CN109783647B - Construction method of intelligent learning model - Google Patents
Construction method of intelligent learning model Download PDFInfo
- Publication number
- CN109783647B CN109783647B CN201811574626.8A CN201811574626A CN109783647B CN 109783647 B CN109783647 B CN 109783647B CN 201811574626 A CN201811574626 A CN 201811574626A CN 109783647 B CN109783647 B CN 109783647B
- Authority
- CN
- China
- Prior art keywords
- knowledge
- individual
- knowledge points
- points
- questions
- 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.)
- Active
Links
Images
Classifications
-
- 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
Abstract
The invention provides a construction method of an intelligent learning model, which is characterized by comprising the following steps: firstly, pushing initial exercises; secondly, collecting exercise data; thirdly, updating the individual knowledge graph; fourth, the problem resource is pushed. Aiming at the defects of the prior art, the invention provides a construction method of an intelligent learning model, realizes personalized learning of students and meets the requirement of self-adaptive learning.
Description
Technical Field
The invention relates to the technical field of big data analysis and artificial intelligence, in particular to a construction method of an intelligent learning model.
Background
The online learning has become a new trend of enterprises and public institutions in learning modes, and the current learning mode is that students are leaned on to select knowledge point catalogues to push problem resources, or a random and extensive pushing exercise bag is used for effectively covering the knowledge point defects of the students and meeting the requirements of individual learning and differential learning of the students, so that the method for constructing the intelligent learning model is invented. The accurate pushing of learning resources of students is realized, and the learning effectiveness, pertinence and learning efficiency are improved.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides a construction method of an intelligent learning model, realizes personalized learning of students and meets the requirement of self-adaptive learning.
The invention provides a construction method of an intelligent learning model, which is characterized by comprising the following steps:
firstly, pushing initial exercises;
secondly, collecting exercise data;
thirdly, updating the individual knowledge graph;
fourth, the problem resource is pushed.
In the technical scheme, the first step comprises the following steps:
if the learner learns for the first time, that is, the individual knowledge graph does not have any data of the learner, the learner selects no more than N/2 knowledge points by himself, and the number of the selected knowledge points is s (0<s is less than or equal to N/2);
1) Question distribution for each knowledge point:
ts i =round (Ns) (1, …, s-1), round is a rounding function.
2) Each knowledge point title selection:
selecting the corresponding knowledge point topic number ts from the topic library by using the knowledge point and the lowest difficulty coefficient 1 i The initial problem pushing is completed.
In the technical scheme, the second step comprises the following steps:
collecting exercise data, namely collecting the answer condition of the student individual to the push problem. The method comprises the steps of question numbering, question names, knowledge points, whether answers are correct, answer time and overtime, and a basic data set for learning behaviors of students is formed.
In the above technical solution, the third step includes the following steps:
after each practice is finished, the individual knowledge graph base is updated in batches, the average value of the difficulty, the error question rate, the delay rate and the severity index of the existing knowledge points are required to be updated, and the non-stored knowledge points are added. After the index updating is completed, the knowledge graph is reordered from large to small according to the severity index;
the calculation rule of each index in the individual knowledge graph base is as follows
1) Knowledge point severity index a i
a i ＝e i ×80+p i ×20
2) Knowledge point error rate e i 。
3) Delay rate p of knowledge point i
4) Difficulty average d of knowledge points i
d i The value of (2) is rounded off and a decimal fraction is reserved.
In the above technical solution, the fourth step includes the following steps:
determining a topic number distribution n:
1) If the individual knowledge graph base has no data, please refer to the initial problem pushing, if the data exists, the knowledge points are set to be N/2, and if the individual knowledge graph base has the data, the knowledge points are acquired from the individual knowledge graph base based on the number of the knowledge points of N/2. If the number of knowledge points in the individual knowledge graph is smaller than N/2, taking the actual number; if greater than or equal to N/2, then n=n/2;
2) Taking the first n serious indexes a of the ranking of the individual knowledge graphs i And are arranged from large to small according to the severity index itself, so that the severity index is the first most significant. If the severity index a i All 0, then an average allocation is used to correspond to topic 2 for each knowledge pair. If at least 1 condition other than 0 exists, calculating the weight coefficient ra of the severity index i (i＝1,…n)；
3) Calculating weight coefficient ra of severity index i (i＝1,…n)：
Assuming weight coefficient ra i A number of 0 is k (1.ltoreq.k)<n), each of the k knowledge points is assigned 1 question first;
ts i ＝round[(N-k)×ra i ](i＝2,··,n-k)
wherein round is a rounding function, considering that 1 is easily added to the rounding function, adding more is compensated for ts with highest weight 1 ；
Determining the source of the title:
the sources of questions are knowledge points, difficulty average values and numbers of questions are taken as basic conditions, the knowledge points are extracted from a problem resource library in a comparison mode, the knowledge points are N/2 knowledge points before ranking individual knowledge graphs, and the numbers of questions ts i The value-taking method of (2) is already defined in the step one, and the difficulty average value dn i Converting based on a difficulty average value D (i) of corresponding knowledge points in the knowledge graph;
the conversion method is as follows:
1) As the knowledge point z, the corresponding severity index d i Is 0.
Dn i ＝1；
2) As the knowledge point z, the corresponding severity index d i And is not 0.
Dn i ＝floor(d i )
floor is a function of the rounding down;
in screening the questions, it is necessary to filter out the questions that have been done in the problem package.
On the basis of establishing the question bank model, the invention establishes the individual knowledge graph of the student by collecting the learning record of the student, and dynamically pushes the learning resource by constructing the resource pushing engine, thereby realizing the personalized learning of the student and meeting the requirement of self-adaptive learning.
Drawings
FIG. 1 is a schematic diagram of the structure of the present invention
Detailed Description
The invention will now be described in further detail with reference to the drawings and specific examples, which are given for clarity of understanding and are not to be construed as limiting the invention.
As shown in fig. 1, the invention provides a method for constructing an intelligent learning model, which is characterized by comprising the following steps:
firstly, pushing initial exercises;
secondly, collecting exercise data;
thirdly, updating the individual knowledge graph;
fourth, the problem resource is pushed.
In the technical scheme, the first step comprises the following steps:
if the learner learns for the first time, that is, the individual knowledge graph does not have any data of the learner, the learner selects no more than N/2 knowledge points by himself, and the number of the selected knowledge points is s (0<s is less than or equal to N/2);
3) Question distribution for each knowledge point:
ts i =round (N/s) (1, …, s-1), round is a rounding function.
4) Each knowledge point title selection:
selecting the corresponding knowledge point topic number ts from the topic library by using the knowledge point and the lowest difficulty coefficient 1 i The initial problem pushing is completed.
In the technical scheme, the second step comprises the following steps:
collecting exercise data, namely collecting the answer condition of the student individual to the push problem. The method comprises the steps of question numbering, question names, knowledge points, whether answers are correct, answer time and overtime, and a basic data set for learning behaviors of students is formed.
In the above technical solution, the third step includes the following steps:
after each exercise is finished, the individual knowledge graph base is updated in batches based on the first table, the average value of the difficulty, the error question rate, the delay rate and the serious index of the existing knowledge points are required to be updated, and the non-stored knowledge points are added. After the index updating is completed, the knowledge graph is reordered from large to small according to the severity index;
list one
| Question(s) | Knowledge surface | Conventional duration of time | Actual duration of time | Whether or not to be correct | Whether or not to timeout |
| Subject 1 | Z1 | L1 | T1 | Is that | Whether or not |
| Subject 2 | Z2 | L2 | T2 | Whether or not | Invalidation of |
| Subject 3 | Z3 | L3 | T3 | Is that | T2>L2 |
| Title 4 | Z4+Z5 | L4 | T4 | Is that | T4<L4 |
| ...... | ..... | ...... | ..... | ...... | ..... |
The calculation rule of each index in the individual knowledge graph base is as follows
2) Knowledge point severity index a i
a i ＝e i ×80+p i ×20
2) Knowledge point error rate e i 。
3) Delay rate p of knowledge point i
4) Difficulty average d of knowledge points i
d i The value of (2) is rounded off and a decimal fraction is reserved.
In the above technical solution, the fourth step includes the following steps:
determining a topic number distribution n:
1) If the individual knowledge graph base has no data, please refer to the initial problem pushing, if the data exists, the knowledge points are set to be N/2, and if the individual knowledge graph base has the data, the knowledge points are acquired from the individual knowledge graph base based on the number of the knowledge points of N/2. If the number of knowledge points in the individual knowledge graph is smaller than N/2, taking the actual number; if greater than or equal to N/2, then n=n/2;
2) Taking the first n serious indexes a of the ranking of the individual knowledge graphs i And are arranged from large to small according to the severity index itself, so that the severity index is the first most significant. If the severity index a i All 0, then an average allocation is used to correspond to topic 2 for each knowledge pair. If at least 1 condition other than 0 exists, calculating the weight coefficient ra of the severity index i (i＝1,…n)；
3) Calculating weight coefficient ra of severity index i (i＝1,…n)：
Assuming weight coefficient ra i A number of 0 is k (1.ltoreq.k)<n), each of the k knowledge points is assigned 1 question first;
ts i ＝round[(N-k)×ra i ](i＝2,··,n-k)
wherein round is a rounding function, considering that 1 is easily added to the rounding function, adding more is compensated for ts with highest weight 1 ；
Determining the source of the title:
the sources of questions are knowledge points, difficulty average values and numbers of questions are taken as basic conditions, the knowledge points are extracted from a problem resource library in a comparison mode, the knowledge points are N/2 knowledge points before ranking individual knowledge graphs, and the numbers of questions ts i The value method of (2) is already defined in the step one, and the difficulty mean dn of the sources of the questions i Converting based on the difficulty average value d (i) of the corresponding knowledge points in the knowledge graph;
the conversion method is as follows:
3) If the knowledge point z, the corresponding severity index is 0.
Dn i ＝1；
4) If the knowledge point z, the corresponding severity index is not 0.
Dn i ＝floor(d i )
floor is a function of the rounding down;
in screening the questions, it is necessary to filter out the questions that have been done in the problem package.
What is not described in detail in this specification is prior art known to those skilled in the art.
Claims (1)
1. The construction method of the intelligent learning model is characterized by comprising the following steps:
firstly, pushing initial exercises;
secondly, collecting exercise data;
thirdly, updating the individual knowledge graph;
step four, pushing problem resources;
the first step comprises the following steps:
if the learner learns for the first time, that is, the individual knowledge graph does not have any data of the learner, the learner selects no more than N/2 knowledge points by himself, and the number of the selected knowledge points is assumed to be s;0<s N/2;
1) Question distribution for each knowledge point:
ts i =round (Ns) (1, …, s-1), round is a rounding function;
2) Each knowledge point title selection:
selecting the corresponding knowledge point topic number ts from the topic library by using the knowledge point and the lowest difficulty coefficient 1 i Completing the initial problem pushing;
the second step comprises the following steps:
collecting exercise data, namely collecting answer situations of individual students to the push exercises; the method comprises the steps of forming a basic data set of learning behaviors of students, wherein the basic data set comprises question numbers, question names, knowledge points, whether answers are correct, answer time and overtime;
the third step comprises the following steps:
after each practice is finished, batch updating is carried out on the individual knowledge graphs, the average value of the difficulty, the error question rate, the delay rate and the severity index of the existing knowledge points are required to be updated, and the non-stored knowledge points are added; after the index updating is completed, the knowledge graph is reordered from large to small according to the severity index;
the calculation rule of each index in the individual knowledge graph is as follows
1) Knowledge point severity index a i
a i ＝e i ×80+p i ×20
2) Knowledge point error rate e i
3) Delay rate p of knowledge point i
4) Difficulty average d of knowledge points i
d i Rounding the value of (2), and reserving a decimal place;
the fourth step comprises the following steps:
determining a topic number distribution:
1) If the individual knowledge graph does not have data, please refer to the initial problem pushing, if the data exists, the number of knowledge points is set to be N/2, and if the individual knowledge graph has data, the number of knowledge points is based on the number of the knowledge points of N/2, and the knowledge points are obtained from the individual knowledge graph; if the number of knowledge points in the individual knowledge graph is smaller than N/2, taking the actual number; if greater than or equal to N/2, then n=n/2;
2) Taking the first n serious indexes a of the ranking of the individual knowledge graphs i And the severity indexes are arranged from large to small according to the severity indexes; if the severity index a i All 0, then adopting average distribution to correspond to the title 2 for each knowledge face; if there are at least 1 other than 0, then computing severe meansA weighting coefficient of the number;
3) Calculating weight coefficient ra of severity index i ,i＝1,...,n：
Assuming weight coefficient ra i The number of knowledge points of 0 is k; k is more than or equal to 1<n, each of the k knowledge points is assigned 1 question;
ts i ＝round[(N-k)×ra i ]
i＝2,...,n-k
wherein round is a rounding function, considering that 1 is easily added to the rounding function, adding more is compensated for ts with highest weight 1 ；
Determining the source of the title:
the sources of questions are that knowledge points and difficulty mean values are taken as basic conditions, and the questions are extracted from a problem resource library in a comparison mode, wherein the knowledge points are N/2 knowledge points before ranking individual knowledge graphs, and the questions ts i The value method of (2) is already defined in the step one, and the difficulty mean dn of the sources of the questions i Converting based on the difficulty average value d (i) of the corresponding knowledge points in the knowledge graph;
the conversion method is as follows:
1) As the knowledge point z, the corresponding severity index a i Is set to be 0, the number of the components is set to be 0,
dn i ＝1；
2) As the knowledge point z, the corresponding severity index a i Is not the same as the value of 0,
dn i ＝floor(d i )
floor is a function of the rounding down;
in screening the questions, it is necessary to filter out the questions that have been done in the problem package.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811574626.8A CN109783647B (en) | 2018-12-21 | 2018-12-21 | Construction method of intelligent learning model |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811574626.8A CN109783647B (en) | 2018-12-21 | 2018-12-21 | Construction method of intelligent learning model |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN109783647A CN109783647A (en) | 2019-05-21 |
| CN109783647B true CN109783647B (en) | 2023-06-13 |
Family
ID=66497625
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811574626.8A Active CN109783647B (en) | 2018-12-21 | 2018-12-21 | Construction method of intelligent learning model |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN109783647B (en) |
Family Cites Families (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101477755A (en) * | 2009-01-06 | 2009-07-08 | 陈诗周 | Error problem database system for student |
| US10290221B2 (en) * | 2012-04-27 | 2019-05-14 | Aptima, Inc. | Systems and methods to customize student instruction |
| CN104966427A (en) * | 2015-05-27 | 2015-10-07 | 北京创数教育科技发展有限公司 | Self-adaptation teaching interaction system and method |
| CN106469169A (en) * | 2015-08-19 | 2017-03-01 | 阿里巴巴集团控股有限公司 | Information processing method and device |
| US10769539B2 (en) * | 2016-08-16 | 2020-09-08 | International Business Machines Corporation | Automatic evaluation of a knowledge canvassing application |
| CN106599999A (en) * | 2016-12-09 | 2017-04-26 | 北京爱论答科技有限公司 | Evaluation method and system for using small amount of questions to accurately detect segmented weak knowledge points of student |
| CN107085803B (en) * | 2017-03-31 | 2020-09-08 | 弘成科技发展有限公司 | Individualized teaching resource recommendation system based on knowledge graph and ability evaluation |
| CN108053130A (en) * | 2017-12-25 | 2018-05-18 | 郑州威科姆科技股份有限公司 | A kind of multimedia teaching analysis system and analysis method |
| CN108647363A (en) * | 2018-05-21 | 2018-10-12 | 安徽知学科技有限公司 | Map construction, display methods, device, equipment and storage medium |
-
2018
- 2018-12-21 CN CN201811574626.8A patent/CN109783647B/en active Active
Also Published As
| Publication number | Publication date |
|---|---|
| CN109783647A (en) | 2019-05-21 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN109582864B (en) | Course recommendation method and system based on big data science and dynamic weight adjustment | |
| Pardos et al. | KT-IDEM: Introducing item difficulty to the knowledge tracing model | |
| CN107665473B (en) | Learning path planning method and device | |
| CN107067849B (en) | Parent end system based on cloud teaching platform | |
| CN107909520B (en) | Paper output method and device based on test question correlation | |
| CN108920663B (en) | Method for automatically recommending title content | |
| US10216827B2 (en) | Participant grouping for enhanced interactive experience | |
| CN107133894B (en) | Online learning grouping method based on complex network theory | |
| CN110929020B (en) | Knowledge point mastering degree analysis method based on test results | |
| WO2016197855A1 (en) | Learning attainment recommendation method and device | |
| CN110659311A (en) | Topic pushing method and device, electronic equipment and storage medium | |
| CN111539581A (en) | Intelligent class scheduling method and system for different shifts | |
| CN109783647B (en) | Construction method of intelligent learning model | |
| CN113269532B (en) | College course arrangement method based on urgency and backtracking algorithm | |
| CN113052316B (en) | Knowledge tracking method, system, equipment and storage medium based on causal reasoning | |
| CN108073603A (en) | Operation distribution method and device | |
| CN111311997B (en) | Interaction method based on network education resources | |
| CN116028602A (en) | Question recommending method and device, computer equipment and storage medium | |
| CN111898803B (en) | Problem prediction method, system, equipment and storage medium | |
| KR102146112B1 (en) | Method for providing an information of a question analysis using big data | |
| KR20190125055A (en) | Learning coaching method using big data | |
| CN115100912A (en) | Teaching activity design system based on big data | |
| WO2014193219A1 (en) | Method for collaborative learning based on divide-master-lead strategy in an intelligent collaborative learning system | |
| CN107423311B (en) | Method and apparatus for providing online diagnostic learning | |
| CN113902596A (en) | Post-class service method and system using information matching |
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 |