart of the identity authentication method based on natural language processing proposed by the present invention;

[0046] Figure 2 It is a schematic structural diagram of the identity authentication system based on natural language processing proposed by the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0047] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are some, but not all, of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts fall within the scope of protection of the present invention.

[0048] In response to the above problems, the following will describe the identity authentication system and method based on natural language processing of the present invention in conjunction with Figure 1 - Figure 2 Describe the identity authentication system and method based on natural language processing of the present invention.

[0049] As Figure 1 shown, in one embodiment, an identity authentication method based on natural language processing, the method includes:

[0050] Step S100, collecting user information of enterprise users.

[0051] Specifically, an online form and a questionnaire can be set on its website or mobile application, allowing users to voluntarily fill in personal information when they first log in to collect the basic information and work resumes of users, and obtaining the behavior data of users according to the behavior log data of users.

[0052] Step S200, constructing a user portrait according to the user information of enterprise users.

[0053] Step S300: Generate a verification question bank according to the user portrait corresponding to the logged-in user account. The verification question bank includes general questions, skill questions, and corresponding correct answers. The skill questions include skill questions and corresponding self-evaluation options for difficulty levels.

[0054] Specifically, the ratio of general questions to skill questions is preferably 3:2.

[0055] Step S400: Adjust the weight coefficient of the skill questions according to the difficulty level feedback from the client. The value of the weight coefficient ranges from 0 to 1.

[0056] Step S500: Collect the answering information feedback from the client to obtain the correct coefficients of each group of general questions and the correct coefficients of each group of skill questions.

[0057] Step S600: Generate a certification pass coefficient based on the correct coefficients of each group of general questions, the correct coefficients of each group of skill questions, and the proportion weight of the skill questions.

[0058] Step S700: Analyze whether to pass the certification based on the certification pass coefficient.

[0059] Specifically, generate a user portrait by collecting the historical user information of the user platform. So that when the user logs in, the corresponding user portrait can be directly retrieved according to the user account. When the user logs in to the enterprise management system through password login or fingerprint recognition, the system retrieves the user portrait according to the user account, generates a verification question bank according to the user portrait information in the ratio of 3:2 for general questions and skill questions. The user answers questions on the mobile phone and scores the skill questions according to the question requirements. The system background collects the user's answering information, compares the user's answering information with the correct answers to obtain the correct coefficients of each group of questions, and determines the weight coefficients of each group of skill questions according to the difficulty level feedback. Finally, calculate and generate a certification pass coefficient based on the correct coefficients of each group of questions and the weight coefficients of the skill questions, and determine whether it is the user's own login based on the certification pass coefficient.

[0060] In one embodiment, the user information includes the user's basic information, work resume, and behavior data.

[0061] In one embodiment, the steps of constructing a user portrait according to the user information of enterprise users include:

[0062] Step S210: Use natural language processing to parse the user's work resume and extract key features. The key features include years of work experience, position, number of projects, project types, and skill tags.

[0063] Step S220: Integrate the key features with the basic information and behavior data to obtain a complete user portrait.

[0064] Specifically, preprocess the personal information filled in by users themselves, and extract entities in the text (such as names of people, department names, position names, and project names). After extracting the entities, further analyze the relationships between the entities, extract all the position names of the user from the resume, arrange them in chronological order, and calculate the total working years based on the user's earliest and latest working times. At the same time, extract the number of projects according to the project names, and determine the corresponding project types and skill tags based on the keywords in the text.

[0065] Finally, integrate the above-mentioned key features extracted with the user's basic information (such as name, age, and education) and behavioral data (such as behavioral log data) to further enrich the user profile, and store the integrated data in a structured manner to form a complete user profile.

[0066] In this embodiment, the steps of generating a verification question bank according to the user profile corresponding to the logged-in user account include:

[0067] Step S310, generate general questions using a language generation model according to the basic information, working years, positions, number of projects, and project types.

[0068] Step S320, retrieve the correct answers from the key information and relevant relationships according to the general questions.

[0069] Step S330, generate relevant distractors according to the correct answers.

[0070] Specifically, first, collect the user's basic information (such as name, age, gender), working years, positions, number of projects participated, project types, as well as the corresponding general question set related thereto and the correct questions and distractors corresponding to the general question set, and divide the collected data above into a training set and a verification set according to a ratio of seven to three.

[0071] Use the user's basic information (such as name, age, gender), working years, positions, number of projects participated, project types, and the corresponding general question set in the training set to train a text generation model, use the general question set in the training set and the correct answers corresponding to the general question set to train an answer generation model, use the correct questions and distractors in the training set to train a distractor generation model, and finally use the verification set to verify and optimize the trained language generation model.

[0072] In the process of generating general questions, first retrieve the corresponding user profile according to the logged-in user account, then generate corresponding general questions based on the basic information, working years, positions, number of projects, and project types in the user profile information, then retrieve the correct answers from the user profile information according to the general questions, and finally generate corresponding distractors according to the correct answers.

[0073] In this embodiment, the step of generating a verification question bank according to the user portrait corresponding to the logged-in user account further includes:

[0074] Step S340, determining the type of technical knowledge question bank to be retrieved according to the skill tags.

[0075] Specifically, establish a mapping relationship between each skill tag and the type of technical knowledge question bank to form a mapping table, so as to retrieve and determine the corresponding type of technical knowledge question bank according to the skill tags.

[0076] Step S350, retrieving the corresponding skill question bank from the external question bank resources according to the type of technical knowledge question bank.

[0077] Specifically, according to the system requirements and budget, select appropriate external question bank resources, such as public question bank websites and question bank materials provided by educational institutions, as external question bank resources, and establish an interface for the system to interact with the external question bank resources.

[0078] Step S360, matching a difficulty level self-evaluation option for each skill question.

[0079] Specifically, provide a difficulty level self-evaluation option for each skill question on the user interface, allowing the user to select the difficulty level of the question according to their own understanding and experience.

[0080] In one embodiment, the step of adjusting the weight coefficient of the skill question according to the difficulty level feedback by the client includes:

[0081] Step S410, according to the rule that the lower the difficulty level, the higher the weight coefficient, set the weight coefficient according to the preset difficulty level.

[0082] Step S420, determining the corresponding weight coefficient according to the difficulty level corresponding to the skill question.

[0083] For example, if the difficulty level of the skill question is set to 5 groups of levels, there are also 5 groups of weight coefficients, which are 0.2, 0.4, 0.6, 0.8, and 1 in sequence. The highest difficulty level corresponds to a weight coefficient of 0.2, and the lowest difficulty level corresponds to a weight coefficient of 1.

[0084] In this embodiment, the step of comparing the answer information with the correct answer to obtain the correct coefficient of each group of regular questions and the correct coefficient of each group of skill questions includes:

[0085] Compare the answer information with the correct answer belonging to the same question. If the answer information is consistent with the correct answer, the correct coefficient takes the value of 1. If the answer information is inconsistent with the correct answer, the correct coefficient takes the value of 0.

[0086] In this embodiment, the authentication pass coefficient passes through It is calculated that, where C is the authentication passing coefficient, X is the correct coefficient of regular questions, i is the number of regular questions, Y is the correct coefficient of skill questions, k is the weight coefficient of skill questions, and n is the number of skill questions.

[0087] Specifically, for skill questions, the correct rate of the questions may be affected due to the too high difficulty level. Therefore, it is considered to reduce the weight of the questions with too high difficulty level to obtain the authentication coefficient more reasonably. Regular questions are common sense questions related to the user himself, so the authentication coefficient can be directly obtained according to the correct rate.

[0088] In this implementation, the steps of analyzing whether to pass the authentication according to the authentication passing coefficient include:

[0089] If the authentication passing coefficient ≥ 0.8, it indicates that the authentication is passed.

[0090] If the authentication passing coefficient < 0.8, it indicates that the authentication fails and a warning message is sent to the system management terminal.

[0091] Specifically, by setting the weight coefficient between 0 and 1 and setting the correct coefficient of each group of questions to 1, it is ensured that the authentication passing coefficient is between 0 and 1.

[0092] As Figure 2 shown, in one embodiment, the present invention also provides an identity authentication system based on natural language processing, and the system includes:

[0093] A collection module 10 for collecting user information of enterprise users.

[0094] A construction module 20 for constructing a user portrait according to the user information of enterprise users.

[0095] A generation module 30 for generating a verification question bank according to the user portrait corresponding to the logged-in user account. The verification question bank includes regular questions, skill questions and corresponding correct answers. The skill questions include skill questions and corresponding difficulty level self-evaluation options.

[0096] A matching module 40 for adjusting the weight coefficient of skill questions according to the difficulty level feedback by the client. The value of the weight coefficient is between 0 and 1.

[0097] An analysis module 50 for collecting the answer information feedback by the client to obtain the correct coefficient of each group of regular questions and the correct coefficient of each group of skill questions.

[0098] An operation module 60 for generating an authentication passing coefficient according to the correct coefficient of each group of regular questions, the correct coefficient of each group of skill questions and the proportion weight of skill questions.

[0099] A determination module 70, configured to analyze whether authentication is passed according to an authentication passing coefficient.

[0100] The technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope recorded in this specification.

[0101] The above-described embodiments merely represent several implementation manners of the present invention, and the description thereof is relatively specific and detailed. However, it should not be construed as a limitation to the scope of the present invention. It should be noted that for those of ordinary skill in the art, without departing from the concept of the present invention, several modifications and improvements can still be made, and these all belong to the protection scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the appended claims.

Claims

1. An identity authentication method based on natural language processing, characterized in that: The method comprises: Collect user information of enterprise users; Build user portraits based on the user information of enterprise users; Generate a verification question bank based on the user portrait corresponding to the logged-in user account, the verification question bank includes routine questions, skill questions and corresponding correct answers, and the skill questions include skill questions and corresponding difficulty level self-assessment options; Adjust the weight coefficient of the skill question according to the difficulty level fed back by the client, where the value of the weight coefficient is between 0 and 1; Collect the answer information fed back by the client to obtain the correct coefficients of each group of routine questions and the correct coefficients of each group of skill questions; Generate the certification pass coefficient based on the correct coefficient of each group of routine questions, the correct coefficient of each group of skill questions, and the weight of the skill questions; Analyze whether the certification is passed based on the certification pass coefficient.

2. The identity authentication method based on natural language processing according to claim 1, characterized in that: The user information includes the user's basic information, work history and behavior data.

3. The identity authentication method based on natural language processing according to claim 1, characterized in that: The steps to build a user profile based on the user information of enterprise users include: Use natural language processing to parse the user's work history and extract key features, including length of service, position, number of projects, project type, and skill tags; Integrate key features with basic information and behavioral data to get a complete user portrait.

4. The identity authentication method based on natural language processing according to claim 3 is characterized in that: The steps of generating a verification question bank according to the user profile corresponding to the logged-in user account include: Generate common questions using a language generation model based on basic information, length of service, position, number of projects, and project type; Retrieve the correct answers from key information and related relationships based on routine questions; Generate relevant distractor options based on the correct answer.

5. The identity authentication method based on natural language processing according to claim 4, characterized in that: The step of generating a verification question bank according to the user profile corresponding to the logged-in user account also includes: Determine the type of technical knowledge question bank to be retrieved based on the skill tag; Retrieve the corresponding skill question bank from the external question bank resources according to the technical knowledge question bank type; Match the difficulty level self-assessment option to each skill question.

6. The identity authentication method based on natural language processing according to claim 1, characterized in that: The steps of adjusting the weight coefficient of the skill question according to the difficulty level fed back by the client include: According to the rule that the lower the difficulty level, the higher the weight coefficient, the weight coefficient is set according to the preset difficulty level, and the value of the weight coefficient is between 0 and 1; The corresponding weight coefficient is determined according to the difficulty level of the skill question.

7. The identity authentication method based on natural language processing according to claim 6, characterized in that: The steps of comparing the answer information with the correct answer to obtain the correct coefficient of each group of routine questions and the correct coefficient of each group of skill questions include: The answer information is compared with the correct answer belonging to the same question. If the answer information is consistent with the correct answer, the correct coefficient is 1. If the answer information is inconsistent with the correct answer, the correct coefficient is 0.

8. The identity authentication method based on natural language processing according to claim 7, characterized in that: The certification pass coefficient is Calculated, where C is the certification pass coefficient, X is the correct coefficient of common questions, i is the number of common questions, Y is the correct coefficient of skill questions, k is the weight coefficient of skill questions, and n is the number of skill questions.

9. The identity authentication method based on natural language processing according to claim 8, characterized in that: The steps to analyze whether the certification is passed according to the certification pass coefficient include: If the authentication pass coefficient is ≥ 0.8, the authentication is passed; If the authentication pass coefficient is less than 0.8, the authentication will be displayed as failed and a warning message will be sent to the system management end.

10. An identity authentication system based on natural language processing, applied to an identity authentication method based on natural language processing according to any one of claims 1 to 9, characterized in that: The system comprises: A collection module is used to collect user information of enterprise users; A construction module, used to construct a user profile based on the user information of the enterprise user; A generation module, used to generate a verification question bank according to the user portrait corresponding to the logged-in user account, wherein the verification question bank includes routine questions, skill questions and corresponding correct answers, and the skill questions include skill questions and corresponding difficulty level self-assessment options; A matching module, used to adjust the weight coefficient of the skill question according to the difficulty level fed back by the client, wherein the value of the weight coefficient is between 0 and 1; The analysis module is used to collect the answer information fed back by the client and obtain the correct coefficient of each group of routine questions and the correct coefficient of each group of skill questions; A calculation module is used to generate a certification pass coefficient based on the correct coefficient of each group of routine questions, the correct coefficient of each group of skill questions, and the weight of the skill questions; The judgment module is used to analyze whether the authentication is passed according to the authentication pass coefficient.

Images