CN116578694A - Disentangled knowledge tracking method, system, device and medium - Google Patents

Disentangled knowledge tracking method, system, device and medium Download PDF

Info

Publication number
CN116578694A
CN116578694A CN202310440461.XA CN202310440461A CN116578694A CN 116578694 A CN116578694 A CN 116578694A CN 202310440461 A CN202310440461 A CN 202310440461A CN 116578694 A CN116578694 A CN 116578694A
Authority
CN
China
Prior art keywords
target object
state
preference
type
answer state
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.)
Granted
Application number
CN202310440461.XA
Other languages
Chinese (zh)
Other versions
CN116578694B (en
Inventor
吴正洋
周金维
汤庸
王冬青
刘远卓
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.)
South China Normal University
Original Assignee
South China Normal University
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 South China Normal University filed Critical South China Normal University
Priority to CN202310440461.XA priority Critical patent/CN116578694B/en
Publication of CN116578694A publication Critical patent/CN116578694A/en
Application granted granted Critical
Publication of CN116578694B publication Critical patent/CN116578694B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/30—Information retrieval; Database structures therefor; File system structures therefor of unstructured textual data
    • G06F16/33—Querying
    • G06F16/335—Filtering based on additional data, e.g. user or group profiles
    • G06F16/337—Profile generation, learning or modification
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/044—Recurrent networks, e.g. Hopfield networks
    • G06N3/0442—Recurrent networks, e.g. Hopfield networks characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/02—Neural networks
    • G06N3/04—Architecture, e.g. interconnection topology
    • G06N3/045—Combinations of networks
    • G06N3/0455—Auto-encoder networks; Encoder-decoder networks
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N5/00—Computing arrangements using knowledge-based models
    • G06N5/02—Knowledge representation; Symbolic representation
    • G06N5/022—Knowledge engineering; Knowledge acquisition
    • 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
    • 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

Landscapes

  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Business, Economics & Management (AREA)
  • General Health & Medical Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Computation (AREA)
  • Health & Medical Sciences (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Tourism & Hospitality (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Molecular Biology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Databases & Information Systems (AREA)
  • Educational Technology (AREA)
  • Educational Administration (AREA)
  • Economics (AREA)
  • Human Resources & Organizations (AREA)
  • Marketing (AREA)
  • Primary Health Care (AREA)
  • Strategic Management (AREA)
  • General Business, Economics & Management (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The application discloses a disentangled knowledge tracking method, a disentangled knowledge tracking system, a disentangled knowledge tracking device and a disentangled knowledge tracking medium. The method comprises the following steps: acquiring a current answer state of a target object, wherein the current answer state comprises a practice problem type and preference of the target object to the practice problem type; the method comprises the steps of performing preference de-entanglement on the exercise types corresponding to the current answer state, and obtaining a reconstructed answer state; and inputting the reconstructed answer state into a gating circulation unit network to obtain the predicted knowledge state of the target object at the future moment. According to the application, the current answer state containing the exercise type is disentangled, so that the preference of the target object to the exercise type in the answer process is obtained, the reconstructed answer state is input into the gating circulation unit network after the reconstructed answer state is obtained based on the preference of the exercise type, and the predicted knowledge state of the target object at the future moment can be obtained through the potential factors of the target object in the answer process, so that the accuracy and the interpretability of the knowledge tracking result are effectively improved.

Description

Disentangled knowledge tracking method, system, device and medium
Technical Field
The application relates to the technical field of education information processing, in particular to a method, a system, a device and a medium for tracking disentangled knowledge.
Background
In the related art, knowledge tracking is a basic and key task for supporting intelligent education service application, and aims to monitor the continuously developed knowledge state of target objects, so that the purposes of providing optimal and adaptive learning experience for each target object, reasonably configuring learning time and improving teaching quality and efficiency are achieved. Knowledge tracking adopts a series of machine learning methods oriented to sequence modeling so as to achieve the purpose of dynamically predicting the knowledge state of a target object by utilizing learning interaction data, and is widely applied to intelligent education systems at present. The realization of knowledge tracking needs to use the interactive data of the target object and the educational resources, however, the prior knowledge tracking method does not consider the influence of potential factors on knowledge mastering in the analysis process, so that the accuracy and the interpretability of the knowledge tracking result are not high.
Disclosure of Invention
The present application aims to solve at least one of the technical problems existing in the prior art. Therefore, the application provides a method, a system, a device and a medium for disentangled knowledge tracking, which can effectively improve the accuracy and the interpretability of knowledge tracking results.
In one aspect, an embodiment of the present application provides a method for tracking disentangled knowledge, including the following steps:
obtaining a current answer state of a target object, wherein the current answer state comprises a practice problem type and a target object pair
Preference of the exercise question type;
the exercise type preference disentanglement corresponding to the current answer state is disentangled in an exercise type preference disentanglement module, and a reconstruction answer state is obtained;
and inputting the reconstruction answer state into a gating circulation unit network to obtain the predicted knowledge state of the target object at the future moment.
In some embodiments, after the obtaining the current answer state of the target object, the method further includes the following steps:
distinguishing the current answer state according to the exercise problem types, and carrying out vector representation on each exercise problem type;
and fusing the corresponding vector of each exercise question type with the current answer state to obtain a new answer state.
In some embodiments, the disentangling the preference of the exercise type corresponding to the current answer state, and obtaining the reconstructed answer state includes:
generating variation distribution of the target object on the preference of the preset type exercise problem by adopting an encoder;
and inputting the variation distribution into a decoder to obtain a reconstruction answering state.
In some embodiments, the generating, with an encoder, a variation distribution of the target object over a preset type of practice problem preference includes:
taking a preset matrix and the new answer state as priori conditions, wherein the preset matrix comprises a matrix formed by all training exercises represented by type vectors;
and obtaining the variation distribution of the target object on the preference of the exercise problem of the preset type by adopting the re-reference skills according to the prior condition.
In some embodiments, the optimization strategy of the practice problem type preference disentanglement module is as follows:
wherein ,representing the practice problemsType preference de-entanglement module loss function, < ->Representing a reconstruction error term, D KL (q φ (z s |X)||p θ (z s ) Represents a KL divergence regularization term, and β represents a factor used to strengthen the KL divergence regularization term.
In some embodiments, the optimization strategy of the gated loop cell network is as follows:
wherein ,representing a loss function of a network of gated loop units, r i Representing the true knowledge state of the current time period, +.>Representing the predicted knowledge state for the current time period.
In another aspect, an embodiment of the present application provides a system for tracking disentangled knowledge, including:
the first module is used for acquiring the current answer state of the target object, wherein the current answer state comprises a practice problem type and preference of the target object to the practice problem type;
the second module is used for disentangling the exercise problem type preference corresponding to the current answer state in the exercise problem type preference disentangling module, and obtaining a reconstructed answer state;
and the third module is used for inputting the reconstruction answer state into a gating circulation unit network to obtain the predicted knowledge state of the target object at the future moment.
In some embodiments, the optimization strategy of the practice problem type preference disentanglement module is as follows:
wherein ,a loss function representing the exercise question type preference de-entanglement module, +.>Representing a reconstruction error term, D KL (q φ (z s |X)||p θ (z s ) Represents a KL divergence regularization term, and β represents a factor used to strengthen the KL divergence regularization term.
In another aspect, an embodiment of the present application provides a device for tracking disentangled knowledge, including:
at least one memory for storing a program;
at least one processor for loading the program to perform the disentangled knowledge tracking method.
In another aspect, an embodiment of the present application provides a storage medium in which a computer-executable program is stored, where the computer-executable program is used to implement the disentangled knowledge tracking method when executed by a processor.
The embodiment of the application provides a disentangled knowledge tracking method, which has the following beneficial effects:
according to the application, the current answer state containing the exercise type is disentangled, so that the preference of the target object to the exercise type in the answer process is obtained, the reconstructed answer state is input into the gating circulation unit network after the reconstructed answer state is obtained based on the preference of the exercise type, and the predicted knowledge of the target object at the future moment can be obtained through the potential factors of the target object in the answer process, so that the accuracy and the interpretability of the knowledge tracking result are effectively improved.
Additional aspects and advantages of the application will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the application.
Drawings
The application is further described with reference to the accompanying drawings and examples, in which:
FIG. 1 is a flowchart of a method for tracking disentangled knowledge according to an embodiment of the present application;
fig. 2 is a schematic diagram of answer status update according to an embodiment of the present application;
FIG. 3 is a schematic illustration of disentanglement according to an embodiment of the present application;
FIG. 4 is a schematic diagram illustrating a process of a prediction model according to an embodiment of the present application.
Detailed Description
Embodiments of the present application are described in detail below, examples of which are illustrated in the accompanying drawings, wherein like or similar reference numerals refer to like or similar elements or elements having like or similar functions throughout. The embodiments described below by referring to the drawings are illustrative only and are not to be construed as limiting the application.
In the description of the present application, the meaning of a number is one or more, the meaning of a number is two or more, and greater than, less than, exceeding, etc. are understood to exclude the present number, and the meaning of a number is understood to include the present number. The description of the first and second is for the purpose of distinguishing between technical features only and should not be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated or implicitly indicating the precedence of the technical features indicated.
In the description of the present application, unless explicitly defined otherwise, terms such as arrangement, connection, etc. should be construed broadly and the specific meaning of the terms in the present application can be reasonably determined by a person skilled in the art in combination with the specific contents of the technical solution.
In the description of the present application, a description of the terms "one embodiment," "some embodiments," "an exemplary embodiment," "an example," "a particular example," or "some examples," etc., means that a particular feature or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the present application. In this specification, schematic representations of the above terms do not necessarily refer to the same embodiments or examples. Furthermore, the particular features or characteristics described may be combined in any suitable manner in any one or more embodiments or examples.
Before proceeding with the description of the specific embodiments, the terms involved in the embodiments of the present application are explained as follows:
and (3) splicing and entanglement: by transforming the data in the characterization space, where the original latent factors are entangled with each other, into a new characterization space, the different latent factors in this space are separated from each other.
Knowledge tracking: the method is to predict the mastering level of a target object on a specific knowledge concept or the reaction to a specific learning interaction item in the next time step according to the historical answer record of the target object.
Knowledge tracking is a basic and key task supporting intelligent education service application, and aims to monitor the knowledge state of the continuous development of target objects, so that the purposes of providing optimal and adaptive learning experience for each target object, reasonably configuring learning time and improving teaching quality and efficiency are supported. Knowledge tracking adopts a series of machine learning methods oriented to sequence modeling so as to achieve the purpose of dynamically predicting the knowledge state of a target object by utilizing learning interaction data, and is widely applied to intelligent education systems at present.
The implementation of knowledge tracking requires interaction data of the target object with the educational resources, and the entangled potential factors present therein influence the outcome of the interaction. The potential factors of the high entanglement include not only knowledge point mastering level of the target object, but also knowledge point difficulty feeling, learning resource adaptability and the like of the target object. Current research on knowledge tracking only relates to whether the target object can answer the practice problem of the next time step correctly, and potential factors for determining whether the answer result of the target object is correct are not quantitatively characterized. Therefore, the research on knowledge tracking of the disentangled characterization is helpful to reveal the microcosmic learning psychology and behavior of the target object, and can provide richer data support for personalized learning.
Referring to fig. 1, an embodiment of the present application provides a method for tracking disentangled knowledge, and the method of the present embodiment may be applied to a processor, a server, or a cloud end corresponding to an educational platform. During application, the method of the present embodiment includes, but is not limited to, the following steps:
step S110, obtaining a current answer state of a target object, wherein the current answer state comprises a practice problem type and preference of the target object to the practice problem type;
step S120, in the exercise type preference disentangling module, the exercise type preference corresponding to the current answering state is disentangled, and a reconstructed answering state is obtained;
and step S130, inputting the reconstructed answer state into a gating circulation unit network to obtain the predicted knowledge state of the target object at the future moment.
In the embodiment of the application, after the current answer state of the target object is obtained, the current answer state is distinguished according to the exercise problem types, and each exercise problem type is subjected to vector representation; and then fusing the corresponding vector of each exercise question type with the current answer state to obtain a new answer state. Illustratively, the true answer state r of the target object in the t time period 1 :r t The distinction is made by the type of practice problem, as shown in FIG. 2. Wherein each type is represented by a separate One-hot vector, such as a representation vector of the practice problem belonging to the problem type 1: x is x 1 ＝[1,0,0]The expression vector belonging to the topic 2: x is x 2 ＝[0,1,0]. Then fusing the exercise problem type vector and the answering state, and updating the exercise problem type vector and the answering state into a new answering state through a formula (1)
In the present embodiment, after completion of updating of the answer state, it is assumed that the observed data is generated by the distribution represented by the formula (2) (for the target object s):
p θ (a s )＝∫p θ (a s |z s ,X)p θ (z s )dz s formula (2)
wherein ,
in the formula (1) and the formula (2), a s Is answer state, X is exercise problem type matrix, z s Is a preferred disentanglement characterization of the exercise problem type. The following are explained separately:
a s,i representing answer state of target object s, a s,i =1 means that the target object s answers the ith exercise question, and a s,i =0 indicates that the target object s has been wrong. For convenience, the present embodiment uses a s To represent all answer states of the target object s.
The practice problem type is represented by a one-hot vector: x is x i ＝[x i,1 ,x i,2 ,x i,3 ,L,x i,C ]If the type of exercise problem i is c, x i,C =1, otherwise 0;represents a matrix of all exercises represented by type vectors (one-hot vectors), where |E| represents the total number of exercises.
z s Is a factorization vector of the vector,representing a total of C types of exercises, < ->Representing the preference of the target object for type c exercise problems,/for the target object>Is a disentangled characterization of the target object' S preference for the type of exercise problem, where |S| represents the total number of target object people.
The implementation of the disentanglement of the preference factors of the exercise problem type is to promote the generationOnly the c-th type of preference of the target object with respect to the practice problem is captured. As shown in fig. 3, when performing the disentanglement, a multi-set conditional variation self-encoder (CVAE) implementation is employed. Specifically, for each CVAE, the qθ (z) generated by the encoder (encoder) is first used s |a s X) approximately represents the preference of the target object for the practice problem, i.e. in a preset matrix X and the new answer state +.>Obtaining the preference of the target object to the preset type exercise problem by adopting the re-parametric skill for the prior condition>Wherein the preset matrix comprises a matrix of all exercises represented by type vectors; then using decoder to obtain the reconstruction answering state +.>Approximation a s And by calculating a s To optimize the exercise problem type preference de-entanglement module. Specifically, the optimization strategy of the exercise problem type preference disentanglement module is shown in the formula (3):
wherein ,loss function representing preference de-entanglement module for exercise problem type, +.>Representing a reconstruction error term, D KL (q φ (z s |X)||p θ (z s ) Represents a KL divergence regularization term, and β represents a factor used to strengthen the KL divergence regularization term. In this embodiment, to encourage z s The independence between the values of the KL divergence regularization term can be reduced, and the regularization term can be enhanced by a factor beta > 1.
After obtaining the reconstruction answer state, the embodiment predicts the predicted knowledge state of the target object at the future time through a gate-controlled loop unit (GRU) network. Illustratively, as shown in fig. 4,the reconstruction answering state of the target object s at the moment of 1-t is input with the sequence +.>After the GRU network is reached, outputting a predicted knowledge state vector of the target object at the moment of 2-t+1>Wherein the v-th element, i.e.)>Is the predicted value of the grasping level of the knowledge point v at the next time step of t for the target object s. The optimization strategy of the gating loop cell network is shown in formula (4):
wherein ,representing a loss function of a network of gated loop units, r i Representing the true knowledge state of the current time period, +.>Representing the predictive knowledge state of the current time period, r i Is the real knowledge state of the target object at the time of 2-t+1. The optimization strategy of the problem type preference entanglement-solving module is synthesized, and the overall optimization strategy is obtained as shown in a formula (5):
in some embodiments, the data set is: assist09, assist12 and ednet, students are taken as target objects, and experimental information is as follows: the specific experimental steps include the following steps:
step one, processing a data set: obtaining a practice problem-type matrix X, X is a matrix of dimensions |E|×|C| and X if the type of practice problem E is C e,c Otherwise, 0, each row of the preset matrix represents a training problem type characterization vector. The answer state of the target object is a vector with a dimension of 2|K |, if the exercise question contains the kth knowledge point, the kth element of the vector is 1, and the other elements are 0; if the practice question is answered correctly, then the |K|+k bit element of the vector is 1 and the other elements are 0.
Generating a fusion answer state: and fusing the training problem type characterization vector represented by each row in the X with the corresponding answering state to obtain a new answering state.
Step three, inputting the new answer state vector into the encoder network to generate n-dimensional hidden stateAnd then->Inputting decoder, combining with x corresponding to practice problem s Generating a reconstructed answer state vector +.> To practice withThe problem type preference factor disentangles the token vector.
Step five, willInputting GRU network to obtain knowledge state prediction +.>
Experiments show that the embodiment can provide more characterization data for downstream tasks, so that the accuracy of knowledge tracking results can be effectively improved, and better robustness and interpretation can be achieved.
The embodiment of the application provides a disentangled knowledge tracking system, which comprises the following components:
the first module is used for acquiring the current answer state of the target object, wherein the current answer state comprises a practice problem type and preference of the target object to the practice problem type;
the second module is used for preferentially disentangling the exercise problem type corresponding to the current answer state in the exercise problem type preference disentangling module and obtaining a reconstructed answer state;
and the third module is used for inputting the reconstruction answer state into the gating circulation unit network to obtain the predicted knowledge state of the target object at the future moment.
In some embodiments, the optimization strategy of the practice problem type preference disentanglement module is as follows:
wherein ,loss function representing preference de-entanglement module for exercise problem type, +.>Representing a reconstruction error term, D KL (q φ (z s |X)||p θ (z s ) Represents a KL divergence regularization term, and β represents a factor used to strengthen the KL divergence regularization term.
The content of the method embodiment of the application is suitable for the system embodiment, the specific function of the system embodiment is the same as that of the method embodiment, and the achieved beneficial effects are the same as those of the method.
The embodiment of the application provides a disentangled knowledge tracking device, which comprises:
at least one memory for storing a program;
at least one processor for loading the program to perform the disentangled knowledge tracking method shown in fig. 1.
The content of the method embodiment of the application is suitable for the device embodiment, the specific function of the device embodiment is the same as that of the method embodiment, and the achieved beneficial effects are the same as those of the method.
An embodiment of the present application provides a storage medium in which a computer-executable program is stored, which when executed by a processor is configured to implement the disentangled knowledge tracking method shown in fig. 1.
The content of the method embodiment of the application is applicable to the storage medium embodiment, the specific function of the storage medium embodiment is the same as that of the method embodiment, and the achieved beneficial effects are the same as those of the method.
Embodiments of the present application also provide a computer program product or computer program comprising computer instructions stored in a computer readable storage medium. The computer instructions may be read from a computer-readable storage medium by a processor of a computer device, and executed by the processor, to cause the computer device to perform the disentangled knowledge tracking method shown in fig. 1.
The embodiments of the present application have been described in detail with reference to the accompanying drawings, but the present application is not limited to the above embodiments, and various changes can be made within the knowledge of one of ordinary skill in the art without departing from the spirit of the present application. Furthermore, embodiments of the application and features of the embodiments may be combined with each other without conflict.

Claims (10)

1. A method of disentangled knowledge tracking, comprising the steps of:
obtaining a current answer state of a target object, wherein the current answer state comprises a practice problem type and preference of the target object to the practice problem type;
the exercise type preference disentanglement corresponding to the current answer state is disentangled in an exercise type preference disentanglement module, and a reconstruction answer state is obtained;
and inputting the reconstruction answer state into a gating circulation unit network to obtain the predicted knowledge state of the target object at the future moment.
2. The method of claim 1, further comprising the steps of, after the obtaining the current answer state of the target object:
distinguishing the current answer state according to the exercise problem types, and carrying out vector representation on each exercise problem type;
and fusing the corresponding vector of each exercise question type with the current answer state to obtain a new answer state.
3. The method of claim 2, wherein the performing the disentangling of the exercise type preference corresponding to the current answer state and obtaining the reconstructed answer state includes:
generating variation distribution of the target object on the preference of the preset type exercise problem by adopting an encoder;
and inputting the variation distribution into a decoder to obtain a reconstruction answering state.
4. A method of de-entanglement knowledge tracking according to claim 3, wherein said generating, with an encoder, a distribution of variations of the target object's preference for a pre-set type of exercises, comprises:
taking a preset matrix and the new answer state as priori conditions, wherein the preset matrix comprises a matrix formed by all training exercises represented by type vectors;
and obtaining the variation distribution of the target object on the preference of the exercise problem of the preset type by adopting the re-reference skills according to the prior condition.
5. The method of claim 1, wherein the optimization strategy of the practice problem type preference disentanglement module is as follows:
wherein ,a loss function representing the exercise question type preference de-entanglement module, +.>Representing a reconstruction error term, D KL (q φ (z s |X)||p θ (z s ) Represents a KL divergence regularization term, and β represents a factor used to strengthen the KL divergence regularization term.
6. The method of claim 5, wherein the optimization strategy of the gated loop unit network is as follows:
wherein ,representing a loss function of a network of gated loop units, r i Representing the true knowledge state of the current time period, +.>Representing the predicted knowledge state for the current time period.
7. A disentangled knowledge tracking system, comprising:
the first module is used for acquiring the current answer state of the target object, wherein the current answer state comprises a practice problem type and preference of the target object to the practice problem type;
the second module is used for disentangling the exercise problem type preference corresponding to the current answer state in the exercise problem type preference disentangling module, and obtaining a reconstructed answer state;
and the third module is used for inputting the reconstruction answer state into a gating circulation unit network to obtain the predicted knowledge state of the target object at the future moment.
8. The system of claim 7, wherein the optimization strategy of the practice problem type preference de-entanglement module is as follows:
wherein ,a loss function representing the exercise question type preference de-entanglement module, +.>Representing a reconstruction error term, D KL (q φ (z s |X)||p θ (z s ) Represents a KL divergence regularization term, and β represents a factor used to strengthen the KL divergence regularization term.
9. A de-entanglement knowledge tracking device, comprising:
at least one memory for storing a program;
at least one processor for loading the program to perform the disentanglement knowledge tracking method of any one of claims 1-6.
10. A storage medium having stored therein a computer executable program for implementing the disentangled knowledge tracking method of any one of claims 1-6 when executed by a processor.
CN202310440461.XA 2023-04-21 2023-04-21 Disentangled knowledge tracking method, system, device and medium Active CN116578694B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202310440461.XA CN116578694B (en) 2023-04-21 2023-04-21 Disentangled knowledge tracking method, system, device and medium

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202310440461.XA CN116578694B (en) 2023-04-21 2023-04-21 Disentangled knowledge tracking method, system, device and medium

Publications (2)

Publication Number Publication Date
CN116578694A true CN116578694A (en) 2023-08-11
CN116578694B CN116578694B (en) 2024-01-09

Family

ID=87535041

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202310440461.XA Active CN116578694B (en) 2023-04-21 2023-04-21 Disentangled knowledge tracking method, system, device and medium

Country Status (1)

Country Link
CN (1) CN116578694B (en)

Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112116092A (en) * 2020-08-11 2020-12-22 浙江师范大学 Interpretable knowledge level tracking method, system and storage medium
CN113361685A (en) * 2021-05-25 2021-09-07 华中师范大学 Knowledge tracking method and system based on learner knowledge state evolution expression
CN113724110A (en) * 2021-08-27 2021-11-30 中国海洋大学 Interpretable depth knowledge tracking method and system and application thereof
CN114372137A (en) * 2022-01-11 2022-04-19 重庆邮电大学 Dynamic perception test question recommendation method and system integrating depth knowledge tracking
CN114971066A (en) * 2022-06-16 2022-08-30 兰州理工大学 Knowledge tracking method and system integrating forgetting factor and learning ability
US20220357177A1 (en) * 2021-05-10 2022-11-10 André SKUPIN Knowledge Space Analytics
CN115329959A (en) * 2022-07-19 2022-11-11 华中师范大学 Learning target recommendation method based on double-flow knowledge embedded network

Patent Citations (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112116092A (en) * 2020-08-11 2020-12-22 浙江师范大学 Interpretable knowledge level tracking method, system and storage medium
US20220357177A1 (en) * 2021-05-10 2022-11-10 André SKUPIN Knowledge Space Analytics
CN113361685A (en) * 2021-05-25 2021-09-07 华中师范大学 Knowledge tracking method and system based on learner knowledge state evolution expression
CN113724110A (en) * 2021-08-27 2021-11-30 中国海洋大学 Interpretable depth knowledge tracking method and system and application thereof
CN114372137A (en) * 2022-01-11 2022-04-19 重庆邮电大学 Dynamic perception test question recommendation method and system integrating depth knowledge tracking
CN114971066A (en) * 2022-06-16 2022-08-30 兰州理工大学 Knowledge tracking method and system integrating forgetting factor and learning ability
CN115329959A (en) * 2022-07-19 2022-11-11 华中师范大学 Learning target recommendation method based on double-flow knowledge embedded network

Non-Patent Citations (1)

* Cited by examiner, † Cited by third party
Title
吴正洋: ""个性化学习推荐研究综述"", 《计算机科学与探索》, vol. 16, no. 1, pages 21 - 40 *

Also Published As

Publication number Publication date
CN116578694B (en) 2024-01-09

Similar Documents

Publication Publication Date Title
CN111897941B (en) Dialogue generation method, network training method, device, storage medium and equipment
CN111506814B (en) Sequence recommendation method based on variational self-attention network
CN110941723A (en) Method, system and storage medium for constructing knowledge graph
Amershi et al. Automatic recognition of learner groups in exploratory learning environments
CN113031983B (en) Intelligent software upgrading method and device based on deep reinforcement learning
CN111680216B (en) Test question recommendation method, system, medium and equipment
CN117033802B (en) Teaching subject pushing method and system based on AI assistance
KR102405838B1 (en) Apparatus for providing personalized content by order of questions and method therefor
CN115545160A (en) Knowledge tracking method and system based on multi-learning behavior cooperation
CN111651792A (en) Risk detection and model enhancement method and device in multi-party collaborative learning
CN115329959A (en) Learning target recommendation method based on double-flow knowledge embedded network
JP3258636B2 (en) Learning support device, learning support method, and recording medium
CN116578694B (en) Disentangled knowledge tracking method, system, device and medium
CN112818100A (en) Knowledge tracking method and system fusing question difficulty
JP6832410B1 (en) Learning effect estimation device, learning effect estimation method, program
CN117215789A (en) Resource allocation method and device for data processing task and computer equipment
CN113254836B (en) Intelligent child-care knowledge point information pushing method and system and cloud platform
CN114971066A (en) Knowledge tracking method and system integrating forgetting factor and learning ability
CN112766513B (en) Knowledge tracking method and system for memory collaboration
KR102329611B1 (en) Pre-training modeling system and method for predicting educational factors
WO2021250725A1 (en) System, device, method, and program for personalized e-learning
Cabo Use of Machine Learning to Identify Predictors of Student Performance in Writing Viable Computer Programs with Repetition Loops and Methods
CN114723590B (en) Group-oriented knowledge tracking method, system, device and storage medium
CN110533189A (en) Machine learning system
US20230245578A1 (en) Method, system and non-transitory computer-readable recording medium for providing information on user&#39;s conceptual understanding

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant