WO2023116306A1 - 信息处理方法、装置、可读存储介质及电子设备 - Google Patents

信息处理方法、装置、可读存储介质及电子设备 Download PDF

Info

Publication number
WO2023116306A1
WO2023116306A1 PCT/CN2022/133207 CN2022133207W WO2023116306A1 WO 2023116306 A1 WO2023116306 A1 WO 2023116306A1 CN 2022133207 W CN2022133207 W CN 2022133207W WO 2023116306 A1 WO2023116306 A1 WO 2023116306A1
Authority
WO
WIPO (PCT)
Prior art keywords
estimated
parameter
current
answerer
answer information
Prior art date
Legal status (The legal status 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 status listed.)
Ceased
Application number
PCT/CN2022/133207
Other languages
English (en)
French (fr)
Inventor
彭璐瑶
韦程志
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Beijing Youzhuju Network Technology Co Ltd
Original Assignee
Beijing Youzhuju Network Technology Co Ltd
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 Beijing Youzhuju Network Technology Co Ltd filed Critical Beijing Youzhuju Network Technology Co Ltd
Publication of WO2023116306A1 publication Critical patent/WO2023116306A1/zh
Anticipated expiration legal-status Critical
Ceased legal-status Critical Current

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N7/00—Computing arrangements based on specific mathematical models
    • G06N7/01—Probabilistic graphical models, e.g. probabilistic networks
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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
    • G—PHYSICS
    • G06—COMPUTING OR CALCULATING; 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
    • Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
    • Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
    • Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
    • Y02D10/00—Energy efficient computing, e.g. low power processors, power management or thermal management

Definitions

  • the present disclosure relates to the technical field of information processing, and in particular, to an information processing method, device, readable storage medium, and electronic equipment.
  • the present disclosure provides an information processing method, including:
  • the respondent's initial learning ability in the item response theory model the difficulty of the topic and the performance factor analysis model, the respondent's successful learning effect and failure learning effect on each knowledge point involved in the target question bank are Parameters to be estimated, construct a Bayesian network model;
  • an information processing device including:
  • An acquisition module configured to acquire the first answer information of a preset number of answerers on the target question bank in the previous period and the second answer information of the current answerer on the target question bank in the current period, wherein the current answerer is one of the preset number of respondents;
  • the failure learning effect is a parameter to be estimated, and a Bayesian network model is constructed;
  • a first determination module configured to determine an estimated value of each parameter to be estimated based on the Bayesian network model obtained by the construction module and the first answer information obtained by the acquisition module;
  • the second determination module is configured to determine the current question answerer's current question mark according to the estimated value of each parameter to be estimated determined by the first determination module and the second answer information obtained by the acquisition module. ability.
  • the present disclosure provides a computer-readable medium on which a computer program is stored, and when the program is executed by a processing device, the steps of the method provided in the first aspect of the present disclosure are implemented.
  • an electronic device including:
  • a processing device configured to execute the computer program in the storage device to implement the steps of the method provided in the first aspect of the present disclosure.
  • the respondent's initial learning ability, topic difficulty and performance factors in the item response theory model are used to analyze the respondent's successful learning effect and failure learning effect on each knowledge point involved in the target question bank.
  • the effect is the parameter to be estimated, and the Bayesian network model is constructed; then, based on the constructed Bayesian network model and the preset number of respondents’ first answer information about the target question bank in the previous period, the value of each parameter to be estimated is determined. Estimated value; finally, according to the estimated value of each parameter to be estimated and the current answerer's second answer information about the target question bank in the current period, determine the current ability of the current answerer.
  • Fig. 1 is a flowchart showing an information processing method according to an exemplary embodiment.
  • Fig. 2 is a flowchart showing an information processing method according to another exemplary embodiment.
  • Fig. 3 is a flowchart showing an information processing method according to another exemplary embodiment.
  • Fig. 4 is a block diagram of an information processing device according to an exemplary embodiment.
  • Fig. 5 is a block diagram of an electronic device according to an exemplary embodiment.
  • the term “comprise” and its variations are open-ended, ie “including but not limited to”.
  • the term “based on” is “based at least in part on”.
  • the term “one embodiment” means “at least one embodiment”; the term “another embodiment” means “at least one further embodiment”; the term “some embodiments” means “at least some embodiments.” Relevant definitions of other terms will be given in the description below.
  • Fig. 1 is a flowchart showing an information processing method according to an exemplary embodiment. As shown in FIG. 1 , the above method includes S101 to S104.
  • the current answerer is one of the preset number of answerers.
  • the first answer information and the second answer information can include information such as answer questions and answer situations (such as answering correctly or incorrectly).
  • the unit of day, week, quarter, etc. may be used as a time period.
  • the target question bank and the preset number of answerers can be selected according to actual needs. For example, it is possible to obtain the information on the first answer to the registered accounting question bank by accounting practitioners in city A last week, and the first answer information on subject four of the driver's license examiners in area B in the previous quarter.
  • the answerer is any one of the preset number of answerers mentioned above, and the learning effect represents the change speed of the answerer's ability in the process of accumulating learning times of the answerer.
  • the estimated value of each parameter to be estimated is determined based on the Bayesian network model and the first answer information.
  • the current ability of the current answerer is determined according to the estimated value of each parameter to be estimated and the second answer information.
  • the above S102 may be executed after the above S101 (as shown in FIG. 1 ), may also be executed simultaneously with the above S101, or may be executed before the above S101, which is not specifically limited in the present disclosure.
  • the respondent's initial learning ability, topic difficulty and performance factors in the item response theory model are used to analyze the respondent's successful learning effect and failure learning effect on each knowledge point involved in the target question bank.
  • the effect is the parameter to be estimated, and the Bayesian network model is constructed; then, based on the constructed Bayesian network model and the preset number of respondents’ first answer information about the target question bank in the previous period, the value of each parameter to be estimated is determined. Estimated value; finally, according to the estimated value of each parameter to be estimated and the current answerer's second answer information about the target question bank in the current period, determine the current ability of the current answerer.
  • Bayesian network model is shown in the following equation (1):
  • T is the transpose.
  • K is the number of knowledge points involved in the target question bank.
  • It is equivalent to a binary vector, that is, the position that contains the knowledge point is 1, and the position that does not contain it is 0.
  • the target question bank involves 5 knowledge points, namely knowledge point 1, knowledge point 2, knowledge point 3, knowledge point 4 and knowledge point 5, wherein the jth question involves knowledge point 1, knowledge point 3 and knowledge point 4 , then the distribution vector of the knowledge points involved in the jth topic
  • the more knowledge points involved in the question the higher the probability that the answerer will answer the question.
  • the probability is directly related to the number of knowledge points involved in the question, resulting in low accuracy. Therefore, in another implementation manner, normalization processing is performed on the above-mentioned dichotomous vectors, and the distribution vector obtained after the normalization processing is determined as Wherein, the sum of each element in the distribution vector obtained after the normalization processing is equal to 1. In this way, the probability will not increase due to the increase in the number of knowledge points involved in the topic, so that the accuracy of probability prediction can be improved.
  • the target question bank involves 5 knowledge points, namely knowledge point 1, knowledge point 2, knowledge point 3, knowledge point 4 and knowledge point 5, wherein the jth question involves knowledge point 1, knowledge point 3 and knowledge point 4 , then the distribution vector of the knowledge points involved in the jth topic
  • the above distribution vectors can also be adjusted according to the importance of different knowledge points. Specifically, the element values at positions containing important knowledge points are relatively larger, and the element values at positions containing non-important knowledge points are relatively small. Thus reflecting the primary and secondary relationship of knowledge points.
  • a Markov Chain Monte Carlo (MCMC) parameter estimation model can be used to determine the estimated value of each parameter to be estimated.
  • an expectation propagation (Expectation Propagation, EP) parameter estimation model can be used to determine the estimated value of each parameter to be estimated.
  • a variational inference method may be used to determine the estimated value of each parameter to be estimated.
  • the variational inference method is used to calculate the variational lower bound.
  • parameter estimation may be performed with the maximization of the variational lower bound as the objective function, and an estimated value of each parameter to be estimated may be obtained by a stochastic gradient descent method. Since the specific implementation manner of using the stochastic gradient descent method to obtain the estimated value of each parameter to be estimated is well known to those skilled in the art, the present disclosure will not repeat it here.
  • the prior distribution of the parameter to be estimated in the current period is determined, which can make the variational inference method suitable for dynamic and continuous parameter estimation scenarios, And the amount of calculation and memory usage are better than the MCMC and EP parameter estimation methods mentioned above, and the estimated value of each parameter to be estimated can be estimated relatively accurately and quickly.
  • the approximate posterior distribution of the parameter to be estimated in the previous period can be determined as the prior distribution of the parameter to be estimated in the current period.
  • the prior distribution of the parameter to be estimated in the current period can be determined by the following equation (2) according to the approximate posterior distribution of the parameter to be estimated in the previous period:
  • p(parameter) is the prior distribution of the parameter to be estimated in the current period
  • q m (parameter) is the approximate posterior distribution of the parameter to be estimated in the previous period
  • decay is the weight coefficient
  • the prior distribution of the parameter to be estimated in the current period is not directly replaced by the approximate posterior distribution of the parameter to be estimated in the previous period, but the weighted average method is used to determine the parameter to be estimated
  • the prior distribution in the current period so that the approximate posterior distribution of the parameters to be estimated in the past period can gradually affect the update of the approximate posterior distribution of the current period and the future period, so that the initial stage (that is, the first few periods ) due to the lack of answer information and the influence of unstable posterior distribution estimation, thereby improving the accuracy of parameter estimation.
  • step (2) based on the Bayesian network model, the first answer information, and the prior distribution of each parameter to be estimated in the current period, the variational inference method is used to calculate the specific implementation of the variational lower bound. illustrate.
  • a variational inference method can be used based on the Bayesian network model, the first answer information, and the prior distribution of each parameter to be estimated in the current period. , to calculate the variational lower bound by the following equation (3):
  • ELBO is the variational lower bound; is the vector formed by the initial ability of each respondent among the preset number of respondents; is a vector composed of the difficulty of each question in the target question bank; is a vector composed of each respondent's successful learning effect on each knowledge point involved in the target question bank; is a vector composed of each respondent’s failure learning effect on each knowledge point involved in the target question bank; likelihood is a reconstruction likelihood function based on the variational posterior distribution, according to the Bayesian network model and each parameter to be estimated Sure; for right expectations; for The KL divergence of the approximate posterior distribution of and its prior distribution; for The KL divergence of the approximate posterior distribution of and its prior distribution; for The KL divergence of the approximate posterior distribution of and its prior distribution; for The KL divergence of the approximate posterior distribution of and its prior distribution.
  • the variational inference method can be used to calculate the variable by the following equation (4): Sub-boundary:
  • shrink and enhance are hyperparameters; g is equal to the product of the preset number and the number of questions corresponding to the first answer information; max is equal to the product of the total number of answerers on the target question bank and the number of questions contained in the target question bank.
  • the initial learning ability and difficulty of the questions are mainly estimated.
  • the initial learning ability and difficulty of the questions are basically fixed. It can make the variational inference method more focused on estimating the two parameters to be estimated, the successful learning effect and the failure learning effect of each knowledge point involved in the target question bank, so as to improve the accuracy of parameter estimation.
  • the KL divergence corresponding to the learned parameters to be estimated is enhanced, namely and
  • the shrink of each time period is a preset value.
  • the prior distribution that is, the prior distribution that retains the future period contains the learned historical information, thereby improving the accuracy of parameter estimation.
  • Second answer information determine the number of successful learning and the number of learning failures of the current answerer under each knowledge point involved in the topic corresponding to the second answer information.
  • the number of successful learning is the number of correct answers
  • the number of learning failures is the number of wrong answers.
  • the number of learning successes and the number of learning failures determine the current ability of the current respondent.
  • Fig. 2 is a flowchart showing an information processing method according to another exemplary embodiment. As shown in FIG. 2 , the above method further includes S105.
  • Bayesian network model can be simplified as:
  • the probability of the current answerer correctly answering the candidate questions is obtained. Since the current ability of the answerer and the difficulty of the candidate questions can be accurately evaluated, the accuracy of the probability prediction can be guaranteed.
  • the questions can be automatically pushed to the answerers according to the above probability.
  • the above method further includes S106.
  • the preset condition may be that the above probability is within a preset probability range, for example, 0.5-0.9.
  • the greater the entropy value of a candidate question the more information the answerer can obtain by practicing the question. Therefore, when H is greater than the preset threshold, the candidate question is pushed to the current answerer.
  • Fig. 4 is a block diagram of an information processing device according to an exemplary embodiment. As shown in Figure 4, the device 400 includes:
  • the acquisition module 401 is used to acquire the first answer information of the preset number of answerers on the target question bank in the previous period and the second answer information of the current answerer on the target question bank in the current period, wherein the current answerer is one of the preset number of respondents;
  • the building block 402 is used to analyze the respondent's successful learning of each knowledge point involved in the target question bank with the respondent's initial learning ability, topic difficulty and performance factors in the item response theory model
  • the effect and failure learning effect are the parameters to be estimated, and the Bayesian network model is constructed
  • the first determination module 403 is configured to determine the estimated value of each parameter to be estimated based on the Bayesian network model obtained by the construction module 402 and the first answer information obtained by the acquisition module 401 ;
  • the second determination module 404 is configured to determine the current answer according to the estimated value of each parameter to be estimated determined by the first determination module 403 and the second answer information acquired by the acquisition module 401 the current capabilities of the
  • the respondent's initial learning ability, topic difficulty and performance factors in the item response theory model are used to analyze the respondent's successful learning effect and failure learning effect on each knowledge point involved in the target question bank.
  • the effect is the parameter to be estimated, and the Bayesian network model is constructed; then, based on the constructed Bayesian network model and the preset number of respondents’ first answer information about the target question bank in the previous period, the value of each parameter to be estimated is determined. Estimated value; finally, according to the estimated value of each parameter to be estimated and the current answerer's second answer information about the target question bank in the current period, determine the current ability of the current answerer.
  • the first determining module 403 is configured to use a Markov chain Monte Carlo parameter estimation model to determine the value of each parameter to be estimated based on the Bayesian network model and the first answer information. estimated value.
  • the first determining module 403 is configured to determine an estimated value of each parameter to be estimated by using an expected propagation parameter estimation model based on the Bayesian network model and the first answer information.
  • the first determination module 403 is configured to determine the estimated value of each parameter to be estimated by using a variational inference method based on the Bayesian network model and the first answer information.
  • the first determining module 403 includes:
  • the first determining submodule is configured to, for each of the parameters to be estimated, determine the prior distribution of the parameter to be estimated in the current period according to the approximate posterior distribution of the parameter to be estimated in the previous period;
  • a calculation submodule configured to calculate a variational lower bound by using a variational inference method based on the Bayesian network model, the first answer information, and the prior distribution of each parameter to be estimated in the current period;
  • the estimation sub-module is used to perform parameter estimation with maximization of the variational lower bound as the objective function, and obtain an estimated value of each parameter to be estimated.
  • the first determination submodule is configured to determine the prior distribution of the parameter to be estimated in the current period by the above equation (2) according to the approximate posterior distribution of the parameter to be estimated in the previous period. test distribution.
  • the calculation submodule is configured to use a variational inference method based on the Bayesian network model, the first answer information, and the prior distribution of each parameter to be estimated in the current period , to compute the variational lower bound by the above equation (4).
  • shrink 1.
  • the second determining module 404 includes:
  • the second determination sub-module is used to determine the number of times of successful learning and the number of times of learning failure of the current answerer under each knowledge point involved in the topic corresponding to the second answer information according to the second answer information;
  • the third determining submodule is used to determine the current ability of the current answerer according to the estimated value of each parameter to be estimated, the number of successful learning and the number of failed learning.
  • the Bayesian network model is the above equation (1).
  • the distribution vector obtained after normalization is the distribution vector obtained after normalization.
  • the device 400 also includes:
  • the third determination module is configured to determine the probability of the current answerer correctly answering the candidate questions according to the current ability of the current answerer, the difficulty of the candidate questions in the target question bank, and the Bayesian network model.
  • the device 400 also includes:
  • a push module configured to push the candidate questions to the current answerer if the probability satisfies a preset condition.
  • the present disclosure also provides a computer-readable medium on which a computer program is stored, and when the program is executed by a processing device, the steps of the above-mentioned information processing method provided by the present disclosure are realized.
  • FIG. 5 it shows a schematic structural diagram of an electronic device (such as a terminal device or a server) 600 suitable for implementing an embodiment of the present disclosure.
  • the terminal equipment in the embodiment of the present disclosure may include but not limited to such as mobile phone, notebook computer, digital broadcast receiver, PDA (personal digital assistant), PAD (tablet computer), PMP (portable multimedia player), vehicle terminal (such as mobile terminals such as car navigation terminals) and fixed terminals such as digital TVs, desktop computers and the like.
  • the electronic device shown in FIG. 5 is only an example, and should not limit the functions and scope of use of the embodiments of the present disclosure.
  • an electronic device 600 may include a processing device (such as a central processing unit, a graphics processing unit, etc.) 601, which may be randomly accessed according to a program stored in a read-only memory (ROM) 602 or loaded from a storage device 608. Various appropriate actions and processes are executed by programs in the memory (RAM) 603 . In the RAM 603, various programs and data necessary for the operation of the electronic device 600 are also stored.
  • the processing device 601, ROM 602, and RAM 603 are connected to each other through a bus 604.
  • An input/output (I/O) interface 605 is also connected to the bus 604 .
  • the following devices can be connected to the I/O interface 605: input devices 606 including, for example, a touch screen, touchpad, keyboard, mouse, camera, microphone, accelerometer, gyroscope, etc.; including, for example, a liquid crystal display (LCD), speaker, vibration an output device 607 such as a computer; a storage device 608 including, for example, a magnetic tape, a hard disk, etc.; and a communication device 609.
  • the communication means 609 may allow the electronic device 600 to communicate with other devices wirelessly or by wire to exchange data. While FIG. 5 shows electronic device 600 having various means, it should be understood that implementing or having all of the means shown is not a requirement. More or fewer means may alternatively be implemented or provided.
  • embodiments of the present disclosure include a computer program product, which includes a computer program carried on a non-transitory computer readable medium, where the computer program includes program code for executing the method shown in the flowchart.
  • the computer program may be downloaded and installed from a network via communication means 609, or from storage means 608, or from ROM 602.
  • the processing device 601 When the computer program is executed by the processing device 601, the above-mentioned functions defined in the methods of the embodiments of the present disclosure are performed.
  • the computer-readable medium mentioned above in the present disclosure may be a computer-readable signal medium or a computer-readable storage medium or any combination of the two.
  • a computer readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device, or device, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to, electrical connections with one or more wires, portable computer diskettes, hard disks, random access memory (RAM), read-only memory (ROM), erasable Programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination of the above.
  • a computer-readable storage medium may be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.
  • a computer-readable signal medium may include a data signal propagated in baseband or as part of a carrier wave carrying computer-readable program code therein. Such propagated data signals may take many forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination of the foregoing.
  • a computer-readable signal medium may also be any computer-readable medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in conjunction with an instruction execution system, apparatus, or device .
  • Program code embodied on a computer readable medium may be transmitted by any appropriate medium, including but not limited to: wires, optical cables, RF (radio frequency), etc., or any suitable combination of the above.
  • the client and the server can communicate using any currently known or future network protocols such as HTTP (HyperText Transfer Protocol, Hypertext Transfer Protocol), and can communicate with digital data in any form or medium
  • HTTP HyperText Transfer Protocol
  • the communication eg, communication network
  • Examples of communication networks include local area networks (“LANs”), wide area networks (“WANs”), internetworks (e.g., the Internet), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future developed network of.
  • the above-mentioned computer-readable medium may be included in the above-mentioned electronic device, or may exist independently without being incorporated into the electronic device.
  • the above-mentioned computer-readable medium carries one or more programs, and when the above-mentioned one or more programs are executed by the electronic device, the electronic device: obtains the first answer of the preset number of answerers on the target question bank in the previous period information and the current answerer's second answer information about the target question bank in the current period, wherein the current answerer is one of the preset number of answerers; the answerer in the item response theory model
  • the respondent's initial learning ability, topic difficulty, and performance factor analysis model include the successful learning effect and failure learning effect of each knowledge point involved in the target question bank as parameters to be estimated, and a Bayesian network model is constructed. ; Based on the Bayesian network model and the first answer information, determine the estimated value of each parameter to be estimated; according to the estimated value of each parameter to be estimated and the second answer information, determine the estimated value Describe the current abilities of the current respondent.
  • Computer program code for carrying out operations of the present disclosure may be written in one or more programming languages, or combinations thereof, including but not limited to object-oriented programming languages—such as Java, Smalltalk, C++, and Includes conventional procedural programming languages - such as "C" or similar programming languages.
  • the program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server.
  • the remote computer may be connected to the user computer through any kind of network, including a local area network (LAN) or a wide area network (WAN), or may be connected to an external computer (for example, using an Internet service provider to connected via the Internet).
  • LAN local area network
  • WAN wide area network
  • Internet service provider for example, using an Internet service provider to connected via the Internet.
  • each block in a flowchart or block diagram may represent a module, program segment, or portion of code that contains one or more logical functions for implementing specified executable instructions.
  • the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or they may sometimes be executed in the reverse order, depending upon the functionality involved.
  • each block of the block diagrams and/or flowchart illustrations, and combinations of blocks in the block diagrams and/or flowchart illustrations can be implemented by a dedicated hardware-based system that performs the specified functions or operations , or may be implemented by a combination of dedicated hardware and computer instructions.
  • the modules involved in the embodiments described in the present disclosure may be implemented by software or by hardware.
  • the name of the module does not constitute a limitation of the module itself under certain circumstances.
  • the obtaining module can also be described as "obtaining the first answer information and A module of the second answer information of the current answerer on the target question bank in the current period".
  • FPGAs Field Programmable Gate Arrays
  • ASICs Application Specific Integrated Circuits
  • ASSPs Application Specific Standard Products
  • SOCs System on Chips
  • CPLDs Complex Programmable Logical devices
  • a machine-readable medium may be a tangible medium that may contain or store a program for use by or in conjunction with an instruction execution system, apparatus, or device.
  • a machine-readable medium may be a machine-readable signal medium or a machine-readable storage medium.
  • a machine-readable medium may include, but is not limited to, electronic, magnetic, optical, electromagnetic, infrared, or semiconductor systems, apparatus, or devices, or any suitable combination of the foregoing.
  • machine-readable storage media would include one or more wire-based electrical connections, portable computer discs, hard drives, random access memory (RAM), read only memory (ROM), erasable programmable read only memory (EPROM or flash memory), optical fiber, compact disk read only memory (CD-ROM), optical storage, magnetic storage, or any suitable combination of the foregoing.
  • RAM random access memory
  • ROM read only memory
  • EPROM or flash memory erasable programmable read only memory
  • CD-ROM compact disk read only memory
  • magnetic storage or any suitable combination of the foregoing.
  • Example 1 provides an information processing method, including: obtaining the first answer information of a preset number of respondents in the previous period about the target question bank and the current respondent's information about the target question bank in the current period
  • the second answer information of the target question bank wherein, the current answerer is one of the preset number of answerers; the initial learning ability, difficulty level and The successful learning effect and the failure learning effect of the respondent in the performance factor analysis model on each knowledge point involved in the target question bank are parameters to be estimated, and a Bayesian network model is constructed; based on the Bayesian network model and the first answer information to determine the estimated value of each parameter to be estimated; according to the estimated value of each parameter to be estimated and the second answer information, determine the current ability of the current answerer.
  • Example 2 provides the method of Example 1, wherein the estimated value of each parameter to be estimated is determined based on the Bayesian network model and the first answer information, The method includes: determining the estimated value of each parameter to be estimated by using a variational inference method based on the Bayesian network model and the first answer information.
  • Example 3 provides the method of Example 1. Based on the Bayesian network model and the first answer information, the variational inference method is used to determine each of the estimated The estimated value of the parameter includes: for each parameter to be estimated, according to the approximate posterior distribution of the parameter to be estimated in the previous period, determining the prior distribution of the parameter to be estimated in the current period; based on the The Bayesian network model, the first answer information and the prior distribution of each parameter to be estimated in the current period, using the variational inference method to calculate the variational lower bound; maximize the variational lower bound Parameter estimation is performed for the objective function, and an estimated value of each parameter to be estimated is obtained.
  • Example 4 provides the method of Example 3, wherein the parameter to be estimated in the current period is determined according to the approximate posterior distribution of the parameter to be estimated in the previous period Prior distributions, including:
  • the prior distribution of the parameter to be estimated in the current period is determined by the following formula, including:
  • p(parameter) is the prior distribution of the parameter to be estimated in the current period
  • q m (parameter) is the approximate posterior distribution of the parameter to be estimated in the previous period
  • decay is a weight coefficient
  • Example 5 provides the method of Example 3, wherein based on the Bayesian network model, the first answer information and each of the parameters to be estimated in the current period The prior distribution of , using the variational inference method to calculate the variational lower bound, including:
  • the variational inference method is used to calculate the variational lower bound by the following formula:
  • ELBO is the variational lower bound; is a vector of initial abilities for each said respondent; is a vector formed by the difficulty of each question in the target question bank; is a vector formed by each answerer's successful learning effect on each knowledge point involved in the target question bank; is a vector composed of each respondent's failure learning effect on each knowledge point involved in the target question bank; likelihood is a reconstruction likelihood function based on the variational posterior distribution, according to the Bayesian network model and each of the parameters to be estimated; shrink and enhance are hyperparameters; g is equal to the product of the number of questions corresponding to the preset number and the first answer information; max is equal to the total number of answerers about the target question bank The product of the number of questions contained in the target question bank; for right expectations; for The KL divergence of the approximate posterior distribution of and its prior distribution; for The KL divergence of the approximate posterior distribution of and its prior distribution; for The KL divergence of the approximate posterior distribution of and its prior distribution; for The KL divergence of the approximate
  • Example 7 provides the method of Example 1, wherein the current ability of the current answerer is determined according to the estimated value of each of the parameters to be estimated and the second answer information , including: according to the second answer information, determine the number of successful learning and the number of learning failures of the current answerer under each knowledge point involved in the topic corresponding to the second answer information; The estimated value of the parameter to be estimated, the number of successful learning and the number of failed learning determine the current ability of the current answerer.
  • Example 8 provides the method of Example 1, and the Bayesian network model is:
  • ⁇ i is the initial learning ability of the i-th answerer
  • b j is the difficulty of the j-th question
  • Example 9 provides the method of Example 8, is the distribution vector obtained after normalization.
  • Example 10 provides the method of any one of Example 1-Example 9, the method further includes: according to the current ability of the current answerer and the candidate questions in the target question bank The difficulty of the question and the Bayesian network model determine the probability that the current answerer answers the candidate question correctly.
  • Example 11 provides the method of Example 10, the method further comprising: if the probability satisfies a preset condition, pushing the candidate question to the current answerer.
  • Example 12 provides an information processing device, including: an acquisition module, configured to acquire the first answer information and the current answer of a preset number of answerers on the target question bank in the previous period The respondent's second answer information about the target question bank in the current period, wherein the current respondent is one of the preset number of respondent; a building block for responding to the item in the theoretical model
  • the respondent's initial learning ability, topic difficulty and performance factor analysis model, the successful learning effect and failure learning effect of the respondent on each knowledge point involved in the target question bank are parameters to be estimated, and a Bayesian network is constructed.
  • the second determination module is configured to determine the current question answerer's current question mark according to the estimated value of each parameter to be estimated determined by the first determination module and the second answer information obtained by the acquisition module.
  • Example 13 provides a computer-readable medium on which a computer program is stored, and when the program is executed by a processing device, the steps of the method in any one of Examples 1-11 are provided.
  • Example 14 provides an electronic device, including: a storage device, on which a computer program is stored; a processing device, configured to execute the computer program in the storage device, to Implement the steps of any one of the methods in Examples 1-11.

Landscapes

  • Business, Economics & Management (AREA)
  • Engineering & Computer Science (AREA)
  • Human Resources & Organizations (AREA)
  • Strategic Management (AREA)
  • Educational Administration (AREA)
  • Economics (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Tourism & Hospitality (AREA)
  • General Physics & Mathematics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Marketing (AREA)
  • General Business, Economics & Management (AREA)
  • Development Economics (AREA)
  • Game Theory and Decision Science (AREA)
  • Operations Research (AREA)
  • Quality & Reliability (AREA)
  • Educational Technology (AREA)
  • Primary Health Care (AREA)
  • General Health & Medical Sciences (AREA)
  • Health & Medical Sciences (AREA)
  • Mathematical Analysis (AREA)
  • Mathematical Optimization (AREA)
  • Artificial Intelligence (AREA)
  • Computational Mathematics (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Probability & Statistics with Applications (AREA)
  • Algebra (AREA)
  • Pure & Applied Mathematics (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

本公开涉及一种信息处理方法、装置、可读存储介质及电子设备。方法包括：获取预设数量的答题者在上一时段关于目标题库的第一答题信息和当前答题者在当前时段关于目标题库的第二答题信息；以项目反应理论模型中答题者的初始学习能力、题目难度和表现因素分析模型中答题者对目标题库所涉及的各知识点的成功学习效应、失败学习效应为待估计参数，构建贝叶斯网络模型；基于模型和第一答题信息，确定各待估计参数的估计值；根据各估计值和第二答题信息，确定当前答题者的当前能力。这样，随着答题者针对目标题库的答题练习积累，成功、失败学习效应持续更新，进而实现动态追踪答题者学习能力变化的功能，适用线上练习场景中答题者学习能力的动态追踪。

Description

信息处理方法、装置、可读存储介质及电子设备
相关申请的交叉引用
本申请要求于2021年12月21日提交的，申请号为202111574789.8、发明名称为“信息处理方法、装置、可读存储介质及电子设备”的中国专利申请的优先权，该申请的全部内容通过引用结合在本申请中。
技术领域
本公开涉及信息处理技术领域，具体地，涉及一种信息处理方法、装置、可读存储介质及电子设备。
背景技术
在线上练习的场景中，具有时间序列性质的答题数据越来越普遍，对于这样的数据，动态追踪答题者的实时变化的能力对于评估学习效果具有重要的作用和意义。现阶段，主要通过项目反应理论(Item Response Theory，IRT)、表现因素分析(Performance Factor Analysis，PFA)、知识追踪模型(Knowledge Tracing Model，KT)等模型来评估答题者能力，但它们均无法动态追踪答题者能力的变化。因此，如何实现答题者能力的动态追踪成为研究的重点。
发明内容
提供该发明内容部分以便以简要的形式介绍构思，这些构思将在后面的具体实施方式部分被详细描述。该发明内容部分并不旨在标识要求保护的技术方案的关键特征或必要特征，也不旨在用于限制所要求的保护的技术方案的范围。
第一方面，本公开提供一种信息处理方法，包括：
获取预设数量的答题者在上一时段关于目标题库的第一答题信息和当前答题者在当前时段关于所述目标题库的第二答题信息，其中，所述当前答题者为所述预设数量的答题者中的一者；
以项目反应理论模型中的所述答题者的初始学习能力、题目难度和表现因素分析模型中的所述答题者对所述目标题库所涉及的每一知识点的成功学习效应、失败学习效应为待估计参数，构建贝叶斯网络模型；
基于所述贝叶斯网络模型和所述第一答题信息，确定每一所述待估计参数的估计值；
根据每一所述待估计参数的估计值和所述第二答题信息，确定所述当前答题者的当前 能力。
第二方面，本公开提供一种信息处理装置，包括：
获取模块，用于获取预设数量的答题者在上一时段关于目标题库的第一答题信息和当前答题者在当前时段关于所述目标题库的第二答题信息，其中，所述当前答题者为所述预设数量的答题者中的一者；
构建模块，用于以项目反应理论模型中的所述答题者的初始学习能力、题目难度和表现因素分析模型中的所述答题者对所述目标题库所涉及的每一知识点的成功学习效应、失败学习效应为待估计参数，构建贝叶斯网络模型；
第一确定模块，用于基于所述构建模块得到的所述贝叶斯网络模型和所述获取模块获取到的所述第一答题信息，确定每一所述待估计参数的估计值；
第二确定模块，用于根据所述第一确定模块确定出的每一所述待估计参数的估计值和所述获取模块获取到的所述第二答题信息，确定所述当前答题者的当前能力。
第三方面，本公开提供一种计算机可读介质，其上存储有计算机程序，该程序被处理装置执行时实现本公开第一方面提供的所述方法的步骤。
第四方面，本公开提供一种电子设备，包括：
存储装置，其上存储有计算机程序；
处理装置，用于执行所述存储装置中的所述计算机程序，以实现本公开第一方面提供的所述方法的步骤。
在上述技术方案中，首先，以项目反应理论模型中的答题者的初始学习能力、题目难度和表现因素分析模型中的答题者对目标题库所涉及的每一知识点的成功学习效应、失败学习效应为待估计参数，构建贝叶斯网络模型；然后，基于构建的贝叶斯网络模型和预设数量的答题者在上一时段关于目标题库的第一答题信息，确定每一待估计参数的估计值；最后，根据每一待估计参数的估计值和当前答题者在当前时段关于目标题库的第二答题信息，确定当前答题者的当前能力。这样，随着答题者针对目标题库的答题练习积累，答题者对目标题库所涉及的每一知识点的成功学习效应、失败学习效应持续更新，进而实现动态追踪答题者学习能力变化的功能，适用于线上练习场景中答题者学习能力的动态追踪。另外，在确定当前答题者的当前能力时，参考了其初始学习能力，即参考了在当前时段之前当前答题者已获得的能力水平，这样，更贴近实际学习场景，从而能够准确评估当前答题者的当前能力。
本公开的其他特征和优点将在随后的具体实施方式部分予以详细说明。
附图说明
结合附图并参考以下具体实施方式，本公开各实施例的上述和其他特征、优点及方面将变得更加明显。贯穿附图中，相同或相似的附图标记表示相同或相似的元素。应当理解附图是示意性的，原件和元素不一定按照比例绘制。在附图中：
图1是根据一示例性实施例示出的一种信息处理方法的流程图。
图2是根据另一示例性实施例示出的一种信息处理方法的流程图。
图3是根据另一示例性实施例示出的一种信息处理方法的流程图。
图4是根据一示例性实施例示出的一种信息处理装置的框图。
图5是根据一示例性实施例示出的一种电子设备的框图。
具体实施方式
下面将参照附图更详细地描述本公开的实施例。虽然附图中显示了本公开的某些实施例，然而应当理解的是，本公开可以通过各种形式来实现，而且不应该被解释为限于这里阐述的实施例，相反提供这些实施例是为了更加透彻和完整地理解本公开。应当理解的是，本公开的附图及实施例仅用于示例性作用，并非用于限制本公开的保护范围。
应当理解，本公开的方法实施方式中记载的各个步骤可以按照不同的顺序执行，和/或并行执行。此外，方法实施方式可以包括附加的步骤和/或省略执行示出的步骤。本公开的范围在此方面不受限制。
本文使用的术语“包括”及其变形是开放性包括，即“包括但不限于”。术语“基于”是“至少部分地基于”。术语“一个实施例”表示“至少一个实施例”；术语“另一实施例”表示“至少一个另外的实施例”；术语“一些实施例”表示“至少一些实施例”。其他术语的相关定义将在下文描述中给出。
需要注意，本公开中提及的“第一”、“第二”等概念仅用于对不同的装置、模块或单元进行区分，并非用于限定这些装置、模块或单元所执行的功能的顺序或者相互依存关系。
需要注意，本公开中提及的“一个”、“多个”的修饰是示意性而非限制性的，本领域技术人员应当理解，除非在上下文另有明确指出，否则应该理解为“一个或多个”。
本公开实施方式中的多个装置之间所交互的消息或者信息的名称仅用于说明性的目的，而并不是用于对这些消息或信息的范围进行限制。
图1是根据一示例性实施例示出的一种信息处理方法的流程图。如图1所示，上述方法包括S101～S104。
在S101中，获取预设数量的答题者在上一时段关于目标题库的第一答题信息和当前答题者在当前时段关于目标题库的第二答题信息。
在本公开中，当前答题者为预设数量的答题者中的一者。第一答题信息和第二答题信 息可以包括作答题目及答题情况(如答对或答错)等信息。另外，可以天、周、季度等为单位作为一个时段。
另外，可根据实际需求选取目标题库以及答题者的预设数量。例如，可以获取A市会计从业人员在上周关于注册会计题库的第一答题信息，还可获取B地区驾照考取人员在上季度关于科目四的第一答题信息。
在S102中，以项目反应理论模型中的答题者的初始学习能力、题目难度和表现因素分析模型中的答题者对目标题库所涉及的每一知识点的成功学习效应、失败学习效应为待估计参数，构建贝叶斯网络模型。
在本公开中，答题者为上述预设数量的答题者中的任一者，学习效应表征在答题者学习次数累计的过程中，其能力的变化速度。
在S103中，基于贝叶斯网络模型和第一答题信息，确定每一待估计参数的估计值。
在S104中，根据每一待估计参数的估计值和第二答题信息，确定当前答题者的当前能力。
另外，需要说明的是，上述S102可以在上述S101之后执行(如图1所示)，也可以与上述S101同时执行，还可以在上述S101之前执行，本公开不作具体限定。
在上述技术方案中，首先，以项目反应理论模型中的答题者的初始学习能力、题目难度和表现因素分析模型中的答题者对目标题库所涉及的每一知识点的成功学习效应、失败学习效应为待估计参数，构建贝叶斯网络模型；然后，基于构建的贝叶斯网络模型和预设数量的答题者在上一时段关于目标题库的第一答题信息，确定每一待估计参数的估计值；最后，根据每一待估计参数的估计值和当前答题者在当前时段关于目标题库的第二答题信息，确定当前答题者的当前能力。这样，随着答题者针对目标题库的答题练习积累，答题者对目标题库所涉及的每一知识点的成功学习效应、失败学习效应持续更新，进而实现动态追踪答题者学习能力变化的功能，适用于线上练习场景中答题者学习能力的动态追踪。另外，在确定当前答题者的当前能力时，参考了其初始学习能力，即参考了在当前时段之前当前答题者已获得的能力水平，这样，更贴近实际学习场景，从而能够准确评估当前答题者的当前能力。
示例地，贝叶斯网络模型如下等式(1)所示：
Figure PCTCN2022133207-appb-000001
其中，
Figure PCTCN2022133207-appb-000002
为第i个答题者答对第j个题目的概率，y ij＝1；θ i为所述第i个答题者的初始学习能力；b j为所述第j个题目的题目难度；
Figure PCTCN2022133207-appb-000003
为由所述第i个答题者在所述目标答题库所涉及的每一知识点下学习成功的次数构成的向量；
Figure PCTCN2022133207-appb-000004
为由所述第i个答题者在所述目标答题库所涉及的每一知识点下学习失败的次数构成的向量；
Figure PCTCN2022133207-appb-000005
为所述第i个答题者对所述目标题库所涉及的每一知识点的成功学习效应；
Figure PCTCN2022133207-appb-000006
为所述第i个答题者对所述目标题库中所涉及的每一知识点的失败学习效应；
Figure PCTCN2022133207-appb-000007
为所述第j个题目的所涉及的知识点的分布向量；T为转置。
其中，
Figure PCTCN2022133207-appb-000008
以及
Figure PCTCN2022133207-appb-000009
可以均为K*1的向量，K为目标题库中所涉及的知识点的数量。
在一种实施方式中，
Figure PCTCN2022133207-appb-000010
等同于一个二分向量，即包含该知识点的位置为1，不包含的位置为0。
示例地，目标题库涉及5个知识点，即知识点1、知识点2、知识点3、知识点4以及知识点5，其中，第j个题目涉及知识点1、知识点3以及知识点4，则第j个题目所涉及的知识点的分布向量
Figure PCTCN2022133207-appb-000011
采用上述实施方式，涉及的知识点越多的题目，答题者答对该题目的概率越高，这样，概率与题目所涉及的知识点的数量直接相关，导致准确度不高。因此，在另一种实施方式中，对上述二分向量进行归一化处理，并将归一化处理后所得的分布向量确定为
Figure PCTCN2022133207-appb-000012
其中，归一化处理后所得的分布向量中各元素之和等于1。这样，概率不会受题目所涉及的知识点的数量增多而增大，从而可以提升概率预测的准确度。
示例地，目标题库涉及5个知识点，即知识点1、知识点2、知识点3、知识点4以及知识点5，其中，第j个题目涉及知识点1、知识点3以及知识点4，则第j个题目所涉及的知识点的分布向量
Figure PCTCN2022133207-appb-000013
另外，还可以根据不同知识点的重要程度来调整上述分布向量，具体来说，包含重要知识点的位置处的元素值相对大些，包含非重要知识点的位置处的元素值相对小些，从而 体现知识点的主次关系。
下面针对上述S103中的基于贝叶斯网络模型和第一答题信息，确定每一待估计参数的估计值的具体实施方式进行详细说明。
在一种实施方式中，可以基于贝叶斯网络模型和第一答题信息，采用马尔科夫链蒙特卡洛(Markov Chain Monte Carlo，MCMC)参数估计模型来确定每一待估计参数的估计值。
在另一种实施方式中，可以基于贝叶斯网络模型和第一答题信息，采用期望传播(Expectation Propagation，EP)参数估计模型来确定每一待估计参数的估计值。
在又一种实施方式中，可以基于贝叶斯网络模型和第一答题信息，采用变分推断方法确定每一待估计参数的估计值。
具体来说，可以通过以下步骤(1)～步骤(3)来实现：
(1)针对每一待估计参数，根据该待估计参数在上一时段的近似后验分布，确定该待估计参数在当前时段的先验分布。
(2)基于贝叶斯网络模型、第一答题信息以及每一待估计参数在当前时段的先验分布，采用变分推断方法，计算变分下界。
(3)以变分下界最大化为目标函数进行参数估计，得到每一待估计参数的估计值。
示例地，可以以变分下界最大化为目标函数进行参数估计，通过随机梯度下降法来得到每一待估计参数的估计值。由于采用随机梯度下降法来得到每一待估计参数的估计值的具体实施方式属于本领域技术人员公知的，本公开不再赘述。
在上述实施方式中，根据该待估计参数在上一时段的近似后验分布，确定该待估计参数在当前时段的先验分布，可以使得变分推断方法适用于动态和连续的估参场景，且计算量和内存占用量都优于上述提到过的MCMC和EP估参方法，可以相对准确且快速地估计每一待估计参数的估计值。
下面针对上述步骤(1)中的根据该待估计参数在上一时段的近似后验分布，确定该待估计参数在当前时段的先验分布的具体实施方式进行详细说明。
具体来说，可以通过多种方来实现，在一种实施方式中，可以将该待估计参数在上一时段的近似后验分布确定为该待估计参数在当前时段的先验分布。
在另一种实施方式中，可以根据该待估计参数在上一时段的近似后验分布，通过以下等式(2)来确定该待估计参数在当前时段的先验分布：
p(parameter)＝(1-decay)*q m(parameter)+decay*p(parameter)   (2)
其中，p(parameter)为该待估计参数parameter在当前时段的先验分布； q m(parameter)为该待估计参数parameter在上一时段的近似后验分布；decay为权重系数。
在该种实施方式中，没有将该待估计参数在当前时段的先验分布直接替换为该待估计参数在上一时段的近似后验分布，而是采用加权平均的方式来确定该待估计参数在当前时段的先验分布，从而可以使得过去时段的待估计参数的近似后验分布逐渐影响当前时段和未来时段的近似后验分布的更新，这样，可以避免因初始阶段(即前几个时段)答题信息少，后验分布估计不稳定所带来的影响，进而提升参数估计的准确度。
下面针对上述步骤(2)中的基于贝叶斯网络模型、第一答题信息以及每一待估计参数在当前时段的先验分布，采用变分推断方法，计算变分下界的具体实施方式进行详细说明。
具体来说，可以通过多种方式来实现，在一种实施方式中，可以基于贝叶斯网络模型、第一答题信息以及每一待估计参数在当前时段的先验分布，采用变分推断方法，通过以下等式(3)来计算变分下界：
Figure PCTCN2022133207-appb-000014
其中，ELBO为变分下界；
Figure PCTCN2022133207-appb-000015
为由预设数量的答题者中每一答题者的初始能力构成的向量；
Figure PCTCN2022133207-appb-000016
为由目标题库中每一题目的题目难度构成的向量；
Figure PCTCN2022133207-appb-000017
为由每一答题者对目标题库所涉及的每一知识点的成功学习效应构成的向量；
Figure PCTCN2022133207-appb-000018
为由每一答题者对目标题库所涉及的每一知识点的失败学习效应构成的向量；likelihood为基于变分后验分布的重建似然函数，根据贝叶斯网络模型和每一待估计参数确定；
Figure PCTCN2022133207-appb-000019
为对
Figure PCTCN2022133207-appb-000020
的期望；
Figure PCTCN2022133207-appb-000021
为
Figure PCTCN2022133207-appb-000022
的近似后验分布与其先验分布的KL散度；
Figure PCTCN2022133207-appb-000023
为
Figure PCTCN2022133207-appb-000024
的近似后验分布与其先验分布的KL散度；
Figure PCTCN2022133207-appb-000025
为
Figure PCTCN2022133207-appb-000026
的近似后验分布与其先验分布的KL散度；
Figure PCTCN2022133207-appb-000027
为
Figure PCTCN2022133207-appb-000028
的近似后验分布与其先验分布的KL散度。
示例地，
Figure PCTCN2022133207-appb-000029
其中，
Figure PCTCN2022133207-appb-000030
为
Figure PCTCN2022133207-appb-000031
的后验联合分布，
Figure PCTCN2022133207-appb-000032
为对
Figure PCTCN2022133207-appb-000033
的期望，通过将每一待估计参数、
Figure PCTCN2022133207-appb-000034
代入上述贝叶斯网络而计算得到
Figure PCTCN2022133207-appb-000035
在另一种实施方式中，可以基于贝叶斯网络模型、第一答题信息以及每一待估计参数在当前时段的先验分布，采用变分推断方法，通过以下等式(4)来计算变分下界：
Figure PCTCN2022133207-appb-000036
其中，shrink和enhance为超参数；g等于预设数量与第一答题信息对应的题目的数量的乘积；max等于关于目标题库的答题者总数与目标题库所包含题目的数量的乘积。
在该种实施方式中，在参数估计过程中，结合贝叶斯网络模型的特点，设置shrink、enhance为超参数，能够提升参数估计的准确度。另外，由于初始阶段答题次数和信息少，主要对答题者的初始学习能力、题目难度进行估计，但随着答题者针对目标题库的答题练习积累，答题者的初始学习能力、题目难度基本固定，可以使得变分推断方法后期更加专注于估计答题者对目标题库所涉及的每一知识点的成功学习效应、失败学习效应这两个待估计参数，以提升参数估计的准确度，因此，在变分下界ELBO中，加强(enhance)了已经学习好的待估计参数对应的KL散度，即
Figure PCTCN2022133207-appb-000037
和
Figure PCTCN2022133207-appb-000038
另外，在一种实施方式中，各时段的shrink均为预设值。
在另一种实施方式中，初始时段(即第1个时段)的shrink为预设值，在后续时段shrink＝1，这样，当前时段的先验分布已经包含前一时段学习到的参数近似后验分布，即保留未来时段的先验分布包含学习到的历史信息，从而提升参数估计的准确度。
下面针对上述S104中的根据每一待估计参数的估计值和第二答题信息，确定当前答题者的当前能力的具体实施方式进行详细说明。具体来说，可以通过以下步骤来实现：
首先，根据第二答题信息，确定当前答题者在第二答题信息对应的题目所涉及的每一知识点下学习成功的次数和学习失败的次数。
其中，学习成功的次数即答对的次数，学习失败的次数即答错的次数。
然后，根据每一待估计参数的估计值、学习成功的次数以及学习失败的次数，确定当 前答题者的当前能力。
示例地，当前答题者的当前能力
Figure PCTCN2022133207-appb-000039
图2是根据另一示例性实施例示出的一种信息处理方法的流程图。如图2所示，上述方法还包括S105。
在S105中，根据当前答题者的当前能力和目标题库中候选题目的题目难度以及贝叶斯网络模型，确定当前答题者答对候选题目的概率。
在本公开中，可以从上述
Figure PCTCN2022133207-appb-000040
中查找到候选题目的题目难度。
具体来说，贝叶斯网络模型可以简化为：
Figure PCTCN2022133207-appb-000041
这样，通过将当前答题者的当前能力和目标题库中候选题目的题目难度代入到该简化后所得的贝叶斯网络模型中，得到当前答题者答对候选题目的概率。由于能够准确评估答题者的当前能力和候选题目的题目难度，进而能够保证概率预测的准确度。
另外，可以根据上述概率，向答题者自动推送题目。具体来说，如图3所示，上述方法还包括S106。
在S106中，若当前答题者答对候选题目的概率满足预设条件，则向当前答题者推送候选题目。
在一种实施方式中，预设条件可以为上述概率处于预设概率范围内，例如，0.5～0.9。
在另一种实施方式中，预设条件可以为候选题目的熵值大于预设阈值，其中，候选题目的熵值H＝-plogp-(1-p)log(1-p)，其中，p为上述概率，即
Figure PCTCN2022133207-appb-000042
根据最大熵原理，候选题目的熵值越大，则答题者练习该题目所能获取的信息量越多，所以当H大于预设阈值时，向当前答题者推送候选题目。
图4是根据一示例性实施例示出的一种信息处理装置的框图。如图4所示，该装置400包括：
获取模块401，用于获取预设数量的答题者在上一时段关于目标题库的第一答题信息和当前答题者在当前时段关于所述目标题库的第二答题信息，其中，所述当前答题者为所述预设数量的答题者中的一者；
构建模块402，用于以项目反应理论模型中的所述答题者的初始学习能力、题目难度和表现因素分析模型中的所述答题者对所述目标题库所涉及的每一知识点的成功学习效应、失败学习效应为待估计参数，构建贝叶斯网络模型；
第一确定模块403，用于基于所述构建模块402得到的所述贝叶斯网络模型和所述获取模块401获取到的所述第一答题信息，确定每一所述待估计参数的估计值；
第二确定模块404，用于根据所述第一确定模块403确定出的每一所述待估计参数的估计值和所述获取模块401获取到的所述第二答题信息，确定所述当前答题者的当前能力。
在上述技术方案中，首先，以项目反应理论模型中的答题者的初始学习能力、题目难度和表现因素分析模型中的答题者对目标题库所涉及的每一知识点的成功学习效应、失败学习效应为待估计参数，构建贝叶斯网络模型；然后，基于构建的贝叶斯网络模型和预设数量的答题者在上一时段关于目标题库的第一答题信息，确定每一待估计参数的估计值；最后，根据每一待估计参数的估计值和当前答题者在当前时段关于目标题库的第二答题信息，确定当前答题者的当前能力。这样，随着答题者针对目标题库的答题练习积累，答题者对目标题库所涉及的每一知识点的成功学习效应、失败学习效应持续更新，进而实现动态追踪答题者学习能力变化的功能，适用于线上练习场景中答题者学习能力的动态追踪。另外，在确定当前答题者的当前能力时，参考了其初始学习能力，即参考了在当前时段之前当前答题者已获得的能力水平，这样，更贴近实际学习场景，从而能够准确评估当前答题者的当前能力。
可选地，所述第一确定模块403用于基于所述贝叶斯网络模型和所述第一答题信息，采用马尔科夫链蒙特卡洛参数估计模型来确定每一所述待估计参数的估计值。
可选地，所述第一确定模块403用于基于所述贝叶斯网络模型和所述第一答题信息，采用期望传播参数估计模型来确定每一所述待估计参数的估计值。
可选地，所述第一确定模块403用于基于所述贝叶斯网络模型和所述第一答题信息，采用变分推断方法确定每一所述待估计参数的估计值。
可选地，所述第一确定模块403包括：
第一确定子模块，用于针对每一所述待估计参数，根据该待估计参数在所述上一时段的近似后验分布，确定该待估计参数在所述当前时段的先验分布；
计算子模块，用于基于所述贝叶斯网络模型、所述第一答题信息以及每一所述待估计参数在所述当前时段的先验分布，采用变分推断方法，计算变分下界；
估计子模块，用于以所述变分下界最大化为目标函数进行参数估计，得到每一所述待估计参数的估计值。
可选地，所述第一确定子模块用于根据该待估计参数在所述上一时段的近似后验分布， 通过以上等式(2)来确定该待估计参数在所述当前时段的先验分布。
可选地，所述计算子模块，用于基于所述贝叶斯网络模型、所述第一答题信息以及每一所述待估计参数在所述当前时段的先验分布，采用变分推断方法，通过以上等式(4)来计算变分下界。
可选地，shrink＝1。
可选地，所述第二确定模块404包括：
第二确定子模块，用于根据所述第二答题信息，确定所述当前答题者在所述第二答题信息对应的题目所涉及的每一知识点下学习成功的次数和学习失败的次数；
第三确定子模块，用于根据每一所述待估计参数的估计值、所述学习成功的次数以及所述学习失败的次数，确定所述当前答题者的当前能力。
可选地，所述贝叶斯网络模型为以上等式(1)。
可选地，
Figure PCTCN2022133207-appb-000043
为归一化处理后所得的分布向量。
可选地，所述装置400还包括：
第三确定模块，用于根据所述当前答题者的当前能力和所述目标题库中候选题目的题目难度以及所述贝叶斯网络模型，确定所述当前答题者答对所述候选题目的概率。
可选地，所述装置400还包括：
推送模块，用于若所述概率满足预设条件，则向所述当前答题者推送所述候选题目。
本公开还提供一种计算机可读介质，其上存储有计算机程序，该程序被处理装置执行时实现本公开提供的上述信息处理方法的步骤。
下面参考图5，其示出了适于用来实现本公开实施例的电子设备(例如终端设备或服务器)600的结构示意图。本公开实施例中的终端设备可以包括但不限于诸如移动电话、笔记本电脑、数字广播接收器、PDA(个人数字助理)、PAD(平板电脑)、PMP(便携式多媒体播放器)、车载终端(例如车载导航终端)等等的移动终端以及诸如数字TV、台式计算机等等的固定终端。图5示出的电子设备仅仅是一个示例，不应对本公开实施例的功能和使用范围带来任何限制。
如图5所示，电子设备600可以包括处理装置(例如中央处理器、图形处理器等)601，其可以根据存储在只读存储器(ROM)602中的程序或者从存储装置608加载到随机访问存储器(RAM)603中的程序而执行各种适当的动作和处理。在RAM 603中，还存储有电子设备600操作所需的各种程序和数据。处理装置601、ROM 602以及RAM 603通过总线604彼此相连。输入/输出(I/O)接口605也连接至总线604。
通常，以下装置可以连接至I/O接口605：包括例如触摸屏、触摸板、键盘、鼠标、摄像头、麦克风、加速度计、陀螺仪等的输入装置606；包括例如液晶显示器(LCD)、扬声 器、振动器等的输出装置607；包括例如磁带、硬盘等的存储装置608；以及通信装置609。通信装置609可以允许电子设备600与其他设备进行无线或有线通信以交换数据。虽然图5示出了具有各种装置的电子设备600，但是应理解的是，并不要求实施或具备所有示出的装置。可以替代地实施或具备更多或更少的装置。
特别地，根据本公开的实施例，上文参考流程图描述的过程可以被实现为计算机软件程序。例如，本公开的实施例包括一种计算机程序产品，其包括承载在非暂态计算机可读介质上的计算机程序，该计算机程序包含用于执行流程图所示的方法的程序代码。在这样的实施例中，该计算机程序可以通过通信装置609从网络上被下载和安装，或者从存储装置608被安装，或者从ROM 602被安装。在该计算机程序被处理装置601执行时，执行本公开实施例的方法中限定的上述功能。
需要说明的是，本公开上述的计算机可读介质可以是计算机可读信号介质或者计算机可读存储介质或者是上述两者的任意组合。计算机可读存储介质例如可以是——但不限于——电、磁、光、电磁、红外线、或半导体的系统、装置或器件，或者任意以上的组合。计算机可读存储介质的更具体的例子可以包括但不限于：具有一个或多个导线的电连接、便携式计算机磁盘、硬盘、随机访问存储器(RAM)、只读存储器(ROM)、可擦式可编程只读存储器(EPROM或闪存)、光纤、便携式紧凑磁盘只读存储器(CD-ROM)、光存储器件、磁存储器件、或者上述的任意合适的组合。在本公开中，计算机可读存储介质可以是任何包含或存储程序的有形介质，该程序可以被指令执行系统、装置或者器件使用或者与其结合使用。而在本公开中，计算机可读信号介质可以包括在基带中或者作为载波一部分传播的数据信号，其中承载了计算机可读的程序代码。这种传播的数据信号可以采用多种形式，包括但不限于电磁信号、光信号或上述的任意合适的组合。计算机可读信号介质还可以是计算机可读存储介质以外的任何计算机可读介质，该计算机可读信号介质可以发送、传播或者传输用于由指令执行系统、装置或者器件使用或者与其结合使用的程序。计算机可读介质上包含的程序代码可以用任何适当的介质传输，包括但不限于：电线、光缆、RF(射频)等等，或者上述的任意合适的组合。
在一些实施方式中，客户端、服务器可以利用诸如HTTP(HyperText Transfer Protocol，超文本传输协议)之类的任何当前已知或未来研发的网络协议进行通信，并且可以与任意形式或介质的数字数据通信(例如，通信网络)互连。通信网络的示例包括局域网(“LAN”)，广域网(“WAN”)，网际网(例如，互联网)以及端对端网络(例如，ad hoc端对端网络)，以及任何当前已知或未来研发的网络。
上述计算机可读介质可以是上述电子设备中所包含的；也可以是单独存在，而未装配入该电子设备中。
上述计算机可读介质承载有一个或者多个程序，当上述一个或者多个程序被该电子设备执行时，使得该电子设备：获取预设数量的答题者在上一时段关于目标题库的第一答题信息和当前答题者在当前时段关于所述目标题库的第二答题信息，其中，所述当前答题者为所述预设数量的答题者中的一者；以项目反应理论模型中的所述答题者的初始学习能力、题目难度和表现因素分析模型中的所述答题者对所述目标题库所涉及的每一知识点的成功学习效应、失败学习效应为待估计参数，构建贝叶斯网络模型；基于所述贝叶斯网络模型和所述第一答题信息，确定每一所述待估计参数的估计值；根据每一所述待估计参数的估计值和所述第二答题信息，确定所述当前答题者的当前能力。
可以以一种或多种程序设计语言或其组合来编写用于执行本公开的操作的计算机程序代码，上述程序设计语言包括但不限于面向对象的程序设计语言—诸如Java、Smalltalk、C++，还包括常规的过程式程序设计语言——诸如“C”语言或类似的程序设计语言。程序代码可以完全地在用户计算机上执行、部分地在用户计算机上执行、作为一个独立的软件包执行、部分在用户计算机上部分在远程计算机上执行、或者完全在远程计算机或服务器上执行。在涉及远程计算机的情形中，远程计算机可以通过任意种类的网络——包括局域网(LAN)或广域网(WAN)——连接到用户计算机，或者，可以连接到外部计算机(例如利用因特网服务提供商来通过因特网连接)。
附图中的流程图和框图，图示了按照本公开各种实施例的系统、方法和计算机程序产品的可能实现的体系架构、功能和操作。在这点上，流程图或框图中的每个方框可以代表一个模块、程序段、或代码的一部分，该模块、程序段、或代码的一部分包含一个或多个用于实现规定的逻辑功能的可执行指令。也应当注意，在有些作为替换的实现中，方框中所标注的功能也可以以不同于附图中所标注的顺序发生。例如，两个接连地表示的方框实际上可以基本并行地执行，它们有时也可以按相反的顺序执行，这依所涉及的功能而定。也要注意的是，框图和/或流程图中的每个方框、以及框图和/或流程图中的方框的组合，可以用执行规定的功能或操作的专用的基于硬件的系统来实现，或者可以用专用硬件与计算机指令的组合来实现。
描述于本公开实施例中所涉及到的模块可以通过软件的方式实现，也可以通过硬件的方式来实现。其中，模块的名称在某种情况下并不构成对该模块本身的限定，例如，获取模块还可以被描述为“获取预设数量的答题者在上一时段关于目标题库的第一答题信息和当前答题者在当前时段关于所述目标题库的第二答题信息的模块”。
本文中以上描述的功能可以至少部分地由一个或多个硬件逻辑部件来执行。例如，非限制性地，可以使用的示范类型的硬件逻辑部件包括：现场可编程门阵列(FPGA)、专用集成电路(ASIC)、专用标准产品(ASSP)、片上系统(SOC)、复杂可编程逻辑设备(CPLD) 等等。
在本公开的上下文中，机器可读介质可以是有形的介质，其可以包含或存储以供指令执行系统、装置或设备使用或与指令执行系统、装置或设备结合地使用的程序。机器可读介质可以是机器可读信号介质或机器可读储存介质。机器可读介质可以包括但不限于电子的、磁性的、光学的、电磁的、红外的、或半导体系统、装置或设备，或者上述内容的任何合适组合。机器可读存储介质的更具体示例会包括基于一个或多个线的电气连接、便携式计算机盘、硬盘、随机存取存储器(RAM)、只读存储器(ROM)、可擦除可编程只读存储器(EPROM或快闪存储器)、光纤、便捷式紧凑盘只读存储器(CD-ROM)、光学储存设备、磁储存设备、或上述内容的任何合适组合。
根据本公开的一个或多个实施例，示例1提供了一种信息处理方法，包括：获取预设数量的答题者在上一时段关于目标题库的第一答题信息和当前答题者在当前时段关于所述目标题库的第二答题信息，其中，所述当前答题者为所述预设数量的答题者中的一者；以项目反应理论模型中的所述答题者的初始学习能力、题目难度和表现因素分析模型中的所述答题者对所述目标题库所涉及的每一知识点的成功学习效应、失败学习效应为待估计参数，构建贝叶斯网络模型；基于所述贝叶斯网络模型和所述第一答题信息，确定每一所述待估计参数的估计值；根据每一所述待估计参数的估计值和所述第二答题信息，确定所述当前答题者的当前能力。
根据本公开的一个或多个实施例，示例2提供了示例1的方法，所述基于所述贝叶斯网络模型和所述第一答题信息，确定每一所述待估计参数的估计值，包括：基于所述贝叶斯网络模型和所述第一答题信息，采用变分推断方法确定每一所述待估计参数的估计值。
根据本公开的一个或多个实施例，示例3提供了示例1的方法，所述基于所述贝叶斯网络模型和所述第一答题信息，采用变分推断方法确定每一所述待估计参数的估计值，包括：针对每一所述待估计参数，根据该待估计参数在所述上一时段的近似后验分布，确定该待估计参数在所述当前时段的先验分布；基于所述贝叶斯网络模型、所述第一答题信息以及每一所述待估计参数在所述当前时段的先验分布，采用变分推断方法，计算变分下界；以所述变分下界最大化为目标函数进行参数估计，得到每一所述待估计参数的估计值。
根据本公开的一个或多个实施例，示例4提供了示例3的方法，所述根据该待估计参数在所述上一时段的近似后验分布，确定该待估计参数在所述当前时段的先验分布，包括：
根据该待估计参数在所述上一时段的近似后验分布，通过以下公式来确定该待估计参数在所述当前时段的先验分布，包括：
p(parameter)＝(1-decay)*q m(parameter)+decay*p(parameter)
其中，p(parameter)为该待估计参数parameter在所述当前时段的先验分布； q m(parameter)为该待估计参数parameter在所述上一时段的近似后验分布；decay为权重系数。
根据本公开的一个或多个实施例，示例5提供了示例3的方法，所述基于所述贝叶斯网络模型、所述第一答题信息以及每一所述待估计参数在所述当前时段的先验分布，采用变分推断方法，计算变分下界，包括：
基于所述贝叶斯网络模型、所述第一答题信息以及每一所述待估计参数在所述当前时段的先验分布，采用变分推断方法，通过以下公式来计算变分下界：
Figure PCTCN2022133207-appb-000044
其中，ELBO为所述变分下界；
Figure PCTCN2022133207-appb-000045
为由每一所述答题者的初始能力构成的向量；
Figure PCTCN2022133207-appb-000046
为由所述目标题库中每一题目的题目难度构成的向量；
Figure PCTCN2022133207-appb-000047
为由每一所述答题者对所述目标题库所涉及的每一知识点的成功学习效应构成的向量；
Figure PCTCN2022133207-appb-000048
为由每一所述答题者对所述目标题库所涉及的每一知识点的失败学习效应构成的向量；likelihood为基于变分后验分布的重建似然函数，根据所述贝叶斯网络模型和每一所述待估计参数确定；shrink和enhance为超参数；g等于所述预设数量与所述第一答题信息对应的题目的数量的乘积；max等于关于所述目标题库的答题者总数与所述目标题库所包含题目的数量的乘积；
Figure PCTCN2022133207-appb-000049
为对
Figure PCTCN2022133207-appb-000050
的期望；
Figure PCTCN2022133207-appb-000051
为
Figure PCTCN2022133207-appb-000052
的近似后验分布与其先验分布的KL散度；
Figure PCTCN2022133207-appb-000053
为
Figure PCTCN2022133207-appb-000054
的近似后验分布与其先验分布的KL散度；
Figure PCTCN2022133207-appb-000055
为
Figure PCTCN2022133207-appb-000056
的近似后验分布与其先验分布的KL散度；
Figure PCTCN2022133207-appb-000057
为
Figure PCTCN2022133207-appb-000058
的近似后验分布与其先验分布的KL散度。
根据本公开的一个或多个实施例，示例6提供了示例5的方法，shrink＝1。
根据本公开的一个或多个实施例，示例7提供了示例1的方法，所述根据每一所述待估计参数的估计值和所述第二答题信息，确定所述当前答题者的当前能力，包括：根据所述第二答题信息，确定所述当前答题者在所述第二答题信息对应的题目所涉及的每一知识点下学习成功的次数和学习失败的次数；根据每一所述待估计参数的估计值、所述学习成功的次数以及所述学习失败的次数，确定所述当前答题者的当前能力。
根据本公开的一个或多个实施例，示例8提供了示例1的方法，所述贝叶斯网络模型为：
Figure PCTCN2022133207-appb-000059
其中，
Figure PCTCN2022133207-appb-000060
为第i个答题者答对第j个题目的概率，y ij＝1；θ i为所述第i个答题者的初始学习能力；b j为所述第j个题目的题目难度；
Figure PCTCN2022133207-appb-000061
为由所述第i个答题者在所述目标答题库所涉及的每一知识点下学习成功的次数构成的向量；
Figure PCTCN2022133207-appb-000062
为由所述第i个答题者在所述目标答题库所涉及的每一知识点下学习失败的次数构成的向量；
Figure PCTCN2022133207-appb-000063
为所述第i个答题者对所述目标题库所涉及的每一知识点的成功学习效应；
Figure PCTCN2022133207-appb-000064
为所述第i个答题者对所述目标题库中所涉及的每一知识点的失败学习效应；
Figure PCTCN2022133207-appb-000065
为所述第j个题目的所涉及的知识点的分布向量。
根据本公开的一个或多个实施例，示例9提供了示例8的方法，
Figure PCTCN2022133207-appb-000066
为归一化处理后所得的分布向量。
根据本公开的一个或多个实施例，示例10提供了示例1-示例9任一项的方法，所述方法还包括：根据所述当前答题者的当前能力和所述目标题库中候选题目的题目难度以及所述贝叶斯网络模型，确定所述当前答题者答对所述候选题目的概率。
根据本公开的一个或多个实施例，示例11提供了示例10的方法，所述方法还包括：若所述概率满足预设条件，则向所述当前答题者推送所述候选题目。
根据本公开的一个或多个实施例，示例12提供了一种信息处理装置，包括：获取模块，用于获取预设数量的答题者在上一时段关于目标题库的第一答题信息和当前答题者在当前时段关于所述目标题库的第二答题信息，其中，所述当前答题者为所述预设数量的答题者中的一者；构建模块，用于以项目反应理论模型中的所述答题者的初始学习能力、题目难度和表现因素分析模型中的所述答题者对所述目标题库所涉及的每一知识点的成功学习效应、失败学习效应为待估计参数，构建贝叶斯网络模型；第一确定模块，用于基于所述构建模块得到的所述贝叶斯网络模型和所述获取模块获取到的所述第一答题信息，确定每一所述待估计参数的估计值；第二确定模块，用于根据所述第一确定模块确定出的每一所述 待估计参数的估计值和所述获取模块获取到的所述第二答题信息，确定所述当前答题者的当前能力。
根据本公开的一个或多个实施例，示例13提供了一种计算机可读介质，其上存储有计算机程序，该程序被处理装置执行时示例1-11中任一项所述方法的步骤。
根据本公开的一个或多个实施例，示例14提供了一种电子设备，包括：存储装置，其上存储有计算机程序；处理装置，用于执行所述存储装置中的所述计算机程序，以实现示例1-11中任一项所述方法的步骤。
以上描述仅为本公开的较佳实施例以及对所运用技术原理的说明。本领域技术人员应当理解，本公开中所涉及的公开范围，并不限于上述技术特征的特定组合而成的技术方案，同时也应涵盖在不脱离上述公开构思的情况下，由上述技术特征或其等同特征进行任意组合而形成的其它技术方案。例如上述特征与本公开中公开的(但不限于)具有类似功能的技术特征进行互相替换而形成的技术方案。
此外，虽然采用特定次序描绘了各操作，但是这不应当理解为要求这些操作以所示出的特定次序或以顺序次序执行来执行。在一定环境下，多任务和并行处理可能是有利的。同样地，虽然在上面论述中包含了若干具体实现细节，但是这些不应当被解释为对本公开的范围的限制。在单独的实施例的上下文中描述的某些特征还可以组合地实现在单个实施例中。相反地，在单个实施例的上下文中描述的各种特征也可以单独地或以任何合适的子组合的方式实现在多个实施例中。
尽管已经采用特定于结构特征和/或方法逻辑动作的语言描述了本主题，但是应当理解所附权利要求书中所限定的主题未必局限于上面描述的特定特征或动作。相反，上面所描述的特定特征和动作仅仅是实现权利要求书的示例形式。关于上述实施例中的装置，其中各个模块执行操作的具体方式已经在有关该方法的实施例中进行了详细描述，此处将不做详细阐述说明。

Claims (14)

  1. 一种信息处理方法，其特征在于，包括：
    获取预设数量的答题者在上一时段关于目标题库的第一答题信息和当前答题者在当前时段关于所述目标题库的第二答题信息，其中，所述当前答题者为所述预设数量的答题者中的一者；
    以项目反应理论模型中的所述答题者的初始学习能力、题目难度和表现因素分析模型中的所述答题者对所述目标题库所涉及的每一知识点的成功学习效应、失败学习效应为待估计参数，构建贝叶斯网络模型；
    基于所述贝叶斯网络模型和所述第一答题信息，确定每一所述待估计参数的估计值；
    根据每一所述待估计参数的估计值和所述第二答题信息，确定所述当前答题者的当前能力。
  2. 根据权利要求1所述的方法，其特征在于，所述基于所述贝叶斯网络模型和所述第一答题信息，确定每一所述待估计参数的估计值，包括：
    基于所述贝叶斯网络模型和所述第一答题信息，采用变分推断方法确定每一所述待估计参数的估计值。
  3. 根据权利要求2所述的方法，其特征在于，所述基于所述贝叶斯网络模型和所述第一答题信息，采用变分推断方法确定每一所述待估计参数的估计值，包括：
    针对每一所述待估计参数，根据该待估计参数在所述上一时段的近似后验分布，确定该待估计参数在所述当前时段的先验分布；
    基于所述贝叶斯网络模型、所述第一答题信息以及每一所述待估计参数在所述当前时段的先验分布，采用变分推断方法，计算变分下界；
    以所述变分下界最大化为目标函数进行参数估计，得到每一所述待估计参数的估计值。
  4. 根据权利要求3所述的方法，其特征在于，所述根据该待估计参数在所述上一时段的近似后验分布，确定该待估计参数在所述当前时段的先验分布，包括：
    根据该待估计参数在所述上一时段的近似后验分布，通过以下公式来确定该待估计参数在所述当前时段的先验分布，包括：
    p(parameter)＝(1-decay)*q m(parameter)+decay*p(parameter)
    其中，p(parameter)为该待估计参数parameter在所述当前时段的先验分布； q m(parameter)为该待估计参数parameter在所述上一时段的近似后验分布；decay为权重系数。
  5. 根据权利要求3所述的方法，其特征在于，所述基于所述贝叶斯网络模型、所述第一答题信息以及每一所述待估计参数在所述当前时段的先验分布，采用变分推断方法，计算变分下界，包括：
    基于所述贝叶斯网络模型、所述第一答题信息以及每一所述待估计参数在所述当前时段的先验分布，采用变分推断方法，通过以下公式来计算变分下界：
    Figure PCTCN2022133207-appb-100001
    其中，ELBO为所述变分下界；
    Figure PCTCN2022133207-appb-100002
    为由每一所述答题者的初始能力构成的向量；
    Figure PCTCN2022133207-appb-100003
    为由所述目标题库中每一题目的题目难度构成的向量；
    Figure PCTCN2022133207-appb-100004
    为由每一所述答题者对所述目标题库所涉及的每一知识点的成功学习效应构成的向量；
    Figure PCTCN2022133207-appb-100005
    为由每一所述答题者对所述目标题库所涉及的每一知识点的失败学习效应构成的向量；likelihood为基于变分后验分布的重建似然函数，根据所述贝叶斯网络模型和每一所述待估计参数确定；shrink和enhance为超参数；g等于所述预设数量与所述第一答题信息对应的题目的数量的乘积；max等于关于所述目标题库的答题者总数与所述目标题库所包含题目的数量的乘积；
    Figure PCTCN2022133207-appb-100006
    为对
    Figure PCTCN2022133207-appb-100007
    的期望；
    Figure PCTCN2022133207-appb-100008
    为
    Figure PCTCN2022133207-appb-100009
    的近似后验分布与其先验分布的KL散度；
    Figure PCTCN2022133207-appb-100010
    为
    Figure PCTCN2022133207-appb-100011
    的近似后验分布与其先验分布的KL散度；
    Figure PCTCN2022133207-appb-100012
    为
    Figure PCTCN2022133207-appb-100013
    的近似后验分布与其先验分布的KL散度；
    Figure PCTCN2022133207-appb-100014
    为
    Figure PCTCN2022133207-appb-100015
    的近似后验分布与其先验分布的KL散度。
  6. 根据权利要求5所述的方法，其特征在于，shrink＝1。
  7. 根据权利要求1所述的方法，其特征在于，所述根据每一所述待估计参数的估计值和所述第二答题信息，确定所述当前答题者的当前能力，包括：
    根据所述第二答题信息，确定所述当前答题者在所述第二答题信息对应的题目所涉及的每一知识点下学习成功的次数和学习失败的次数；
    根据每一所述待估计参数的估计值、所述学习成功的次数以及所述学习失败的次数， 确定所述当前答题者的当前能力。
  8. 根据权利要求1所述的方法，其特征在于，所述贝叶斯网络模型为：
    Figure PCTCN2022133207-appb-100016
    其中，
    Figure PCTCN2022133207-appb-100017
    为第i个答题者答对第j个题目的概率，y ij＝1；θ i为所述第i个答题者的初始学习能力；b j为所述第j个题目的题目难度；
    Figure PCTCN2022133207-appb-100018
    为由所述第i个答题者在所述目标答题库所涉及的每一知识点下学习成功的次数构成的向量；
    Figure PCTCN2022133207-appb-100019
    为由所述第i个答题者在所述目标答题库所涉及的每一知识点下学习失败的次数构成的向量；
    Figure PCTCN2022133207-appb-100020
    为所述第i个答题者对所述目标题库所涉及的每一知识点的成功学习效应；
    Figure PCTCN2022133207-appb-100021
    为所述第i个答题者对所述目标题库中所涉及的每一知识点的失败学习效应；
    Figure PCTCN2022133207-appb-100022
    为所述第j个题目的所涉及的知识点的分布向量。
  9. 根据权利要求8所述的方法，其特征在于，
    Figure PCTCN2022133207-appb-100023
    为归一化处理后所得的分布向量。
  10. 根据权利要求1-9中任一项所述的方法，其特征在于，所述方法还包括：
    根据所述当前答题者的当前能力和所述目标题库中候选题目的题目难度以及所述贝叶斯网络模型，确定所述当前答题者答对所述候选题目的概率。
  11. 根据权利要求10所述的方法，其特征在于，所述方法还包括：
    若所述概率满足预设条件，则向所述当前答题者推送所述候选题目。
  12. 一种信息处理装置，其特征在于，包括：
    获取模块，用于获取预设数量的答题者在上一时段关于目标题库的第一答题信息和当前答题者在当前时段关于所述目标题库的第二答题信息，其中，所述当前答题者为所述预设数量的答题者中的一者；
    构建模块，用于以项目反应理论模型中的答题者的初始学习能力、题目难度和表现因素分析模型中的答题者对所述目标题库所涉及的每一知识点的成功学习效应、失败学习效应为待估计参数，构建贝叶斯网络模型；
    第一确定模块，用于基于所述构建模块得到的所述贝叶斯网络模型和所述获取模块获取到的所述第一答题信息，确定每一所述待估计参数的估计值；
    第二确定模块，用于根据所述第一确定模块确定出的每一所述待估计参数的估计值和所述获取模块获取到的所述第二答题信息，确定所述当前答题者的当前能力。
  13. 一种计算机可读介质，其上存储有计算机程序，其特征在于，该程序被处理装置执行时实现权利要求1-11中任一项所述方法的步骤。
  14. 一种电子设备，其特征在于，包括：
    存储装置，其上存储有计算机程序；
    处理装置，用于执行所述存储装置中的所述计算机程序，以实现权利要求1-11中任一项所述方法的步骤。
PCT/CN2022/133207 2021-12-21 2022-11-21 信息处理方法、装置、可读存储介质及电子设备 Ceased WO2023116306A1 (zh)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
CN202111574789.8 2021-12-21
CN202111574789.8A CN116341933A (zh) 2021-12-21 2021-12-21 信息处理方法、装置、可读存储介质及电子设备

Publications (1)

Publication Number Publication Date
WO2023116306A1 true WO2023116306A1 (zh) 2023-06-29

Family

ID=86888160

Family Applications (1)

Application Number Title Priority Date Filing Date
PCT/CN2022/133207 Ceased WO2023116306A1 (zh) 2021-12-21 2022-11-21 信息处理方法、装置、可读存储介质及电子设备

Country Status (2)

Country Link
CN (1) CN116341933A (zh)
WO (1) WO2023116306A1 (zh)

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN119273244A (zh) * 2024-12-11 2025-01-07 石家庄铁道大学 一种铁路预制箱梁的性能评价方法和系统
CN120930945A (zh) * 2025-10-10 2025-11-11 浙江海亮科技有限公司 基于智慧教育平台的学习内容确定方法、装置及电子设备

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2001209297A (ja) * 2000-01-24 2001-08-03 Human Resource Research Institute Inc コンピュータ適応型テスト装置、コンピュータ適応型テストシステム、コンピュータ適応型テスト方法およびコンピュータ適応型テストプログラムを格納する記録媒体
CN106599999A (zh) * 2016-12-09 2017-04-26 北京爱论答科技有限公司 少量题目精准探测学生细分薄弱知识点的测评方法及系统
CN108229683A (zh) * 2016-12-22 2018-06-29 上海谦问万答吧云计算科技有限公司 一种基于irt的信息处理方法及装置
US10354544B1 (en) * 2015-02-20 2019-07-16 Snapwiz Inc. Predicting student proficiencies in knowledge components
CN111798138A (zh) * 2020-07-07 2020-10-20 广州视源电子科技股份有限公司 数据处理方法、计算机存储介质及相关设备

Family Cites Families (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
KR101370735B1 (ko) * 2011-09-08 2014-03-07 한양대학교 산학협력단 베이지안 네트워크를 이용한 컴퓨터 기반 적응적 평가 시스템 및 방법
CN108229718B (zh) * 2016-12-22 2020-06-02 北京字节跳动网络技术有限公司 一种信息预测方法及装置
CN108229688B (zh) * 2016-12-22 2020-10-09 北京字节跳动网络技术有限公司 一种基于irt的信息处理方法及装置
CN111582694B (zh) * 2020-04-29 2023-08-08 腾讯科技（深圳）有限公司 一种学习评估方法及装置
CN113033808B (zh) * 2021-03-08 2024-01-19 西北大学 一种基于习题难度和学生能力的深度嵌入知识追踪方法
CN113240323B (zh) * 2021-06-01 2023-08-29 平安科技（深圳）有限公司 基于机器学习的水平评测方法、装置及相关设备

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
JP2001209297A (ja) * 2000-01-24 2001-08-03 Human Resource Research Institute Inc コンピュータ適応型テスト装置、コンピュータ適応型テストシステム、コンピュータ適応型テスト方法およびコンピュータ適応型テストプログラムを格納する記録媒体
US10354544B1 (en) * 2015-02-20 2019-07-16 Snapwiz Inc. Predicting student proficiencies in knowledge components
CN106599999A (zh) * 2016-12-09 2017-04-26 北京爱论答科技有限公司 少量题目精准探测学生细分薄弱知识点的测评方法及系统
CN108229683A (zh) * 2016-12-22 2018-06-29 上海谦问万答吧云计算科技有限公司 一种基于irt的信息处理方法及装置
CN111798138A (zh) * 2020-07-07 2020-10-20 广州视源电子科技股份有限公司 数据处理方法、计算机存储介质及相关设备

Cited By (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN119273244A (zh) * 2024-12-11 2025-01-07 石家庄铁道大学 一种铁路预制箱梁的性能评价方法和系统
CN120930945A (zh) * 2025-10-10 2025-11-11 浙江海亮科技有限公司 基于智慧教育平台的学习内容确定方法、装置及电子设备

Also Published As

Publication number Publication date
CN116341933A (zh) 2023-06-27

Similar Documents

Publication Publication Date Title
CN112966712B (zh) 语言模型训练方法、装置、电子设备和计算机可读介质
US12536990B2 (en) Model training method, speech recognition method, device, medium, and apparatus
CN114398556B (zh) 学习内容的推荐方法、装置、设备及存储介质
CN112650841A (zh) 信息处理方法、装置和电子设备
WO2023116306A1 (zh) 信息处理方法、装置、可读存储介质及电子设备
WO2023274187A1 (zh) 基于自然语言推理的信息处理方法、装置和电子设备
CN117290477B (zh) 一种基于二次检索增强的生成式建筑知识问答方法
US20240331419A1 (en) Training method of object detection model, object detection method, apparatus and device
CN116483891A (zh) 一种信息预测方法、装置、设备和存储介质
CN111353601A (zh) 用于预测模型结构的延时的方法和装置
WO2023226819A1 (zh) 数据匹配方法、装置、可读介质及电子设备
CN113191257B (zh) 笔顺检测方法、装置和电子设备
CN111680754B (zh) 图像分类方法、装置、电子设备及计算机可读存储介质
CN114117219A (zh) 会话推荐方法和推荐模型的训练方法、装置、设备及介质
CN116364066A (zh) 分类模型生成方法、音频分类方法、装置、介质及设备
CN113010784B (zh) 用于生成预测信息的方法、装置、电子设备和介质
CN118673328B (zh) 信息发送方法、装置、电子设备和计算机可读介质
US11470167B2 (en) Method and apparatus for generating information
CN115080724A (zh) 习题推荐方法、装置、电子设备和存储介质
CN115099323B (zh) 内容小组确定方法、装置、介质及电子设备
WO2023125000A1 (zh) 内容输出方法、装置、计算机可读介质及电子设备
CN116010867A (zh) 负样本确定方法、装置、介质及电子设备
CN115619456A (zh) 对象推荐方法、装置、介质以及电子设备
CN116340632A (zh) 对象推荐方法、装置、介质及电子设备
CN115796159A (zh) 题目确定方法、装置、介质及电子设备

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: 22909622

Country of ref document: EP

Kind code of ref document: A1

NENP Non-entry into the national phase

Ref country code: DE

122 Ep: pct application non-entry in european phase

Ref document number: 22909622

Country of ref document: EP

Kind code of ref document: A1