WO2023045193A1 - 基于自适应测评的用户能力定级方法、装置、设备及介质 - Google Patents

基于自适应测评的用户能力定级方法、装置、设备及介质 Download PDF

Info

Publication number
WO2023045193A1
WO2023045193A1 PCT/CN2022/072377 CN2022072377W WO2023045193A1 WO 2023045193 A1 WO2023045193 A1 WO 2023045193A1 CN 2022072377 W CN2022072377 W CN 2022072377W WO 2023045193 A1 WO2023045193 A1 WO 2023045193A1
Authority
WO
WIPO (PCT)
Prior art keywords
ability
user
answering
module
question
Prior art date
Application number
PCT/CN2022/072377
Other languages
English (en)
French (fr)
Inventor
舒畅
陈又新
Original Assignee
平安科技（深圳）有限公司
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by 平安科技（深圳）有限公司 filed Critical 平安科技（深圳）有限公司
Publication of WO2023045193A1 publication Critical patent/WO2023045193A1/zh

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G06Q10/06398—Performance of employee with respect to a job function
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q10/00—Administration; Management
    • G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
    • G06Q10/063—Operations research, analysis or management
    • G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
    • G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
    • G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
    • G06Q50/10—Services
    • G06Q50/20—Education
    • G06Q50/205—Education administration or guidance

Definitions

  • the present application relates to the technical field of computer and artificial intelligence, in particular to a method, device, equipment and medium for grading user capabilities based on adaptive evaluation.
  • the traditional ability grading strategy generally pushes multiple test papers to the user first, after the user completes these questions, scores are given according to the user's performance, and finally the user is rated according to the scoring results.
  • the inventor realized that because the traditional grading strategy requires users to do lengthy questions, when users encounter questions that they cannot do, they need to continue to push questions to users until they complete this set of test questions, which will not only hinder the user's learning Enthusiasm will bring troubles to users, reduce user experience, and even cause users to answer randomly and negatively to the following questions, resulting in inaccurate total test scores, so there is a time when users are finally graded based on the total scores. inaccurate defect.
  • the main purpose of this application is to provide a user ability grading method, device, equipment and medium based on adaptive evaluation, aiming to solve the existing grading strategy that requires users to complete lengthy questions and ignores whether users can do difficult questions , it is impossible to accurately grade the user's ability in various subjects, and the ability grading needs to consume a lot of time and energy of the user.
  • the main purpose of this application is to provide a user ability grading method, device, equipment and medium based on adaptive evaluation, aiming to solve the existing grading strategy that requires users to complete lengthy questions and ignores whether users can do difficult questions , it is impossible to accurately grade the user's ability in various subjects, and the ability grading needs to consume a lot of time and energy of the user.
  • the first aspect of the embodiment of the present application proposes a user ability grading method based on adaptive evaluation, the method includes: recording the user's answering situation characteristic information during the evaluation process, the answering situation characteristic information Including the answering modules answered by the user during the evaluation process, and the difficulty coefficient of the questions corresponding to each of the answering modules, the degree of question differentiation and the answer score; the expected posteriori parameter estimation method is used to calculate the ability of the user after completing each question Estimated value, save the ability estimated value to the corresponding historical ability estimated value list; obtain the historical ability estimated value list of any answering module of the user, and calculate whether the ability estimated value in the historical ability estimated value list Reach the convergence requirement, if so, take the latest capability estimate in the list of historical capability estimates as the preliminary grading result of the answering module; obtain the preliminary grading results of all answering modules, and use the weighted average method to calculate the user Corresponding subject test ability grading results.
  • the second aspect of the embodiment of the present application also proposes a user ability grading device based on adaptive evaluation, including a recording module for recording the characteristic information of the user's answering situation during the evaluation process, and the characteristic information of the answering situation includes the user's evaluation
  • a recording module for recording the characteristic information of the user's answering situation during the evaluation process, and the characteristic information of the answering situation includes the user's evaluation
  • the answering modules answered in the process, and the difficulty coefficients of the questions, question distinctions and answering scores corresponding to each of the answering modules
  • the ability estimation module is used to calculate the user's ability to complete the question each time by using the expected posteriori parameter estimation method.
  • the preliminary grading module is used to obtain the historical ability estimated value list of any answering module of the user, and calculate the historical ability estimated value list.
  • the ability estimated value in the ability estimated value list reaches the convergence requirement, if so, then take the latest ability estimated value in the historical ability estimated value list as the preliminary grading result of the answering module; subject ability grading module, use
  • the weighted average method is used to calculate the user's corresponding subject test ability grading results.
  • the third aspect of the embodiments of the present application also proposes a computer device, including a memory and a processor, the memory stores a computer program, and the processor implements the steps of any one of the methods described above when executing the computer program , including: recording the characteristic information of the user’s answering situation during the evaluation process, the characteristic information of the answering situation includes the answering modules answered by the user during the evaluation process, and the difficulty coefficient of the questions corresponding to each of the answering modules, the degree of question differentiation and the answer score ;Use the expected posteriori parameter estimation method to calculate the estimated ability value of the user after each completion of the question, and save the estimated ability value to the corresponding historical ability estimated value list; obtain the historical ability of any answering module of the user estimated value list, and calculate whether the ability estimated value in the historical ability estimated value list meets the convergence requirement, if so, then take the latest ability estimated value in the historical ability estimated value list as the preliminary determination of the answering module Grading results; obtain the preliminary grading results of all answering modules, and use the weighted average method to calculate the user
  • the fourth aspect of the embodiments of the present application also proposes a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, the steps of any one of the methods described above are implemented, including: recording The characteristic information of the user's answering situation during the evaluation process, the characteristic information of the answering situation includes the answering modules answered by the user during the evaluation process, and the difficulty coefficient of the question, the degree of question differentiation and the answering question score corresponding to each of the answering modules; after adopting the expectation
  • the empirical parameter estimation method calculates the estimated ability value of the user after each completion of the question, and saves the estimated ability value to the corresponding historical ability estimated value list; obtains the historical ability estimated value list of any answering module of the user, And calculate whether the capability estimates in the list of historical capability estimates meet the convergence requirement, if so, take the latest capability estimate in the list of historical capability estimates as the preliminary grading result of the answering module; obtain For the preliminary grading results of all answering modules, the weighted average method is used to calculate the user'
  • the self-adaptive evaluation-based user ability grading method, device, equipment, and medium of the present application record the characteristic information of the user's answering situation during the evaluation process, and the characteristic information of the answering situation includes the answering modules answered by the user during the evaluation process, and each The question difficulty coefficient, question distinction and answer score corresponding to the answering module, so as to record the latest answering situation of the user in real time, and provide an accurate and reliable data basis for the subsequent estimation of the subject ability of the user; the expected posteriori parameter estimation method is used to calculate The estimated ability value of the user after completing the question each time, and save the estimated ability value to the corresponding historical ability estimated value list; through the item response theory, a system that simultaneously describes the user's ability level, the characteristics of the evaluation topic, and the user's answer is constructed.
  • the mathematical model of the relationship between them improves the comprehensiveness and accuracy of the calculation model; obtains the historical ability estimated value list of any answering module of the user, and calculates whether the ability estimated value in the historical ability estimated value list reaches Convergence requirement, if so, take the latest ability estimate in the historical ability estimate list as the preliminary grading result of the answering module, and the calculation method for judging whether the user ability value estimate meets the convergence requirement is relatively simple, No complicated iterative steps are required, which improves the calculation efficiency of the user's preliminary ability grading in each answering module; obtains the preliminary grading results of all answering modules, and uses the weighted average method to calculate the corresponding subject test ability grading result of the user. In this way, the grading of subject ability can be completed quickly and accurately, and the number of user answers is small and the user experience is good, as well as the effect of accurate grading results and high evaluation quality.
  • FIG. 1 is a schematic flow diagram of a method for grading user capabilities based on adaptive evaluation according to an embodiment of the present application
  • Fig. 2 is a schematic block diagram of the structure of a user capability grading device based on adaptive evaluation according to an embodiment of the present application
  • FIG. 3 is a schematic block diagram of a computer device according to an embodiment of the present application.
  • AI artificial intelligence
  • digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence, perceive the environment, acquire knowledge and use knowledge to obtain the best results.
  • the embodiment of the present application can be applied to a server.
  • the server can be an independent server, or it can provide cloud services, cloud databases, cloud computing, cloud functions, cloud storage, network services, cloud communications, middleware services, domain name services, security Cloud servers for basic cloud computing services such as cloud computing services, content delivery network (Content Delivery Network, CDN), and big data and artificial intelligence platforms.
  • an embodiment of the present application provides a method for grading user capabilities based on adaptive evaluation in order to achieve the above-mentioned purpose of the invention.
  • the method includes:
  • the characteristic information of the answering situation includes the answering modules answered by the user during the evaluation process, and the difficulty coefficient of the question, the question differentiation degree and the answering question score corresponding to each of the answering modules;
  • the embodiment of the present application records the characteristic information of the user's answering situation during the evaluation process, and the characteristic information of the answering situation includes the answering modules answered by the user during the evaluation process, as well as the difficulty coefficient of the questions corresponding to each of the answering modules, the degree of question differentiation and the answering questions Scores, so as to record the latest answers of the users in real time, and provide an accurate and reliable data basis for the subsequent estimation of the user's subject ability;
  • the expected posterior parameter estimation method is used to calculate the estimated value of the user's ability after each completion of the question, and the obtained
  • the ability estimates are saved to the corresponding list of historical ability estimates; a mathematical model that simultaneously describes the user's ability level, the characteristics of the evaluation questions, and the relationship between the user's answers is constructed through the item response theory, which improves the comprehensiveness and accuracy of the calculation model property; obtain the historical ability estimated value list of any answering module of the user, and calculate whether the ability estimated value in the described historical ability estimated value list meets the convergence requirement, and if so, then take the historical ability estimated value
  • the calculation method for judging whether the estimated value of the user’s ability meets the convergence requirement is relatively simple and does not require complicated iterative steps, which improves the user’s ability in each answering module.
  • the calculation efficiency of preliminary ability grading obtain the preliminary grading results of all answering modules, and use the weighted average method to calculate the user's corresponding subject test ability grading results, so as to complete the subject ability grading quickly and accurately, and realize the number of user answers Less and better user experience, as well as accurate grading results and high evaluation quality.
  • step S1 during the user’s subject evaluation process, the data of the user’s current answer module field, question difficulty coefficient, question differentiation degree slop, and user answer score (1 for correct, 0 for error) are recorded in real time, and the data of the user is recorded in real time
  • the latest answers to the questions provide an accurate and reliable data basis for the subsequent estimation of the user's subject ability.
  • step S2 after the user completes the latest question in any answering module each time, according to the real-time record of the user’s answering situation feature information and the question parameters of the current answering module, the expected posterior parameter estimation method and the item response principle model are used to calculate The user's current ability estimate, calculates the user's ability estimate after each answer in the current answering module, so as to obtain the latest ability estimate, so that the user's latest ability estimate can be obtained by real-time calculation, After updating the latest estimated ability value, it is saved to the list of historical ability estimated values of the answering question module; through the item response theory, a mathematical model that simultaneously describes the user ability level, the characteristics of the evaluation question, and the relationship between the user's answer is constructed to improve The comprehensiveness and accuracy of the calculation model are improved.
  • step S3 obtain the user’s list of historical ability estimates corresponding to any answering module, and judge whether the convergence requirement is met. If it is judged that the convergence requirement is met, stop answering questions, and take the latest ability estimate in the historical ability estimate list as The current preliminary grading result of the answering module; if it is judged that the convergence requirement has not been met, the user is required to continue answering questions and keep the current preliminary grading result. It is relatively simple and does not require complex iterative steps, which greatly improves the calculation efficiency of the user's preliminary ability grading in each answering module, and estimates the user's ability value by recording the data after each answer of the user in real time, and further improves the user's ability. Timeliness and reliability of capability ratings.
  • step S4 repeat the above steps S1-S3 until the user completes the evaluation of all independent answering modules, obtain the preliminary grading results of all answering modules, and finally use the weighted average method to calculate the grading results of the user's corresponding subject test ability.
  • the weights of each independent answer module between two subjects are different.
  • the scores of each answer module are obtained, and all answer modules
  • the scores of the subject are added and divided by the number of independent answering modules to obtain the weighted average result of the subject, which is used as the subject test ability grading result of the user in this subject, so as to complete the subject ability grading quickly and accurately, and realize the user's ability to answer questions
  • the number is small and the user experience is good, as well as the effect of accurate grading results and high evaluation quality.
  • the Item Response Principle (IRT) model calculates the user's correct answer rate for all unanswered questions, and uses the Argmin function to analyze the question that best matches the user's latest ability estimate:
  • the latest estimated capacity value Calculate the probability P i of the examinee's correct answer for each unanswered question in the question bank, and calculate the next question that makes P i closest to 0.5, that is, the question that best matches the examinee's ability value, where a i and b i are the question parameters , respectively represent the topic discrimination degree slop of the i-th question and the difficulty coefficient difficulty of the i-th question.
  • the convergence requirement is met by obtaining the list of historical ability estimates corresponding to any answering module of the user; If the preset threshold is determined to meet the convergence requirement, answering will be stopped, and the last ability value in the historical ability estimation value list will be used as the preliminary grading result of the answering module; if the standard deviation of the estimated values of the last three user ability values If it is less than the preset threshold, it is judged that the convergence requirement has not been met, and the user is required to continue answering questions and keep the previous preliminary grading result.
  • the calculation method of judging whether the estimated value of the user's ability value meets the convergence requirement is relatively simple, without the need for complicated iterative steps, which greatly improves the calculation efficiency of the user's preliminary ability grading in each answering module, and through real-time recording
  • the data after each answer of the user's question is used to estimate the user's ability value, which further improves the timeliness and reliability of the user's ability grading.
  • the calculation of the estimated ability of the user after each completion of a question by using the expected posteriori parameter estimation method includes:
  • the expected posterior parameter estimation method is used to calculate the user's ability estimate after completing each question value.
  • the user calculates the estimated value of the examinee's ability in real time according to the expected a posteriori parameter estimation method (EAP) after completing a question of any answer module. And save the estimated results each time.
  • EAP expected a posteriori parameter estimation method
  • IRT item response principle
  • this embodiment also adopts the two-parameter logistic model algorithm (The 2-parameter Logistic Model) from the perspective of the proposition mode, the scoring mode, and the robustness of the model, and its formula is as follows:
  • a i and b i are the question parameters, which respectively represent the topic discrimination degree slop of the i-th question and the difficulty coefficient of the i-th question, can be regarded as a known parameter.
  • IRT item response theory
  • the discrimination parameter is calculated according to the following formula,
  • rank(p i ) represents the ranking of the correct rate of item i in the sample set from high to low.
  • the difficulty coefficient of each item falls in the interval [1,12], which is consistent with the ability parameter of the subject remain on the same scale.
  • the characteristic information of the user's answering situation in the process of recording the evaluation it also includes:
  • the initial evaluation settings include answer module association settings and question parameter settings.
  • the initial evaluation setting is first performed on each independent answer module in the same subject evaluation test question, and the division of different independent modules can be carried out according to actual business needs. Division, such as grammar, vocabulary, listening and speaking in the English subject assessment questions, each independent answer module.
  • Initial evaluation settings include answer module association settings and question parameter settings. Among them, answer module association settings include association settings for each answer module in the same subject that can be associated with each other according to preset rules, and question parameter settings include The questions in the test questions Distinguish degree, degree of difficulty and scoring rules, etc. are set. By pre-assessing and evaluating the same subject, each independent answering module in the body performs initial evaluation settings, so that the user can accurately obtain the user's ability level when answering questions in each independent answering module.
  • the initialization evaluation settings are performed on several independent answer modules in the subject evaluation test questions, including:
  • the multiple independent answer modules in the subject assessment test questions need to be initialized and set up according to the following rules,
  • Initial A (diff, slop) represents the topic parameter setting of the initial first m questions of the independent answering module A, including its topic difficulty parameter diff and topic discrimination parameter; is the final ability estimate of the independent answering module B, and f(x) is the correlation function relationship between the independent answering module A ability estimate and the independent answering module B ability estimate, obtained from the following historical data training model;
  • N is the total number of users in the training set
  • ⁇ is a preset coefficient
  • the questions will be released by the system settings, and the difficulty parameters of the questions are taken from The item discrimination parameter is set to 1; otherwise, the initial evaluation setting of the independent answering module A is obtained from historical data training, so as to obtain the optimal configuration of the independent answering module A according to the initial evaluation setting rules.
  • the step of taking the question with the correct answer rate closest to the preset threshold as the next question of the answering module it also includes:
  • steps S1-S3 need to be repeated, and the expected posterior parameter estimation method is used to calculate the user's
  • the ability estimated value after the next question of the answering module is updated to the corresponding historical ability estimated value list; after the update is completed, the user’s historical ability estimated value list of the answering module is obtained in real time, and whether the ability estimated value is calculated
  • update the latest ability estimate as the preliminary grading result of the answering module after judging that the convergence requirements are met, update the latest ability estimate as the preliminary grading result of the answering module, and update the user's latest ability estimate in real time to provide the final grading for the subsequent subject test ability of the user Provide better timeliness and higher accuracy.
  • setting the question parameters of the question-to-be-answered module according to the preliminary grading result corresponding to the associated answering module further includes:
  • the weighted grading results corresponding to the associated multiple answer modules are calculated by the weighted average method, and the question parameters of the to-be-answered module are set according to the weighted grading results.
  • the weighted rating corresponding to the multiple answer modules is calculated by the weighted average method.
  • the question parameters of the module to be answered can be set according to the weighted grading results, so as to better and balance the initial evaluation settings for the current independent answer module, and improve the ability to automatically match the test questions that meet the testee's ability level according to the candidate's ability level accuracy and reliability.
  • this application also proposes a user capability grading device based on adaptive evaluation, including:
  • the recording module 100 is used to record the characteristic information of the user's answering situation during the evaluation process, and the characteristic information of the answering situation includes the answering modules answered by the user during the evaluation process, and the difficulty coefficient of the questions corresponding to each of the answering modules, and the degree of differentiation of the questions and answer scores;
  • the ability estimation module 200 is used to calculate the estimated ability value of the user after each completion of the topic by using the expected a posteriori parameter estimation method, and save the estimated ability value to the corresponding historical ability estimated value list;
  • the preliminary grading module 300 is used to obtain a list of historical ability estimates of any answering module of the user, and calculate whether the ability estimates in the list of historical ability estimates meet the convergence requirements, and if so, take the historical The latest ability estimate in the ability estimate list is used as the preliminary grading result of the answering module;
  • the subject ability grading module 400 is used to obtain the preliminary grading results of all answering modules, and use the weighted average method to calculate the user's corresponding subject test ability grading results.
  • the capacity estimation module 200 includes:
  • the answer correct rate calculation unit is used to calculate the answer correct rate corresponding to all unanswered questions in the answer module according to the latest ability estimate based on the item response principle model;
  • the topic selection unit is used to select the topic whose correct answer rate is closest to the preset threshold as the next topic of the answering module.
  • the preliminary grading module 300 includes:
  • Ability determination unit used to calculate the standard deviation of the historical ability estimated value of the most recent preset times of the historical ability estimated value list of any answering module of the user, and judge whether the standard deviation is less than the convergence threshold
  • the grading result acquisition unit is used to determine that the convergence requirement is met, stop answering the question, and take the latest ability estimate in the historical ability estimate list as the preliminary grading result of the answering module;
  • the convergence calculation unit is configured to determine that if not, the convergence requirement has not been met, and the user is required to continue answering questions until the standard deviation is smaller than the convergence threshold.
  • the capacity estimation module 200 includes:
  • the Poisson calculation unit is used to calculate the prior distribution ability value of the user for the subject evaluation test questions by using the Poisson distribution formula
  • the probability calculation unit is used to obtain the question parameters in any answer module in the subject evaluation test questions and the latest ability estimate value of the answer module corresponding to the user, and calculate the user's answer to any question in the answer module by using the item response principle model The probability;
  • Ability estimated value calculation unit used to calculate the user's ability to use the expected posteriori parameter estimation method to calculate the Ability estimates after each item is completed.
  • an initialization module 500 which is used to: perform initial evaluation settings for several independent answer modules in the subject evaluation test questions; wherein, the initial evaluation settings include answer module association settings and question parameter settings.
  • the initialization module 500 includes:
  • An associated module judging unit used for judging whether there is an associated answering module that has completed preliminary grading when initializing evaluation settings for any of the question-to-be-answered modules
  • the first question parameter setting unit is used to set the question parameters of the question-to-be-answered module according to the preliminary grading result corresponding to the associated answering module;
  • the second question parameter setting unit is configured to, if not, set the question parameters of the question-to-be-answered module according to the historical data training results.
  • the capacity estimation module 200 also includes:
  • An estimated value updating unit which is used to calculate the ability estimated value after the next question of the answering module described by the user by using the expected a posteriori parameter estimation method, and update the calculated latest ability estimated value to the corresponding historical ability estimated value list;
  • a real-time calculation unit configured to obtain a list of historical ability estimates of the user's answering module in real time after the update is completed, and calculate whether the ability estimates meet the convergence requirement;
  • the first grading determination unit is configured to update the latest ability estimate among the historical ability estimates as the preliminary grading result of the answering module if yes;
  • the second grading judging unit is used to, if not, require the user to continue answering questions, and the preliminary grading result of the answering module remains unchanged.
  • This embodiment records the characteristic information of the user's answering situation during the evaluation process, and the characteristic information of the answering situation includes the answering modules answered by the user during the evaluation process, as well as the difficulty coefficient of the question, the question differentiation degree and the answering question score corresponding to each of the answering modules , so as to record the user's latest answering situation in real time, and provide an accurate and reliable data basis for the subsequent estimation of the user's subject ability;
  • the expected posterior parameter estimation method is used to calculate the estimated value of the user's ability after each completion of the question, and the The ability estimate is saved to the corresponding historical ability estimate list;
  • a mathematical model that simultaneously describes the user's ability level, the characteristics of the test questions, and the relationship between the user's answer is constructed through the item response theory, which improves the comprehensiveness and accuracy of the calculation model ;
  • Obtain the historical ability estimated value list of any answering module of the user and calculate whether the ability estimated value in the described historical ability estimated value list meets the convergence requirement, if so, then take the latest in the described historical ability estimated value list
  • the calculation efficiency of the level obtain the preliminary grading results of all answering modules, and use the weighted average method to calculate the grading results of the user's corresponding subject test ability, so as to complete the subject ability grading quickly and accurately, and realize that the number of user answers is small and the user Good experience, accurate grading results and high evaluation quality.
  • the computer device may be a server, and its internal structure may be as shown in FIG. 3 .
  • the computer device includes a processor, memory, network interface and database connected by a system bus. Among them, the processor designed by the computer is used to provide calculation and control capabilities.
  • the memory of the computer device includes a non-volatile storage medium and an internal memory.
  • the non-volatile storage medium stores an operating system, computer programs and databases.
  • the memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium.
  • the database of the computer equipment is used to store data such as user ability grading methods based on self-adaptive evaluation.
  • the network interface of the computer device is used to communicate with an external terminal via a network connection.
  • a user ability rating method based on self-adaptive evaluation is realized.
  • the user ability grading method based on adaptive evaluation includes: recording the characteristic information of the user's answering situation during the evaluation process, and the characteristic information of the answering situation includes the answering modules answered by the user during the evaluation process, and each of the answering modules.
  • Corresponding question difficulty coefficient, question discrimination degree and answer score use the expected posterior parameter estimation method to calculate the estimated ability value of the user after completing each question, and save the estimated ability value to the corresponding historical ability estimated value list ;
  • the one-time ability estimate is used as the preliminary grading result of the answering module; the preliminary grading results of all answering modules are obtained, and the weighted average method is used to calculate the corresponding subject test ability grading result of the user.
  • An embodiment of the present application also provides a computer-readable storage medium, on which a computer program is stored.
  • a method for grading user capabilities based on adaptive evaluation is implemented, including the steps of: recording during the evaluation process The characteristic information of the user's answering situation.
  • the characteristic information of the answering situation includes the answering modules answered by the user during the evaluation process, and the difficulty coefficient of the question, the degree of question differentiation and the answering question score corresponding to each of the answering modules; the expected posteriori parameter estimation is adopted method to calculate the estimated ability value of the user after completing each question, and save the estimated ability value to the corresponding historical ability estimated value list; obtain the historical ability estimated value list of any answering module of the user, and calculate the Whether the ability estimates in the historical ability estimated value list meet the convergence requirements, if so, take the latest ability estimated value in the historical ability estimated value list as the preliminary grading result of the answering module; obtain all answering modules Based on the preliminary grading results, the weighted average method is used to calculate the grading results of the user's corresponding subject test ability.
  • this embodiment records the characteristic information of the user's answering situation during the evaluation process, and the characteristic information of the answering situation includes the answering modules answered by the user during the evaluation process, and each of the answers
  • the difficulty coefficient of the questions, the degree of question distinction and the answer scores corresponding to the modules so as to record the latest answers of the users in real time, and provide an accurate and reliable data basis for the subsequent estimation of the user's subject ability
  • the expected posteriori parameter estimation method is used to calculate the user's After each completion of the topic, the estimated ability value is saved to the corresponding historical ability estimated value list; through the item response theory, the relationship between the user ability level, the characteristics of the evaluation question, and the user's answer is constructed.
  • the mathematical model improves the comprehensiveness and accuracy of the calculation model; obtains a list of historical ability estimates of any answering module of the user, and calculates whether the ability estimates in the list of historical ability estimates meet the convergence requirements, If so, take the latest capability estimate in the historical capability estimate list as the preliminary grading result of the answering module, and the calculation method for judging whether the estimated value of user capability meets the convergence requirement is relatively simple and does not need to be complicated
  • the iterative steps improve the calculation efficiency of the user's preliminary ability grading in each answering module; obtain the preliminary grading results of all answering modules, and use the weighted average method to calculate the user's corresponding subject test ability grading results, thus quickly Accurately complete the grading of subject ability, realize the effects of less number of questions answered by users and good user experience, as well as accurate grading results and high evaluation quality.
  • Nonvolatile memory can include read only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory.
  • Volatile memory can include random access memory (RAM) or external cache memory.
  • RAM is available in many forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (SSRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.
  • SRAM Static RAM
  • DRAM Dynamic RAM
  • SDRAM Synchronous DRAM
  • SSRSDRAM Double Data Rate SDRAM
  • ESDRAM Enhanced SDRAM
  • SLDRAM Synchronous Link (Synchlink) DRAM
  • SLDRAM Synchronous Link (Synchlink) DRAM
  • Rambus direct RAM
  • DRAM direct memory bus dynamic RAM
  • RDRAM memory bus dynamic RAM

Landscapes

  • Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • Engineering & Computer Science (AREA)
  • Educational Administration (AREA)
  • Strategic Management (AREA)
  • Economics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Tourism & Hospitality (AREA)
  • Development Economics (AREA)
  • Marketing (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Business, Economics & Management (AREA)
  • Operations Research (AREA)
  • Educational Technology (AREA)
  • Game Theory and Decision Science (AREA)
  • Quality & Reliability (AREA)
  • Primary Health Care (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Electrically Operated Instructional Devices (AREA)

Abstract

一种基于自适应测评的用户能力定级方法、装置、设备及介质，其中方法包括：记录测评过程用户的答题情况特征信息，采用期望后验参数估计法计算所述用户在每次完成题目后的最新能力估计值并保存至历史能力估计值列表；获取所述用户任一答题模块的历史能力估计值列表，计算该历史能力估计值列表的能力估计值是否达到收敛要求，若是，则取所述历史能力估计值列表中最近一次能力估计值作为该答题模块的初步定级结果；获取所有答题模块的初步定级结果，采用加权平均法计算用户科目测试能力定级结果。从而能够快速准确完成用户学科能力定级，实现用户答题数量少，以及定级结果准确且测评质量高的效果。

Description

基于自适应测评的用户能力定级方法、装置、设备及介质
本申请要求于2021年9月27日提交中国专利局、申请号为202111138372.7，发明名称为“基于自适应测评的用户能力定级方法、装置、设备及介质”的中国专利申请的优先权，其全部内容通过引用结合在本申请中。
技术领域
本申请涉及到计算机及人工智能技术领域，特别是涉及到一种基于自适应测评的用户能力定级方法、装置、设备及介质。
背景技术
在线教育需要对每个用户的教育水平进行定级，从而在后续阶段按照已定的级别为用户推荐相应的学习产品。传统的能力定级策略，一般是先给用户推送多道试题的试卷，在用户完成这些题目后，根据用户的做题情况进行打分，最后根据打分的结果对用户进行定级。
然而，发明人意识到，传统定级策略由于需要用户去做冗长的题目，当用户遇到不会做的题目后还需要继续推送题目至用户，直至完成这套试题，不仅会打击用户的学习积极性，给用户带来烦恼，降低用户体验感，甚至会导致用户对后面的题目随意作答和消极作答，造成了测试总分数不准确的情况，从而存在最终根据总分数对用户进行能力定级时不准确的缺陷。
技术问题
本申请的主要目的为提供一种基于自适应测评的用户能力定级方法、装置、设备及介质，旨在解决现有定级策略需要用户完成冗长题目且忽略用户是否会做难度较高的题目，无法对用户各个学科能力进行准确定级，且能力定级需要耗费用户大量时间和精力的问题。
技术解决方案
本申请的主要目的为提供一种基于自适应测评的用户能力定级方法、装置、设备及介质，旨在解决现有定级策略需要用户完成冗长题目且忽略用户是否会做难度较高的题目，无法对用户各个学科能力进行准确定级，且能力定级需要耗费用户大量时间和精力的问题。
为了实现上述发明目的，本申请实施例的第一方面提出一种基于自适应测评的用户能力定级方法，所述方法包括：记录测评过程中用户的答题情况特征信息，所述答题情况特征信息包括用户测评过程中所回答的答题模块，以及各个所述答题模块所对应的题目难度系数、题目区分度和答题分数；采用期望后验参数估计法计算所述用户在每次完成题目后的能力估计值，将所述能力估计值保存至对应的历史能力估计值列表；获取所述用户任一答题模块的历史能力估计值列表，并计算该所述历史能力估计值列表中的能力估计值是否达到收敛要求，若是，则取所述历史能力估计值列表中的最近一次的能力估计值作为该答题模块的初步定级结果；获取所有答题模块的初步定级结果，采用加权平均法计算出用户对应的科目测试能力定级结果。
本申请实施例的第二方面还提出了一种基于自适应测评的用户能力定级装置，包括记录模块，用于记录测评过程中用户的答题情况特征信息，所述答题情况特征信息包括用户测评过程中所回答的答题模块，以及各个所述答题模块所对应的题目难度系数、题目区分度和答题分数；能力估算模块，用于采用期望后验参数估计法计算所述用户在每次完成题目后的能力估计值，将所述能力估计值保存至对应的历史能力估计值列表；初步定级模块，用于获取所述用户任一答题模块的历史能力估计值列表，并计算该所述历史能力估计值列表中的能力估计值是否达到收敛要求，若是，则取所述历史能力估计值列表中的最近一次的能力估计值作为该答题模块的初步定级结果；科目能力定级模块，用于获取所有答题模块的初步定级结果，采用加权平均法计算出用户对应的科目测试能力定级结果。
本申请实施例的第三方面还提出了一种计算机设备，包括存储器和处理器，所述存储器存储有计算机程序，所述处理器执行所述计算机程序时实现上述任一项所述方法的步骤，包括：记录测评过程中用户的答题情况特征信息，所述答题情况特征信息包括用户测评过程中所回答的答题模块，以及各个所述答题模块所对应的题目难度系数、题目区分度和答题分数；采 用期望后验参数估计法计算所述用户在每次完成题目后的能力估计值，将所述能力估计值保存至对应的历史能力估计值列表；获取所述用户任一答题模块的历史能力估计值列表，并计算该所述历史能力估计值列表中的能力估计值是否达到收敛要求，若是，则取所述历史能力估计值列表中的最近一次的能力估计值作为该答题模块的初步定级结果；获取所有答题模块的初步定级结果，采用加权平均法计算出用户对应的科目测试能力定级结果。
本申请实施例的第四方面还提出了一种计算机可读存储介质，其上存储有计算机程序，所述计算机程序被处理器执行时实现上述任一项所述的方法的步骤，包括：记录测评过程中用户的答题情况特征信息，所述答题情况特征信息包括用户测评过程中所回答的答题模块，以及各个所述答题模块所对应的题目难度系数、题目区分度和答题分数；采用期望后验参数估计法计算所述用户在每次完成题目后的能力估计值，将所述能力估计值保存至对应的历史能力估计值列表；获取所述用户任一答题模块的历史能力估计值列表，并计算该所述历史能力估计值列表中的能力估计值是否达到收敛要求，若是，则取所述历史能力估计值列表中的最近一次的能力估计值作为该答题模块的初步定级结果；获取所有答题模块的初步定级结果，采用加权平均法计算出用户对应的科目测试能力定级结果。
有益效果
本申请的基于自适应测评的用户能力定级方法、装置、设备及介质，记录测评过程中用户的答题情况特征信息，所述答题情况特征信息包括用户测评过程中所回答的答题模块，以及各个所述答题模块所对应的题目难度系数、题目区分度和答题分数，从而实时记录用户的最新答题情况，为后续对用户进行学科能力估计提供准确可靠的数据基础；采用期望后验参数估计法计算所述用户在每次完成题目后的能力估计值，将所述能力估计值保存至对应的历史能力估计值列表；通过项目反应理论构建了同时描述用户能力水平、测评题目特性、与用户作答之间的关系的数学模型，提高了计算模型的全面性和准确性；获取所述用户任一答题模块的历史能力估计值列表，并计算该所述历史能力估计值列表中的能力估计值是否达到收敛要求，若是，则取所述历史能力估计值列表中的最近一次的能力估计值作为该答题模块的初步定级结果，对用户能力值的估计值判断是否达到收敛要求的计算方式较为简单，不需要复杂的迭代步骤，提高了对用户在各个答题模块的初步能力定级的计算效率；获取所有答题模块的初步定级结果，采用加权平均法计算出用户对应的科目测试能力定级结果，从而又快又准地完成学科能力定级，实现用户答题数量少且用户体验佳，以及定级结果准且测评质量高的效果。
附图说明
图1为本申请一实施例的基于自适应测评的用户能力定级方法的流程示意图；
图2为本申请一实施例的基于自适应测评的用户能力定级装置的结构示意框图；
图3为本申请一实施例的计算机设备的结构示意框图。
本申请目的的实现、功能特点及优点将结合实施例，参照附图做进一步说明。
本发明的最佳实施方式
为了使本申请的目的、技术方案及优点更加清楚明白，以下结合附图及实施例，对本申请进行进一步详细说明。应当理解，此处描述的具体实施例仅仅用以解释本申请，并不用于限定本申请。
本申请实施例可以基于人工智能技术对相关的数据进行获取和处理。其中，人工智能(Artificial Intelligence，AI)是利用数字计算机或者数字计算机控制的机器模拟、延伸和扩展人的智能，感知环境、获取知识并使用知识获得最佳结果的理论、方法、技术及应用系统。
本申请实施例可以应用于服务器中，服务器可以是独立的服务器，也可以是提供云服务、云数据库、云计算、云函数、云存储、网络服务、云通信、中间件服务、域名服务、安全服务、内容分发网络(Content Delivery Network，CDN)、以及大数据和人工智能平台等基础 云计算服务的云服务器。
参照图1，本申请实施例中提供一种为了实现上述发明目的，本申请提出一种基于自适应测评的用户能力定级方法，所述方法包括：
S1、记录测评过程中用户的答题情况特征信息，所述答题情况特征信息包括用户测评过程中所回答的答题模块，以及各个所述答题模块所对应的题目难度系数、题目区分度和答题分数；
S2、采用期望后验参数估计法计算所述用户在每次完成题目后的能力估计值，将所述能力估计值保存至对应的历史能力估计值列表；
S3、获取所述用户任一答题模块的历史能力估计值列表，并计算该所述历史能力估计值列表中的能力估计值是否达到收敛要求，若是，则取所述历史能力估计值列表中的最近一次的能力估计值作为该答题模块的初步定级结果；
S4、获取所有答题模块的初步定级结果，采用加权平均法计算出用户对应的科目测试能力定级结果。
随着在线教育的普及，在线教育在为用户推荐相应水平的学习课程或产品之前，都需要对每个用户的教育水平进行定级。现有算法大多基于经典测验理论，但其底层数学模型相对简单和缺乏科学性，如真实分数与观测分数间存在线性关系的假定不符合事实。同时由于传统定级策略需要给用户推送多道试题的试卷，用户全部做完试题才能根据用户的做题情况进行打分和能力定级，这些统一固定的试卷都忽略了用户实际的不同的能力水平，导致有些用户不能很好地完成这些冗长的试题，因此，无法正确评估用户真实水平或答题数量过多影响用户体验，现有算法过于简单及死板的出题流程也会造成分数不准确和后续定级不准确的问题。
本申请实施例记录测评过程中用户的答题情况特征信息，所述答题情况特征信息包括用户测评过程中所回答的答题模块，以及各个所述答题模块所对应的题目难度系数、题目区分度和答题分数，从而实时记录用户的最新答题情况，为后续对用户进行学科能力估计提供准确可靠的数据基础；采用期望后验参数估计法计算所述用户在每次完成题目后的能力估计值，将所述能力估计值保存至对应的历史能力估计值列表；通过项目反应理论构建了同时描述用户能力水平、测评题目特性、与用户作答之间的关系的数学模型，提高了计算模型的全面性和准确性；获取所述用户任一答题模块的历史能力估计值列表，并计算该所述历史能力估计值列表中的能力估计值是否达到收敛要求，若是，则取所述历史能力估计值列表中的最近一次的能力估计值作为该答题模块的初步定级结果，对用户能力值的估计值判断是否达到收敛要求的计算方式较为简单，不需要复杂的迭代步骤，提高了对用户在各个答题模块的初步能力定级的计算效率；获取所有答题模块的初步定级结果，采用加权平均法计算出用户对应的科目测试能力定级结果，从而又快又准地完成学科能力定级，实现用户答题数量少且用户体验佳，以及定级结果准且测评质量高的效果。
对于步骤S1，在用户进行学科测评过程中，实时记录用户当前的答题模块field、题目难度系数difficulty、题目区分度slop及用户答题情况score(正确为1，错误为0)的数据，实时记录用户的最新答题情况，为后续对用户进行学科能力估计提供准确可靠的数据基础。
对于步骤S2，在用户每次完成任一答题模块内的最新一道题目后，根据实时记录用户的答题情况特征信息和当前答题模块的题目参数，采用期望后验参数估计法和项目反应原理模型计算用户当前的能力估计值，计算出用户在当前答题模块中每次答完一道题后的能力估计值，从而得到最新一次的能力估计值，从而能够实时计算获取得到用户的最新一次能力估计值，在更新最新一次的能力估计值后，保存至对应答题模块的历史能力估计值列表；通过项目反应理论构建了同时描述用户能力水平、测评题目特性、与用户作答之间的关系的数学 模型，提高了计算模型的全面性和准确性。
对于步骤S3，获取用户在任一答题模块对应的历史能力估计值列表，判断是否达到收敛要求，若判断为达到收敛要求，则停止答题，取历史能力估计值列表中的最新一次的能力估计值作为该答题模块当前的初步定级结果；若判断为未达到收敛要求，则需要用户继续答题，保持当前的初步定级结果，本实施例中对用户能力值的估计值判断是否达到收敛要求的方式较为简单，而不需要复杂的迭代步骤，大大提高了对用户在各个答题模块的初步能力定级的计算效率，且通过实时记录用户每次答题后的数据进行用户能力值的估计，进一步提高用户能力定级的时效性和可靠性。
对于步骤S4，重复上述步骤S1～S3，直至用户完成所有独立答题模块的测评，获取所有答题模块的初步定级结果，最后采用加权平均法计算出用户对应的科目测试能力定级结果，由于每个学科之间的各个独立答题模块所占的权重是不同的，因此，通过对每个独立答题模块的初步定级结果乘以所占的权重值，得到各个答题模块的分数，将所有答题模块的分数相加再除以独立答题模块的个数，从而得到该学科的加权平均结果，作为用户该学科的科目测试能力定级结果，从而又快又准地完成学科能力定级，实现用户答题数量少且用户体验佳，以及定级结果准且测评质量高的效果。
在一种优选的实施例中，在所述采用期望后验参数估计法计算所述用户在每次完成题目后的能力估计值，将所述能力估计值保存至对应的历史能力估计值列表步骤之后，还包括：
基于项目反应原理模型根据所述最新一次的能力估计值，分别计算该答题模块中所有未作答题目对应的回答正确率；取回答正确率最接近预设阈值的题目作为该答题模块的下一道题目。
在具体的实施例中，在取用户最新能力估计值作为某一答题模块的初步定级结果之后，根据最新一次的估计能力值，并结合命题方式、计分方式、模型的鲁棒性，基于项目反应原理(IRT)模型计算出用户关于所有未作答题目对应的回答正确率，并采用Argmin函数解析最匹配用户最新一次的能力估计值的题目：
Figure PCTCN2022072377-appb-000001
根据最新估计能力值
Figure PCTCN2022072377-appb-000002
计算每一道题库中未作答题目考生回答正确的概率P i，下一题题目则推使得P i最接近0.5的题，即最匹配考生能力值的题，其中，a i，b i为题目参数，分别表示第i题的题目区分度slop及第i题的难度系数difficulty，通过根据考生的能力水平自动匹配符合其能力水平的侧题，实现千人千面和优化用户答题体验的效果。
在一种优选的实施例中，所述获取所述用户任一答题模块的历史能力估计值列表，并计算该所述历史能力估计值列表中的能力估计值是否达到收敛要求，若是，则取所述历史能力估计值列表中的最近一次的能力估计值作为该答题模块的初步定级结果之中，包括：
计算所述用户任一答题模块的历史能力估计值列表的最近预设次数的历史能力估计值的标准差，判断所述标准差是否小于收敛阈值；
若是，则判定为达到收敛要求，停止答题，取所述历史能力估计值列表中的最近一次的能力估计值作为该答题模块的初步定级结果；
若否，则判定为未达到收敛要求，需要用户继续答题，直至所述标准差小于收敛阈值。
在具体的实例中，通过获取用户在任一答题模块对应的历史能力估计值列表，判断是否达到收敛要求；若近三次(该次数可根据实际情况灵活调整)用户能力值的估计值的标准差小于预设阈值，则判定为达到收敛要求，则停止答题，并将历史能力估计值列表中最后一次能力值作为该答题模块的初步定级结果；若近三次用户能力值的估计值的标准差不小于预设阈值，则判断为未达到收敛要求，需要用户继续答题，并保持上一次的初步定级结果。对过 判断用户能力值的估计值是否达到收敛要求的这种计算方式较为简单，而无需复杂的迭代步骤，大大提高了对用户在各个答题模块的初步能力定级的计算效率，且通过实时记录用户每次答题后的数据进行用户能力值的估计，进一步提高用户能力定级的时效性和可靠性。
在一种优选的实施例中，所述采用期望后验参数估计法计算所述用户在每次完成题目后的能力估计值，包括：
采用泊松分布公式计算出用户对科目测评试题的先验分布能力值；
获取科目测评试题中任一答题模块中的题目参数和用户对应的该答题模块的最近一次的能力估计值，采用项目反应原理模型计算出用户答对该答题模块中任一题目的概率；
根据用户已完成试题数量、先验分布能力值、题目参数和当前用户答对该答题模块中任一题目的概率，采用期待后验参数估计法计算出所述用户在每次完成题目后的能力估计值。
在具体的实施例中，用户在每次完成任一答题模块的一道题目后，实时根据期望后验参数估计法(EAP)计算考生能力的估计值
Figure PCTCN2022072377-appb-000003
并保存每次的估计结果。
Figure PCTCN2022072377-appb-000004
其中，L表示考生已完成的试题数量；φ(θ)表示能力值θ的先验分布，根据学科能力分布特性(能力等级低的人群偏多)，假设其符合泊松分布
Figure PCTCN2022072377-appb-000005
式中参数λ可根据用户历史数据获取进行拟合；P i表示考生答对题目i的可能性，其基于项目反应原理(IRT)模型计算。
需要说明的是，本实施例还从命题方式、计分方式、模型的鲁棒性出发，采用二参数模型算法(The 2-parameter Logistic Model)，其公式如下：
Figure PCTCN2022072377-appb-000006
其中，
Figure PCTCN2022072377-appb-000007
表示最近一次的能力估计值，即根据考生L-1道作答记录的能力估计值；a i，b i为题目参数，分别表示第i题的题目区分度slop及第i题的难度系数difficulty，可看作已知参数。通过项目反应理论(IRT)构建了同时描述考生能力水平、测评题目特性、与考生作答之间的关系的数学模型，提高了计算模型的全面性和准确性。
其中，根据以下公式计算求出区分度参数，
Figure PCTCN2022072377-appb-000008
式中
Figure PCTCN2022072377-appb-000009
为样本集中答对题目i的被试测评的该模块平均等级，
Figure PCTCN2022072377-appb-000010
为答错题目i的被试测评的该模块平均等级，s t为全体被试测评的该模块等级标准差，p为答对该题目被试占总被试人数比率，q＝1-p。
根据以下公式计算求出难度系数参数，
Figure PCTCN2022072377-appb-000011
式中rank(p i)表示题目i在样本集中正确率由高至低的排序，通过min-max标准化方 式，使得各题难度系数落在[1,12]区间内，与被试的能力参数保持在同一量表。
在一种优选的实施例中，在所述记录测评过程中用户的答题情况特征信息步骤之前，还包括：
分别对科目测评试题内的若干个独立答题模块进行初始化测评设置；其中，所述初始化测评设置包括答题模块关联设置和题目参数设置。
在具体的实施例中，在用户进入试题分别完成多个局部独立模块的测评前，首先对同一科目测评试题内的各个独立答题模块进行初始化测评设置，不同独立模块的划分可根据实际业务需求进行划分，如英语科目测评试题内的语法、词汇、听力和口语等各个独立答题模块。初始化测评设置包括答题模块关联设置和题目参数设置，其中，答题模块关联设置包括根据预设规则对同一学科中的各个能够两两关联的答题模块进行关联设置，题目参数设置包括对试题内的题目区分度、难度系数和分值规则等等进行设置。通过预先对同一科目测评使体内的各个独立答题模块进行初始化测评设置，能够使得用户在各个独立答题模块进行答题时能够准确获取用户的能力水平。
在一种优选的实施例中，所述分别对科目测评试题内的若干个独立答题模块进行初始化测评设置，包括：
在对任一待答题模块进行初始化测评设置时，判断是否存在关联且已完成初步定级的答题模块；
若是，则根据所关联的答题模块对应的初步定级结果设置该待答题模块的题目参数；
若否，则根据历史数据训练结果设置该待答题模块的题目参数。
在具体的实施例中，在用户进行学科测评前，需要根据以下规则对该科目测评试题内的多个独立答题模块分别进行初始化测评设置，
Figure PCTCN2022072377-appb-000012
其中，Initial A(diff，slop)表示独立答题模块A的初始前m道题目的题目参数设置，包含其题目难度参数diff及题目区分度参数；
Figure PCTCN2022072377-appb-000013
为独立答题模块B的最终能力估计值，f(x)为独立答题模块A能力估计值与独立答题模块B能力估计值的相关函数关系，由以下历史数据训练建模得出；
Figure PCTCN2022072377-appb-000014
式中，N为训练集用户总数目，Δi表示用户i的评级偏差Δi＝User_i predict-User_i true，α为预设系数。
具体的，在对独立答题模块A进行初始化测评设置前，若独立答题模块A存在独立答题模块模块B，且独立答题模块B的定级结果已知，则独立答题模块A的初始前m＝1题将由系统设置推出，其题目的难度参数取
Figure PCTCN2022072377-appb-000015
题目区分度参数取1；否则，独立答题模块A的初始测评设置由历史数据训练得出，从而根据初始测评设置规则求得独立答题模块A的最优配置。
在一种优选的实施例中，在所述取回答正确率最接近预设阈值的题目作为该答题模块的 下一道题目步骤之后，还包括：
采用期望后验参数估计法计算用户所述该答题模块的下一道题目后的能力估计值，将计算后得到的最新能力估计值更新至对应的历史能力估计值列表；
在完成更新后，实时获取用户该答题模块的历史能力估计值列表，计算所述能力估计值是否达到收敛要求；
若是，则更新所述历史能力估计值中的最新能力估计值作为该答题模块的初步定级结果；
若否，则需要用户继续答题，该答题模块的初步定级结果不变。
在具体的实施例中，在每次取回答正确率最接近预设阈值的题目作为用户在该答题模块的下一道题目之后，都需要重复步骤S1～S3，采用期望后验参数估计法计算用户所述该答题模块的下一道题目后的能力估计值，更新至对应的历史能力估计值列表；在完成更新后，实时获取用户该答题模块的历史能力估计值列表，计算所述能力估计值是否达到收敛要求后，在判断达到收敛要求后，则更新最新能力估计值作为该答题模块的初步定级结果，通过实时更新用户的最新能力估计值，为后续对用户进行学科测试能力的最终定级提供更好的时效性和更高的准确性。
在一种优选的实施例中，所述根据所关联的答题模块对应的初步定级结果设置该待答题模块的题目参数，还包括：
当所关联的答题模块的数量为一个时，则根据该关联的答题模块对应的初步定级结果设置该待答题模块的题目参数；
当所关联的答题模块的数量多于一个时，则通过加权平均法计算出所关联的多个答题模块对应的加权定级结果，根据加权定级结果设置该待答题模块的题目参数。
在具体的实施例中，在根据所关联的答题模块对应的初步定级结果设置该待答题模块的题目参数时，若存在当前独立答题模块关联的答题模块为一个时，则仅仅需要根据该关联的答题模块对应的初步定级结果设置该待答题模块的题目参数；若存在当前独立答题模块关联的答题模块多于一个时，则通过加权平均法计算出所关联的多个答题模块对应的加权定级结果，根据加权定级结果设置该待答题模块的题目参数，从而更好地和平衡地对当前独立答题模块进行初始化测评设置，提高根据考生的能力水平自动匹配符合其能力水平的测试题目的准确性和可靠性。
参照图2，本申请还提出了一种基于自适应测评的用户能力定级装置，包括：
记录模块100，用于记录测评过程中用户的答题情况特征信息，所述答题情况特征信息包括用户测评过程中所回答的答题模块，以及各个所述答题模块所对应的题目难度系数、题目区分度和答题分数；
能力估算模块200，用于采用期望后验参数估计法计算所述用户在每次完成题目后的能力估计值，将所述能力估计值保存至对应的历史能力估计值列表；
初步定级模块300，用于获取所述用户任一答题模块的历史能力估计值列表，并计算该所述历史能力估计值列表中的能力估计值是否达到收敛要求，若是，则取所述历史能力估计值列表中的最近一次的能力估计值作为该答题模块的初步定级结果；
科目能力定级模块400，用于获取所有答题模块的初步定级结果，采用加权平均法计算出用户对应的科目测试能力定级结果。
进一步地，能力估算模块200，包括：
回答正确率计算单元，用于基于项目反应原理模型根据所述最新一次的能力估计值，分别计算该答题模块中所有未作答题目对应的回答正确率；
题目选取单元，用于取回答正确率最接近预设阈值的题目作为该答题模块的下一道题目。
进一步地，初步定级模块300，包括：
能力判定单元，用于计算所述用户任一答题模块的历史能力估计值列表的最近预设次数 的历史能力估计值的标准差，判断所述标准差是否小于收敛阈值；
定级结果获取单元，用于若是，则判定为达到收敛要求，停止答题，取所述历史能力估计值列表中的最近一次的能力估计值作为该答题模块的初步定级结果；
收敛计算单元，用于若否，则判定为未达到收敛要求，需要用户继续答题，直至所述标准差小于收敛阈值。
进一步地，能力估算模块200，包括：
泊松计算单元，用于采用泊松分布公式计算出用户对科目测评试题的先验分布能力值；
概率计算单元，用于获取科目测评试题中任一答题模块中的题目参数和用户对应的该答题模块的最近一次的能力估计值，采用项目反应原理模型计算出用户答对该答题模块中任一题目的概率；
能力估计值计算单元，用于根据用户已完成试题数量、先验分布能力值、题目参数和当前用户答对该答题模块中任一题目的概率，采用期待后验参数估计法计算出所述用户在每次完成题目后的能力估计值。
进一步地，还包括初始化模块500，用于：分别对科目测评试题内的若干个独立答题模块进行初始化测评设置；其中，所述初始化测评设置包括答题模块关联设置和题目参数设置。
进一步地，初始化模块500，包括：
关联模块判定单元，用于在对任一待答题模块进行初始化测评设置时，判断是否存在关联且已完成初步定级的答题模块；
第一题目参数设置单元，用于若是，则根据所关联的答题模块对应的初步定级结果设置该待答题模块的题目参数；
第二题目参数设置单元，用于若否，则根据历史数据训练结果设置该待答题模块的题目参数。
进一步地，能力估算模块200，还包括：
估计值更新单元，用于采用期望后验参数估计法计算用户所述该答题模块的下一道题目后的能力估计值，将计算后得到的最新能力估计值更新至对应的历史能力估计值列表；
实时计算单元，用于在完成更新后，实时获取用户该答题模块的历史能力估计值列表，计算所述能力估计值是否达到收敛要求；
第一定级判定单元，用于若是，则更新所述历史能力估计值中的最新能力估计值作为该答题模块的初步定级结果；
第二定级判定单元，用于若否，则需要用户继续答题，该答题模块的初步定级结果不变。
本实施例记录测评过程中用户的答题情况特征信息，所述答题情况特征信息包括用户测评过程中所回答的答题模块，以及各个所述答题模块所对应的题目难度系数、题目区分度和答题分数，从而实时记录用户的最新答题情况，为后续对用户进行学科能力估计提供准确可靠的数据基础；采用期望后验参数估计法计算所述用户在每次完成题目后的能力估计值，将所述能力估计值保存至对应的历史能力估计值列表；通过项目反应理论构建了同时描述用户能力水平、测评题目特性、与用户作答之间的关系的数学模型，提高了计算模型的全面性和准确性；获取所述用户任一答题模块的历史能力估计值列表，并计算该所述历史能力估计值列表中的能力估计值是否达到收敛要求，若是，则取所述历史能力估计值列表中的最近一次的能力估计值作为该答题模块的初步定级结果，用户能力值的估计值判断是否达到收敛要求计算方式较为简单，不需要复杂的迭代步骤，提高了对用户在各个答题模块的初步能力定级的计算效率；获取所有答题模块的初步定级结果，采用加权平均法计算出用户对应的科目测试能力定级结果，从而又快又准地完成学科能力定级，实现用户答题数量少且用户体验佳，以及定级结果准且测评质量高的效果。
参照图3，本申请实施例中还提供一种计算机设备，该计算机设备可以是服务器，其内 部结构可以如图3所示。该计算机设备包括通过系统总线连接的处理器、存储器、网络接口和数据库。其中，该计算机设计的处理器用于提供计算和控制能力。该计算机设备的存储器包括非易失性存储介质、内存储器。该非易失性存储介质存储有操作系统、计算机程序和数据库。该内存器为非易失性存储介质中的操作系统和计算机程序的运行提供环境。该计算机设备的数据库用于储存基于自适应测评的用户能力定级方法等数据。该计算机设备的网络接口用于与外部的终端通过网络连接通信。该计算机程序被处理器执行时以实现一种基于自适应测评的用户能力定级方法。所述基于自适应测评的用户能力定级方法，包括：记录测评过程中用户的答题情况特征信息，所述答题情况特征信息包括用户测评过程中所回答的答题模块，以及各个所述答题模块所对应的题目难度系数、题目区分度和答题分数；采用期望后验参数估计法计算所述用户在每次完成题目后的能力估计值，将所述能力估计值保存至对应的历史能力估计值列表；获取所述用户任一答题模块的历史能力估计值列表，并计算该所述历史能力估计值列表中的能力估计值是否达到收敛要求，若是，则取所述历史能力估计值列表中的最近一次的能力估计值作为该答题模块的初步定级结果；获取所有答题模块的初步定级结果，采用加权平均法计算出用户对应的科目测试能力定级结果。
本申请一实施例还提供一种计算机可读存储介质，其上存储有计算机程序，计算机程序被处理器执行时实现一种基于自适应测评的用户能力定级方法，包括步骤：记录测评过程中用户的答题情况特征信息，所述答题情况特征信息包括用户测评过程中所回答的答题模块，以及各个所述答题模块所对应的题目难度系数、题目区分度和答题分数；采用期望后验参数估计法计算所述用户在每次完成题目后的能力估计值，将所述能力估计值保存至对应的历史能力估计值列表；获取所述用户任一答题模块的历史能力估计值列表，并计算该所述历史能力估计值列表中的能力估计值是否达到收敛要求，若是，则取所述历史能力估计值列表中的最近一次的能力估计值作为该答题模块的初步定级结果；获取所有答题模块的初步定级结果，采用加权平均法计算出用户对应的科目测试能力定级结果。
上述执行的基于自适应测评的用户能力定级方法，本实施例记录测评过程中用户的答题情况特征信息，所述答题情况特征信息包括用户测评过程中所回答的答题模块，以及各个所述答题模块所对应的题目难度系数、题目区分度和答题分数，从而实时记录用户的最新答题情况，为后续对用户进行学科能力估计提供准确可靠的数据基础；采用期望后验参数估计法计算所述用户在每次完成题目后的能力估计值，将所述能力估计值保存至对应的历史能力估计值列表；通过项目反应理论构建了同时描述用户能力水平、测评题目特性、与用户作答之间的关系的数学模型，提高了计算模型的全面性和准确性；获取所述用户任一答题模块的历史能力估计值列表，并计算该所述历史能力估计值列表中的能力估计值是否达到收敛要求，若是，则取所述历史能力估计值列表中的最近一次的能力估计值作为该答题模块的初步定级结果，对用户能力值的估计值判断是否达到收敛要求的计算方式较为简单，不需要复杂的迭代步骤，提高了对用户在各个答题模块的初步能力定级的计算效率；获取所有答题模块的初步定级结果，采用加权平均法计算出用户对应的科目测试能力定级结果，从而又快又准地完成学科能力定级，实现用户答题数量少且用户体验佳，以及定级结果准且测评质量高的效果。
本领域普通技术人员可以理解实现上述实施例方法中的全部或部分流程，是可以通过计算机程序来指令相关的硬件来完成，所述的计算机程序可存储于一非易失性计算机可读取存储介质中，该计算机程序在执行时，可包括如上述各方法的实施例的流程。其中，本申请所提供的和实施例中所使用的对存储器、存储、数据库或其它介质的任何引用，均可包括非易失性和/或易失性存储器。非易失性存储器可以包括只读存储器(ROM)、可编程ROM(PROM)、电可编程ROM(EPROM)、电可擦除可编程ROM(EEPROM)或闪存。易失性存储器可包括随机存取存储器(RAM)或者外部高速缓冲存储器。作为说明而非局限，RAM以多种形式可得，诸如静态RAM(SRAM)、动态RAM(DRAM)、同步DRAM(SDRAM)、双速据率SDRAM(SSRSDRAM)、 增强型SDRAM(ESDRAM)、同步链路(Synchlink)DRAM(SLDRAM)、存储器总线(Rambus)直接RAM(RDRAM)、直接存储器总线动态RAM(DRDRAM)、以及存储器总线动态RAM(RDRAM)等。
需要说明的是，在本文中，术语“包括”、“包含”或者其任何其他变体意在涵盖非排他性的包含，从而使得包括一系列要素的过程、装置、物品或者方法不仅包括那些要素，而且还包括没有明确列出的其他要素，或者是还包括为这种过程、装置、物品或者方法所固有的要素。在没有更多限制的情况下，由语句“包括一个……”限定的要素，并不排除在包括该要素的过程、装置、物品或者方法中还存在另外的相同要素。
以上所述仅为本申请的优选实施例，并非因此限制本申请的专利范围，凡是利用本申请说明书及附图内容所作的等效结构或等效流程变换，或直接或间接运用在其他相关的技术领域，均同理包括在本申请的专利保护范围内。

Claims (22)

  1. 一种基于自适应测评的用户能力定级方法，其中，所述方法包括：
    记录测评过程中用户的答题情况特征信息，所述答题情况特征信息包括用户测评过程中所回答的答题模块，以及各个所述答题模块所对应的题目难度系数、题目区分度和答题分数；
    采用期望后验参数估计法计算所述用户在每次完成题目后的能力估计值，将所述能力估计值保存至对应的历史能力估计值列表；
    获取所述用户任一答题模块的历史能力估计值列表，并计算该所述历史能力估计值列表中的能力估计值是否达到收敛要求，若是，则取所述历史能力估计值列表中的最近一次的能力估计值作为该答题模块的初步定级结果；
    获取所有答题模块的初步定级结果，采用加权平均法计算出用户对应的科目测试能力定级结果。
  2. 根据权利要求1所述的基于自适应测评的用户能力定级方法，其中，在所述采用期望后验参数估计法计算所述用户在每次完成题目后的能力估计值，将所述能力估计值保存至对应的历史能力估计值列表步骤之后，还包括：
    基于项目反应原理模型根据所述最新一次的能力估计值，分别计算该答题模块中所有未作答题目对应的回答正确率；
    取回答正确率最接近预设阈值的题目作为该答题模块的下一道题目。
  3. 根据权利要求1所述的基于自适应测评的用户能力定级方法，其中，所述获取所述用户任一答题模块的历史能力估计值列表，并计算该所述历史能力估计值列表中的能力估计值是否达到收敛要求，若是，则取所述历史能力估计值列表中的最近一次的能力估计值作为该答题模块的初步定级结果步骤之中，包括：
    计算所述用户任一答题模块的历史能力估计值列表的最近预设次数的历史能力估计值的标准差，判断所述标准差是否小于收敛阈值；
    若是，则判定为达到收敛要求，停止答题，取所述历史能力估计值列表中的最近一次的能力估计值作为该答题模块的初步定级结果；
    若否，则判定为未达到收敛要求，需要用户继续答题，直至所述标准差小于收敛阈值。
  4. 根据权利要求1所述的基于自适应测评的用户能力定级方法，其中，所述采用期望后验参数估计法计算所述用户在每次完成题目后的能力估计值，包括：
    采用泊松分布公式计算出用户对科目测评试题的先验分布能力值；
    获取科目测评试题中任一答题模块中的题目参数和用户对应的该答题模块的最近一次的能力估计值，采用项目反应原理模型计算出用户答对该答题模块中任一题目的概率；
    根据用户已完成试题数量、先验分布能力值、题目参数和当前用户答对该答题模块中任一题目的概率，采用期待后验参数估计法计算出所述用户在每次完成题目后的能力估计值。
  5. 根据权利要求1所述的基于自适应测评的用户能力定级方法，其中，在所述记录测评过程中用户的答题情况特征信息步骤之前，还包括：
    分别对科目测评试题内的若干个独立答题模块进行初始化测评设置；其中，所述初始化测评设置包括答题模块关联设置和题目参数设置。
  6. 根据权利要求5所述的基于自适应测评的用户能力定级方法，其中，所述分别对科目测评试题内的若干个独立答题模块进行初始化测评设置，包括：
    在对任一待答题模块进行初始化测评设置时，判断是否存在关联且已完成初步定级的答题模块；
    若是，则根据所关联的答题模块对应的初步定级结果设置该待答题模块的题目参数；
    若否，则根据历史数据训练结果设置该待答题模块的题目参数。
  7. 根据权利要求2所述的基于自适应测评的用户能力定级方法，其中，在所述取回答正确率最接近预设阈值的题目作为该答题模块的下一道题目步骤之后，还包括：
    采用期望后验参数估计法计算用户所述该答题模块的下一道题目后的能力估计值，将计算后得到的最新能力估计值更新至对应的历史能力估计值列表；
    在完成更新后，实时获取用户该答题模块的历史能力估计值列表，计算所述能力估计值是否达到收敛要求；
    若是，则更新所述历史能力估计值中的最新能力估计值作为该答题模块的初步定级结果；
    若否，则需要用户继续答题，该答题模块的初步定级结果不变。
  8. 一种基于自适应测评的用户能力定级装置，其中，包括：
    记录模块，用于记录测评过程中用户的答题情况特征信息，所述答题情况特征信息包括用户测评过程中所回答的答题模块，以及各个所述答题模块所对应的题目难度系数、题目区分度和答题分数；
    能力估算模块，用于采用期望后验参数估计法计算所述用户在每次完成题目后的能力估计值，将所述能力估计值保存至对应的历史能力估计值列表；
    初步定级模块，用于获取所述用户任一答题模块的历史能力估计值列表，并计算该所述历史能力估计值列表中的能力估计值是否达到收敛要求，若是，则取所述历史能力估计值列表中的最近一次的能力估计值作为该答题模块的初步定级结果；
    科目能力定级模块，用于获取所有答题模块的初步定级结果，采用加权平均法计算出用户对应的科目测试能力定级结果。
  9. 一种计算机设备，包括存储器和处理器，所述存储器存储有计算机程序，其中，所述处理器执行所述计算机程序时实现一种基于自适应测评的用户能力定级方法的步骤；
    其中，所述基于自适应测评的用户能力定级方法包括：
    记录测评过程中用户的答题情况特征信息，所述答题情况特征信息包括用户测评过程中所回答的答题模块，以及各个所述答题模块所对应的题目难度系数、题目区分度和答题分数；
    采用期望后验参数估计法计算所述用户在每次完成题目后的能力估计值，将所述能力估计值保存至对应的历史能力估计值列表；
    获取所述用户任一答题模块的历史能力估计值列表，并计算该所述历史能力估计值列表中的能力估计值是否达到收敛要求，若是，则取所述历史能力估计值列表中的最近一次的能力估计值作为该答题模块的初步定级结果；
    获取所有答题模块的初步定级结果，采用加权平均法计算出用户对应的科目测试能力定级结果。
  10. 根据权利要求9所述的计算机设备，其中，在所述采用期望后验参数估计法计算所述用户在每次完成题目后的能力估计值，将所述能力估计值保存至对应的历史能力估计值列表步骤之后，还包括：
    基于项目反应原理模型根据所述最新一次的能力估计值，分别计算该答题模块中所有未作答题目对应的回答正确率；
    取回答正确率最接近预设阈值的题目作为该答题模块的下一道题目。
  11. 根据权利要求9所述的计算机设备，其中，所述获取所述用户任一答题模块的历史能力估计值列表，并计算该所述历史能力估计值列表中的能力估计值是否达到收敛要求，若是，则取所述历史能力估计值列表中的最近一次的能力估计值作为该答题模块的初步定级结果步骤之中，包括：
    计算所述用户任一答题模块的历史能力估计值列表的最近预设次数的历史能力估计值的标准差，判断所述标准差是否小于收敛阈值；
    若是，则判定为达到收敛要求，停止答题，取所述历史能力估计值列表中的最近一次的能力估计值作为该答题模块的初步定级结果；
    若否，则判定为未达到收敛要求，需要用户继续答题，直至所述标准差小于收敛阈值。
  12. 根据权利要求9所述的计算机设备，其中，所述采用期望后验参数估计法计算所述用户在每次完成题目后的能力估计值，包括：
    采用泊松分布公式计算出用户对科目测评试题的先验分布能力值；
    获取科目测评试题中任一答题模块中的题目参数和用户对应的该答题模块的最近一次的能力估计值，采用项目反应原理模型计算出用户答对该答题模块中任一题目的概率；
    根据用户已完成试题数量、先验分布能力值、题目参数和当前用户答对该答题模块中任一题目的概率，采用期待后验参数估计法计算出所述用户在每次完成题目后的能力估计值。
  13. 根据权利要求9所述的计算机设备，其中，在所述记录测评过程中用户的答题情况特征信息步骤之前，还包括：
    分别对科目测评试题内的若干个独立答题模块进行初始化测评设置；其中，所述初始化测评设置包括答题模块关联设置和题目参数设置。
  14. 根据权利要求13所述的计算机设备，其中，所述分别对科目测评试题内的若干个独立答题模块进行初始化测评设置，包括：
    在对任一待答题模块进行初始化测评设置时，判断是否存在关联且已完成初步定级的答题模块；
    若是，则根据所关联的答题模块对应的初步定级结果设置该待答题模块的题目参数；
    若否，则根据历史数据训练结果设置该待答题模块的题目参数。
  15. 根据权利要求10所述的计算机设备，其中，在所述取回答正确率最接近预设阈值的题目作为该答题模块的下一道题目步骤之后，还包括：
    采用期望后验参数估计法计算用户所述该答题模块的下一道题目后的能力估计值，将计算后得到的最新能力估计值更新至对应的历史能力估计值列表；
    在完成更新后，实时获取用户该答题模块的历史能力估计值列表，计算所述能力估计值是否达到收敛要求；
    若是，则更新所述历史能力估计值中的最新能力估计值作为该答题模块的初步定级结果；
    若否，则需要用户继续答题，该答题模块的初步定级结果不变。
  16. 一种计算机可读存储介质，其上存储有计算机程序，其中，所述计算机程序被处理器执行时实现一种基于自适应测评的用户能力定级方法的步骤；
    其中，基于自适应测评的用户能力定级方法包括：
    记录测评过程中用户的答题情况特征信息，所述答题情况特征信息包括用户测评过程中所回答的答题模块，以及各个所述答题模块所对应的题目难度系数、题目区分度和答题分数；
    采用期望后验参数估计法计算所述用户在每次完成题目后的能力估计值，将所述能力估计值保存至对应的历史能力估计值列表；
    获取所述用户任一答题模块的历史能力估计值列表，并计算该所述历史能力估计值列表中的能力估计值是否达到收敛要求，若是，则取所述历史能力估计值列表中的最近一次的能力估计值作为该答题模块的初步定级结果；
    获取所有答题模块的初步定级结果，采用加权平均法计算出用户对应的科目测试能力定级结果。
  17. 根据权利要求16所述的计算机可读存储介质，其中，在所述采用期望后验参数估计法计算所述用户在每次完成题目后的能力估计值，将所述能力估计值保存至对应的历史能力估计值列表步骤之后，还包括：
    基于项目反应原理模型根据所述最新一次的能力估计值，分别计算该答题模块中所有未作答题目对应的回答正确率；
    取回答正确率最接近预设阈值的题目作为该答题模块的下一道题目。
  18. 根据权利要求16所述的计算机可读存储介质，其中，所述获取所述用户任一答题模块的历史能力估计值列表，并计算该所述历史能力估计值列表中的能力估计值是否达到收敛要求，若是，则取所述历史能力估计值列表中的最近一次的能力估计值作为该答题模块的初步定级结果步骤之中，包括：
    计算所述用户任一答题模块的历史能力估计值列表的最近预设次数的历史能力估计值的标准差，判断所述标准差是否小于收敛阈值；
    若是，则判定为达到收敛要求，停止答题，取所述历史能力估计值列表中的最近一次的能力估计值作为该答题模块的初步定级结果；
    若否，则判定为未达到收敛要求，需要用户继续答题，直至所述标准差小于收敛阈值。
  19. 根据权利要求16所述的计算机可读存储介质，其中，所述采用期望后验参数估计法计算所述用户在每次完成题目后的能力估计值，包括：
    采用泊松分布公式计算出用户对科目测评试题的先验分布能力值；
    获取科目测评试题中任一答题模块中的题目参数和用户对应的该答题模块的最近一次的能力估计值，采用项目反应原理模型计算出用户答对该答题模块中任一题目的概率；
    根据用户已完成试题数量、先验分布能力值、题目参数和当前用户答对该答题模块中任一题目的概率，采用期待后验参数估计法计算出所述用户在每次完成题目后的能力估计值。
  20. 根据权利要求16所述的计算机可读存储介质，其中，在所述记录测评过程中用户的答题情况特征信息步骤之前，还包括：
    分别对科目测评试题内的若干个独立答题模块进行初始化测评设置；其中，所述初始化测评设置包括答题模块关联设置和题目参数设置。
  21. 根据权利要求20所述的计算机可读存储介质，其中，所述分别对科目测评试题内的若干个独立答题模块进行初始化测评设置，包括：
    在对任一待答题模块进行初始化测评设置时，判断是否存在关联且已完成初步定级的答题模块；
    若是，则根据所关联的答题模块对应的初步定级结果设置该待答题模块的题目参数；
    若否，则根据历史数据训练结果设置该待答题模块的题目参数。
  22. 根据权利要求17所述的计算机可读存储介质，其中，在所述取回答正确率最接近预设阈值的题目作为该答题模块的下一道题目步骤之后，还包括：
    采用期望后验参数估计法计算用户所述该答题模块的下一道题目后的能力估计值，将计算后得到的最新能力估计值更新至对应的历史能力估计值列表；
    在完成更新后，实时获取用户该答题模块的历史能力估计值列表，计算所述能力估计值是否达到收敛要求；
    若是，则更新所述历史能力估计值中的最新能力估计值作为该答题模块的初步定级结果；
    若否，则需要用户继续答题，该答题模块的初步定级结果不变。
PCT/CN2022/072377 2021-09-27 2022-01-17 基于自适应测评的用户能力定级方法、装置、设备及介质 WO2023045193A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202111138372.7A CN113850512A (zh) 2021-09-27 2021-09-27 基于自适应测评的用户能力定级方法、装置、设备及介质
CN202111138372.7 2021-09-27

Publications (1)

Publication Number Publication Date
WO2023045193A1 true WO2023045193A1 (zh) 2023-03-30

Family

ID=78980191

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/072377 WO2023045193A1 (zh) 2021-09-27 2022-01-17 基于自适应测评的用户能力定级方法、装置、设备及介质

Country Status (2)

Country Link
CN (1) CN113850512A (zh)
WO (1) WO2023045193A1 (zh)

Families Citing this family (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113850512A (zh) * 2021-09-27 2021-12-28 平安科技（深圳）有限公司 基于自适应测评的用户能力定级方法、装置、设备及介质
CN114519534A (zh) * 2022-02-23 2022-05-20 平安科技（深圳）有限公司 能力等级分析方法、装置、电子设备及存储介质

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2008242637A (ja) * 2007-03-26 2008-10-09 Kddi Corp 能力推定システムおよび方法ならびにプログラムおよび記録媒体
CN110378818A (zh) * 2019-07-22 2019-10-25 广西大学 基于难度的个性化习题推荐方法、系统及介质
CN111311055A (zh) * 2019-12-26 2020-06-19 成都终身成长科技有限公司 一种用户能力的评估方法、装置及存储介质
CN111507632A (zh) * 2020-04-21 2020-08-07 北京一起教育信息咨询有限责任公司 一种发展性评测的方法、装置及电子设备
CN111626420A (zh) * 2020-03-31 2020-09-04 杨阳 一种认知能力的自适应测评方法、测评系统及其使用方法
CN113850512A (zh) * 2021-09-27 2021-12-28 平安科技（深圳）有限公司 基于自适应测评的用户能力定级方法、装置、设备及介质

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2008242637A (ja) * 2007-03-26 2008-10-09 Kddi Corp 能力推定システムおよび方法ならびにプログラムおよび記録媒体
CN110378818A (zh) * 2019-07-22 2019-10-25 广西大学 基于难度的个性化习题推荐方法、系统及介质
CN111311055A (zh) * 2019-12-26 2020-06-19 成都终身成长科技有限公司 一种用户能力的评估方法、装置及存储介质
CN111626420A (zh) * 2020-03-31 2020-09-04 杨阳 一种认知能力的自适应测评方法、测评系统及其使用方法
CN111507632A (zh) * 2020-04-21 2020-08-07 北京一起教育信息咨询有限责任公司 一种发展性评测的方法、装置及电子设备
CN113850512A (zh) * 2021-09-27 2021-12-28 平安科技（深圳）有限公司 基于自适应测评的用户能力定级方法、装置、设备及介质

Also Published As

Publication number Publication date
CN113850512A (zh) 2021-12-28

Similar Documents

Publication Publication Date Title
CN109857835B (zh) 一种基于认知诊断理论的自适应网络安全知识测评方法
WO2023045193A1 (zh) 基于自适应测评的用户能力定级方法、装置、设备及介质
CN108229718B (zh) 一种信息预测方法及装置
US11030919B2 (en) Measuring language learning using standardized score scales and adaptive assessment engines
CN110147428B (zh) 知识点掌握度的更新方法、装置、设备和存储介质
CN111651676B (zh) 基于能力模型进行职业推荐的方法、装置、设备和介质
CN107357763B (zh) 一种基于自步学习的众包分类数据质量控制方法
CN113409174A (zh) 知识点测评方法和装置
CN111651677B (zh) 课程内容推荐方法、装置、计算机设备和存储介质
CN111680216B (zh) 一种试题推荐方法、系统、介质及设备
KR20130037541A (ko) 큐벡터를 이용한 문항출제 시스템 및 방법
WO2020135642A1 (zh) 一种基于生成对抗网络的模型训练方法及设备
JP2020047234A (ja) データ評価方法、装置、機器及び読み取り可能な記憶媒体
US20120058459A1 (en) Democratic Process of Testing for Cognitively Demanding Skills and Experiences
US20160086086A1 (en) Multi-media content-recommender system that learns how to elicit user preferences
WO2020206172A1 (en) Confidence evaluation to measure trust in behavioral health survey results
Dehejia Was there a Riverside miracle? A hierarchical framework for evaluating programs with grouped data
CN112446809B (zh) 心理健康综合自适应测评方法及系统
CN110489602B (zh) 知识点能力值预估方法、系统、设备及介质
CN109800880B (zh) 基于动态学习风格信息的自适应学习特征提取系统及应用
US20230351153A1 (en) Knowledge graph reasoning model, system, and reasoning method based on bayesian few-shot learning
CN111259305A (zh) 一种在线智能教育方法
US10854099B2 (en) Adaptive learning machine for score improvement and parts thereof
CN116127029A (zh) 基于能力画像的内容配置方法、装置、电子设备及存储介质
CN114973815A (zh) 一种用于在线教学对学生进行测试的系统、方法

Legal Events

Date Code Title Description
121 Ep: the epo has been informed by wipo that ep was designated in this application

Ref document number: 22871261

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE