CN108614865A - Method is recommended in individualized learning based on deeply study - Google Patents
Method is recommended in individualized learning based on deeply study Download PDFInfo
- Publication number
- CN108614865A CN108614865A CN201810307140.1A CN201810307140A CN108614865A CN 108614865 A CN108614865 A CN 108614865A CN 201810307140 A CN201810307140 A CN 201810307140A CN 108614865 A CN108614865 A CN 108614865A
- Authority
- CN
- China
- Prior art keywords
- topic
- user
- knowledge point
- learning
- difficulty
- 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
Links
- 230000013016 learning Effects 0.000 title claims abstract description 69
- 238000000034 method Methods 0.000 title claims abstract description 28
- 230000006399 behavior Effects 0.000 claims abstract description 26
- 230000003542 behavioural effect Effects 0.000 claims description 7
- 238000010276 construction Methods 0.000 claims description 2
- 238000013499 data model Methods 0.000 claims description 2
- 230000008569 process Effects 0.000 description 7
- 230000000694 effects Effects 0.000 description 2
- 241001269238 Data Species 0.000 description 1
- 230000009471 action Effects 0.000 description 1
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000019771 cognition Effects 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 235000013399 edible fruits Nutrition 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000005065 mining Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- 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
Landscapes
- Business, Economics & Management (AREA)
- Engineering & Computer Science (AREA)
- Educational Administration (AREA)
- Educational Technology (AREA)
- Strategic Management (AREA)
- Tourism & Hospitality (AREA)
- Health & Medical Sciences (AREA)
- Economics (AREA)
- General Health & Medical Sciences (AREA)
- Human Resources & Organizations (AREA)
- Marketing (AREA)
- Primary Health Care (AREA)
- Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- Information Retrieval, Db Structures And Fs Structures Therefor (AREA)
Abstract
The invention discloses a kind of individualized learnings based on deeply study to recommend method, includes the following steps:The difficulty attribute for defining knowledge point and topic builds knowledge point network figure according to the relationship between knowledge point;The relationship under knowledge point between topic is determined according to the relationship between knowledge point, builds topic network;According to user behavior data, obtained in topic network for the subgraph under designated user's current state, as study boundary;And then deeply learning algorithm is used, it is recorded and is modeled using user's history, training obtains how choosing cut set strategy in the subgraph under user's current state.The method of the present invention can be intelligently that user recommends best topic, save user's learning time, its learning efficiency is made to improve, learning experience is promoted.
Description
Technical field
The present invention relates to individualized learnings to recommend research field, more particularly to a kind of personalization based on deeply study
Learn recommendation method.
Background technology
It is released along with present more and more Internet education platforms, Network Learning Resource has also obtained great rich
Richness, user can learn whenever and wherever possible, while can also be tested at any time, it is such experience for users convenience do not say and
Analogy.But student largely affects the effect of study, nothing in the difference of individual difference, interest, learning style etc.
It is relatively low that differentiation teaching presence learning efficiency, it is difficult to the case where accomplishing to teach students in accordance with their aptitude.American Psychologist's Noel's base of a fruit is strange
(Noel Tichy) it is proposed that the most perfect condition of a self-study be often in study things have challenge appropriate
" learning region (stretch zone) ".So, the learning behavior of user is excavated, find the topic of " learning region " to
Recommended to have a very important significance the learning process of user in family.In addition, due to Internet education learning platform
It is universal, can present rapidly the education resource of most suitable user cognition level, find in a large number of homework exercises for primary and middle school students topic of most suitable student into
Row personalized recommendation just seem more it is important.Platform is universal and user volume increase is also accumulated from more and more users' network science
The behavioral data of habit.The behavioral data for how utilizing user is recommended to be suitble to the study teaching material or topic of oneself to user, to change
Learning experience into user has become the hot spot studied at present.
It is the behavioral data for active user to have had correlative study at present, is modeled according to these behavioral datas,
Recommend personalized topic for user, technical solution is mainly there are two aspect, and one is to be based on commending system, the other is based on using
Family behavior patterns mining.The two, which exists, easily ignores the information contained in user behavior, and resource utilization is not high, recommends output unstable
The problems such as fixed and precision is relatively low.
Invention content
It is an object of the invention to overcome the prior art that can not carry out personalized recommendation, provide a kind of based on depth
It can be intelligently that user recommends " learning region " topic that method, this method are recommended in the individualized learning of intensified learning, save user
Learning time makes learning efficiency improve, and learning experience is promoted.
The purpose of the present invention is realized by the following technical solution:Individualized learning recommendation side based on deeply study
Method includes the following steps:
(1) the difficulty attribute for defining knowledge point and topic builds knowledge point network figure according to the relationship between knowledge point;
(2) relationship under knowledge point between topic is determined according to the relationship between knowledge point, builds topic network;
(3) according to user behavior data, the subgraph under designated user's current state is obtained in topic network;
(4) deeply learning algorithm is used, is recorded and is modeled using user's history, training obtains under user's current state
Subgraph in how to choose cut set i.e. user " learning region " strategy.
Preferably, in step (1), the difficulty attribute value of knowledge point relies on expert or user data models to define, topic
Difficulty attribute expert or user data are relied on according to the difficulty of the difficulty attribute value and topic itself of the knowledge point where topic
It models to define.
Preferably, in step (1), knowledge point network figure refers to according to knowledge point as node, the difficulty attribute of knowledge point
It is worth the difficulty attribute value as node, establishes even side according to the relationship between knowledge point, relation between knowledge points degree is as even side
Weighted value, relational dependence expert or user data modeling.
Preferably, in step (2), topic network refers to according to the topic under knowledge point as node, the difficulty of topic
Item difficulty attribute value of the attribute value as node, difficulty of knowledge points of the difficulty of knowledge points attribute value as node where topic
Attribute value, according to have even while knowledge point under relationship is established even between topic between topic under relationship and same knowledge point while, topic
Between degree of relationship as even side weighted value.
Preferably, in step (3), the construction method of the subgraph under user's current state is:According to user behavior data,
Find the forward or a backward node of the topic node answered according to user behavior data in topic network, the node found and
With its company while, even while weight constitute user's current state under subgraph.
Preferably, in step (4), a deeply learning model is built, the history of user is answered into record as depth
The state of intensified learning model, according to the Strategies of Topic of the difficulty attribute of the subgraph interior joint under user's current state as action
Collection, the positive exact figures answered according to user determine return value, and deeply learning training, instruction are carried out by a certain amount of answer process
Practice and chooses cut set strategy from the subgraph under user's current state, the topic of " learning region " during cut set, that is, individualized learning is recommended.
Compared with prior art, the present invention having the following advantages that and advantageous effect:
1, the present invention is modeled according to user's learning behavior, is learnt user behavior using deeply learning algorithm, is used
" learning region " at family so that the topic that final recommended user answers, which reaches, is not only suitble to user capability difficulty, but also can make user's
Answering has preferable accuracy rate, to reach the efficient the destination of study of user.
2, it is based on complex network figure in the present invention, is found in topic network according to user's history behavior and user's history
The associated topic of behavior, can make full use of user's history behavioural information, excavate the effective information of user behavior.
3, the present invention when building deeply learning model, uses user behavior during deeply learning training
Series Modeling is carried out deeply learning training by a certain amount of answer, is all answered recently using user after each answer
Record is used as state, is updated after answering every time, the state chosen in this way can effectively embody the personalization of user.
4, the method for the present invention can intelligent " learning region " for choosing user, that is, utilize deeply learning algorithm, study to give
User carries out the strategy that personalized topic is recommended, and reaches and intelligently recommends topic, i.e. topic in " learning region " range to user
Mesh, allow user experience more preferably.
Description of the drawings
Fig. 1 is the principle schematic of the present embodiment method, and (a) indicates knowledge point network graph structure, (b) indicate same knowledge point
Lower topic network structure, (c) topic network structure under related knowledge point (d) indicate that the user behavior data chosen is being inscribed
Structure in mesh network (e) indicates the forward, backward node that the topic node is found in topic network, (f) indicates to use
The structure of subgraph under the current state of family (g) indicates obtained " learning region " topic.
Fig. 2 is the procedure chart when present invention carries out deeply learning training.
Fig. 3 is the relationship between data, operation etc. in the present embodiment method implementation process.
Specific implementation mode
Present invention will now be described in further detail with reference to the embodiments and the accompanying drawings, but embodiments of the present invention are unlimited
In this.
Embodiment
It is to use complex web to recommend method, step the present embodiment provides a kind of individualized learning based on deeply study
Network figure indicates that the relationship that the relationship between knowledge point is formed between knowledge point network figure and topic forms topic network, passes through user's row
The subgraph under user's current state of the user behavior in topic network is obtained for data, searching " learning region " problem is converted
To find cut set problem in the subgraph under user's current state, using deeply learning algorithm, user behavior data is modeled,
Training obtains choosing the strategy of cut set from the subgraph under user's current state, to realize that carrying out individualized learning to user pushes away
It recommends.Each step is specifically described below in conjunction with the accompanying drawings.
One, the difficulty attribute for defining knowledge point and topic builds knowledge point network figure according to the relationship between knowledge point.
In practical operation, the difficulty attribute of knowledge point and topic can be advance according to the experience with students of oneself by senior teacher
Set or generated using user's history data, the difficulty attribute of topic can be according to the difficulty of the knowledge point in conjunction with where topic
The difficulty of degree attribute value and topic itself relies on expert or user data modeling to define.
In the knowledge point network figure of structure, according to knowledge point as node, the difficulty attribute value of knowledge point is as node
Difficulty attribute value establishes even side, weighted value of the relation between knowledge points degree as even side according to the relationship between knowledge point.Structure
Knowledge point network graph structure is referring to Fig. 1 (a).
Two, the relationship under knowledge point between topic is determined according to the relationship between knowledge point, builds topic network.
In the present embodiment, topic network refers to according to the topic under knowledge point as node, the difficulty attribute value of topic
As the item difficulty attribute value of node, difficulty of knowledge points attribute of the difficulty of knowledge points attribute value as node where topic
Value, according to have even while knowledge point under relationship is established even between topic between topic under relationship and same knowledge point while, close between topic
It is weighted value of the degree as even side.The structure of structure is indicated referring to Fig. 1 (b), Fig. 1 (c), Fig. 1 (b) with topic net under knowledge point
Network structure, topic network structure under the related knowledge points Fig. 1 (c).
Three, according to user behavior data, the subgraph under user's current state is obtained in topic network.
(1) user behavior data is obtained from user behavior library first, chooses nearest answer record, i.e. user's current state
Behavioral data, the structure in topic network is referring to Fig. 1 (d);
(2) the forward, backward section that topic node of answering is found from topic network and then is recorded according to nearest answer
Point finds the rear to node of the topic node if topic is answered correctly specifically, history is answered in topic network, if
History topic of answering is answered mistake, then the forward direction node of the topic node is found in topic network, structure is referring to Fig. 1 (e);
(3) then by the node found and with its company while, even while weight collectively form the son under user's current state
Figure, structure is referring to Fig. 1 (f).
Four, it using deeply learning algorithm, is recorded in conjunction with user's history, training obtains the son under user's current state
How cut set strategy is chosen in figure.
Referring to Fig. 2, the process learnt using deeply learning algorithm is as follows:
(1) deeply study initial model is first built, deeply study is carried out by a certain amount of user's answer
Training answers the history of user state of the record as deeply learning model in training process, by user's current state
Under subgraph interior joint difficulty attribute Strategies of Topic as behavior aggregate, the positive exact figures answered according to user determine return value;
(2) feed back " learning region " topic according to deeply learning model, user obtained after answering strategy return value,
New answer record, the subgraph under new user's current state, former answer record continually enter into depth intensified learning model into
Row training;
(3) final training obtains choosing the strategy of cut set from the subgraph under user's current state, to realize to user
Individualized learning recommendation is carried out, shown in obtained " learning region " topic such as Fig. 1 (g).
Referring to Fig. 3, in method implementation process, user, which answers, is continuously available new historical record, not according to these records
It is disconnected to be input to deeply learning model and be trained, according to training result, new " learning region " topic is obtained, i.e., is worked as from user
The new topic filtered out in subgraph under preceding state, user continue to answer, and by the above process, obtain the best plan for choosing topic
Slightly, realize that individualized learning is recommended.
The neural network that the method for the present invention is learnt based on deeply can adapt to most users by largely training
Behavior models user behavior, is set a question strategy according to user behavior using the study of deeply learning art, to realize
Individualized learning recommendation is carried out according to user, can reach the personalized purpose set a question in the application.
The above embodiment is a preferred embodiment of the present invention, but embodiments of the present invention are not by above-described embodiment
Limitation, it is other it is any without departing from the spirit and principles of the present invention made by changes, modifications, substitutions, combinations, simplifications,
Equivalent substitute mode is should be, is included within the scope of the present invention.
Claims (6)
1. method is recommended in the individualized learning based on deeply study, which is characterized in that include the following steps:
(1) the difficulty attribute for defining knowledge point and topic builds knowledge point network figure according to the relationship between knowledge point;
(2) relationship under knowledge point between topic is determined according to the relationship between knowledge point, builds topic network;
(3) according to user behavior data, the subgraph under designated user's current state is obtained in topic network;
(4) deeply learning algorithm is used, is recorded and is modeled using user's history, training obtains the son under user's current state
How cut set is chosen in figure.
2. method is recommended in the individualized learning according to claim 1 based on deeply study, which is characterized in that step
(1) in, the difficulty attribute value of knowledge point relies on expert or user data modeling to define, and the difficulty attribute of topic is according to topic institute
The difficulty attribute value of knowledge point and the difficulty of topic itself rely on expert or user data modeling defines.
3. method is recommended in the individualized learning according to claim 1 based on deeply study, which is characterized in that step
(1) in, knowledge point network figure refers to according to knowledge point as node, the difficulty attribute of the difficulty attribute value of knowledge point as node
Value establishes even side according to the relationship between knowledge point, relation between knowledge points degree as the weighted value for connecting side, relational dependence expert or
User data models to define.
4. method is recommended in the individualized learning according to claim 1 based on deeply study, which is characterized in that step
(2) in, topic network refers to according to the topic under knowledge point as node, the topic of the difficulty attribute value of topic as node
Difficulty attribute value, difficulty of knowledge points attribute value of the difficulty of knowledge points attribute value as node where topic connect side according to having
Relationship establishes even side between topic under knowledge point between topic under relationship and same knowledge point, and degree of relationship is as company side between topic
Weighted value.
5. method is recommended in the individualized learning according to claim 1 based on deeply study, which is characterized in that step
(3) in, the construction method of the subgraph under user's current state is:According to user behavior data, according to use in topic network
Family behavioral data finds the forward or a backward node for the topic node answered, the node found and with its company while, even while
Weight constitutes the subgraph under user's current state.
6. method is recommended in the individualized learning according to claim 1 based on deeply study, which is characterized in that step
(4) in, a deeply learning model is built, the history of user is answered state of the record as deeply learning model,
According to the Strategies of Topic of the difficulty attribute of the subgraph interior joint under user's current state as behavior aggregate, answer just according to user
Exact figures determine return value, carry out deeply learning training by a certain amount of answer, train from the son under user's current state
Cut set strategy is chosen in figure.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810307140.1A CN108614865B (en) | 2018-04-08 | 2018-04-08 | Personalized learning recommendation method based on deep reinforcement learning |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810307140.1A CN108614865B (en) | 2018-04-08 | 2018-04-08 | Personalized learning recommendation method based on deep reinforcement learning |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108614865A true CN108614865A (en) | 2018-10-02 |
CN108614865B CN108614865B (en) | 2020-12-11 |
Family
ID=63659587
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810307140.1A Active CN108614865B (en) | 2018-04-08 | 2018-04-08 | Personalized learning recommendation method based on deep reinforcement learning |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108614865B (en) |
Cited By (12)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109255994A (en) * | 2018-10-26 | 2019-01-22 | 北京智能优学科技有限公司 | A kind of foreign language teaching adaptive learning method and computer readable storage medium |
CN109543840A (en) * | 2018-11-09 | 2019-03-29 | 北京理工大学 | A kind of Dynamic recommendation design method based on multidimensional classification intensified learning |
CN109859554A (en) * | 2019-03-29 | 2019-06-07 | 上海乂学教育科技有限公司 | Adaptive english vocabulary learning classification pushes away topic device and computer learning system |
CN110009956A (en) * | 2019-04-22 | 2019-07-12 | 上海乂学教育科技有限公司 | English Grammar adaptive learning method and learning device |
CN110223553A (en) * | 2019-05-20 | 2019-09-10 | 北京师范大学 | A kind of prediction technique and system of answering information |
CN110288878A (en) * | 2019-07-01 | 2019-09-27 | 科大讯飞股份有限公司 | Adaptive learning method and device |
CN110399541A (en) * | 2019-05-31 | 2019-11-01 | 平安国际智慧城市科技股份有限公司 | Topic recommended method, device and storage medium based on deep learning |
CN110675295A (en) * | 2019-09-29 | 2020-01-10 | 联想(北京)有限公司 | Processing method and device and electronic equipment |
CN111061694A (en) * | 2019-11-26 | 2020-04-24 | 上海乂学教育科技有限公司 | Student test question sharing system |
CN111428020A (en) * | 2020-04-09 | 2020-07-17 | 圆梦共享教育科技(深圳)有限公司 | Personalized learning test question recommendation method based on artificial intelligence |
CN114595923A (en) * | 2022-01-11 | 2022-06-07 | 电子科技大学 | Group teaching recommendation system based on deep reinforcement learning |
CN117874339A (en) * | 2024-01-03 | 2024-04-12 | 北京华乐思教育科技有限公司 | Intelligent recommendation system and method for testing and analyzing learning content |
-
2018
- 2018-04-08 CN CN201810307140.1A patent/CN108614865B/en active Active
Cited By (14)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN109255994A (en) * | 2018-10-26 | 2019-01-22 | 北京智能优学科技有限公司 | A kind of foreign language teaching adaptive learning method and computer readable storage medium |
CN109543840A (en) * | 2018-11-09 | 2019-03-29 | 北京理工大学 | A kind of Dynamic recommendation design method based on multidimensional classification intensified learning |
CN109543840B (en) * | 2018-11-09 | 2023-01-10 | 北京理工大学 | Dynamic recommendation system design method based on multidimensional classification reinforcement learning |
CN109859554A (en) * | 2019-03-29 | 2019-06-07 | 上海乂学教育科技有限公司 | Adaptive english vocabulary learning classification pushes away topic device and computer learning system |
CN110009956A (en) * | 2019-04-22 | 2019-07-12 | 上海乂学教育科技有限公司 | English Grammar adaptive learning method and learning device |
CN110223553A (en) * | 2019-05-20 | 2019-09-10 | 北京师范大学 | A kind of prediction technique and system of answering information |
CN110399541A (en) * | 2019-05-31 | 2019-11-01 | 平安国际智慧城市科技股份有限公司 | Topic recommended method, device and storage medium based on deep learning |
CN110288878B (en) * | 2019-07-01 | 2021-10-08 | 科大讯飞股份有限公司 | Self-adaptive learning method and device |
CN110288878A (en) * | 2019-07-01 | 2019-09-27 | 科大讯飞股份有限公司 | Adaptive learning method and device |
CN110675295A (en) * | 2019-09-29 | 2020-01-10 | 联想(北京)有限公司 | Processing method and device and electronic equipment |
CN111061694A (en) * | 2019-11-26 | 2020-04-24 | 上海乂学教育科技有限公司 | Student test question sharing system |
CN111428020A (en) * | 2020-04-09 | 2020-07-17 | 圆梦共享教育科技(深圳)有限公司 | Personalized learning test question recommendation method based on artificial intelligence |
CN114595923A (en) * | 2022-01-11 | 2022-06-07 | 电子科技大学 | Group teaching recommendation system based on deep reinforcement learning |
CN117874339A (en) * | 2024-01-03 | 2024-04-12 | 北京华乐思教育科技有限公司 | Intelligent recommendation system and method for testing and analyzing learning content |
Also Published As
Publication number | Publication date |
---|---|
CN108614865B (en) | 2020-12-11 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108614865A (en) | Method is recommended in individualized learning based on deeply study | |
CN108573628B (en) | Learning ability evaluation and knowledge point set extension recommendation method based on learning track | |
Gärdenfors et al. | The archaeology of teaching and the evolution of Homo docens | |
Yufeia et al. | Review of the application of artificial intelligence in education | |
CN104765842B (en) | A kind of optimal Learning Scheme method for pushing and system | |
Grivokostopoulou et al. | Utilizing semantic web technologies and data mining techniques to analyze students learning and predict final performance | |
CN109800300A (en) | learning content recommendation method and system | |
CN111461442A (en) | Knowledge tracking method and system based on federal learning | |
Perkins | Education in process systems engineering: past, present and future | |
Alshalabi et al. | Automated Adaptive Mobile Learning System using Shortest Path Algorithm and Learning Style. | |
Kerfoot | Participator education in a South African context: Contradictions and challenges | |
OROZOVA et al. | Generalized net model for dynamic decision making and prognoses | |
CN108647788A (en) | A kind of automatic improved method of association type knowledge base | |
Limongelli et al. | Fuzzy student modeling for personalization of e-learning courses | |
Hamada et al. | A study of a learning style index to support an intelligent and adaptive learning systems | |
Cronjé | What should we be teaching if Google gives the answer before we have even finished typing the question? | |
Tacadao et al. | A generic model for the group formation problem using constraint logic programming | |
Thanyaphongphat | Effects of personalized learning with preferred digital media types on learning motivation | |
Zhang | Research on e-learning in college education: A model on personalized resource recommendation based on learning situation | |
Bhattacharya et al. | An effective e-learning system through learners' scaffolding | |
Cambridge | Learning, knowing, and reflecting: Literacies for the 21st century | |
Shin et al. | Self-learning system based on metadata management module (MMM) for providing self-learning service | |
Márquez et al. | A Framework for Automatic Identification of Learning Styles in Learning Management Systems. | |
Gill | Graphs of affine transformations, with applications to sequential circuits | |
Bakar et al. | Education 4.0: Prediction model towards usage on internet and services for young learners |
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 |