CN113254629B - Learning content recommendation method and system based on artificial intelligence - Google Patents
Learning content recommendation method and system based on artificial intelligence Download PDFInfo
- Publication number
- CN113254629B CN113254629B CN202110629607.6A CN202110629607A CN113254629B CN 113254629 B CN113254629 B CN 113254629B CN 202110629607 A CN202110629607 A CN 202110629607A CN 113254629 B CN113254629 B CN 113254629B
- Authority
- CN
- China
- Prior art keywords
- test
- knowledge
- point
- points
- question
- 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
- 238000000034 method Methods 0.000 title claims abstract description 34
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 28
- 238000012360 testing method Methods 0.000 claims abstract description 282
- 238000000605 extraction Methods 0.000 claims description 46
- 238000012545 processing Methods 0.000 claims description 16
- 230000000007 visual effect Effects 0.000 claims description 15
- 238000013145 classification model Methods 0.000 claims description 14
- 238000007637 random forest analysis Methods 0.000 claims description 12
- 238000010276 construction Methods 0.000 claims description 6
- 230000000875 corresponding effect Effects 0.000 claims 14
- 230000002596 correlated effect Effects 0.000 claims 1
- 230000000694 effects Effects 0.000 abstract description 4
- 238000003066 decision tree Methods 0.000 description 6
- 238000011835 investigation Methods 0.000 description 6
- 238000012549 training Methods 0.000 description 6
- 230000009471 action Effects 0.000 description 3
- 230000002349 favourable effect Effects 0.000 description 3
- 230000008569 process Effects 0.000 description 3
- 230000009286 beneficial effect Effects 0.000 description 2
- 238000005056 compaction Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000012905 input function Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 210000003462 vein Anatomy 0.000 description 2
- 210000001072 colon Anatomy 0.000 description 1
- 238000000354 decomposition reaction Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000007786 learning performance Effects 0.000 description 1
- 238000005259 measurement Methods 0.000 description 1
- 230000003340 mental effect Effects 0.000 description 1
- 230000002093 peripheral effect Effects 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
Images
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
-
- 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
- G06F16/35—Clustering; Classification
-
- 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
- G06F16/36—Creation of semantic tools, e.g. ontology or thesauri
- G06F16/367—Ontology
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- 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
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Databases & Information Systems (AREA)
- Evolutionary Biology (AREA)
- Strategic Management (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computational Linguistics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Artificial Intelligence (AREA)
- Educational Administration (AREA)
- Educational Technology (AREA)
- Evolutionary Computation (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)
- General Business, Economics & Management (AREA)
- Animal Behavior & Ethology (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The application provides a learning content recommendation method and system based on artificial intelligence, which are used for counting the scoring condition of each knowledge point in a test answer sheet and outputting a graph statistical result by acquiring the test answer sheet of a student, a knowledge point distribution network corresponding to the test answer sheet and the knowledge point distribution network. The output of the statistical result of the pictures can be reflected to students, so that the students can know the knowledge mastering conditions of the students. Then determining knowledge points to be promoted of the students according to the statistical results of the graphs; therefore, the learning content for learning and/or testing the knowledge points to be improved is determined and pushed to the student for learning. Therefore, automatic and accurate pushing of the learning content can be achieved, manpower and material resources are saved, artificial deviation can be overcome, artificial careless omission can be overcome, the more valuable learning content can be pushed to students, the learning efficiency of the students is improved, and the method is very practical and effective without the need of committing to activities of teachers.
Description
Technical Field
The application relates to the technical field of artificial intelligence, in particular to a learning content recommendation method and system based on artificial intelligence.
Background
In view of the great difference between the teaching management mode of colleges and universities and middle schools, some college students are difficult to adapt to the learning tasks (the learning tasks are heavy and the difficulty of course contents is high) and the learning modes (the college students do not rely on the learning consciousness of students any more instead of the supervised teaching mode) of colleges and universities after departing from the supervised learning mode of middle schools, so that the learning performance is not ideal, the professional knowledge is not firmly mastered, the assessment of the learning tasks of the colleges and the universities cannot reach the standard, the college students leave the schools to enter the society later, and the lack of professional knowledge is not favorable for the growth of the college students when engaged in related work.
Therefore, how to provide a learning scheme more suitable for the learning mode of the college students (or a learning scheme capable of being simultaneously suitable for the college students and the middle students) according to the characteristics of the college students is a problem to be solved at present.
Disclosure of Invention
An object of the embodiments of the present application is to provide a learning content recommendation method and system based on artificial intelligence, so as to recommend appropriate learning content in a targeted manner according to knowledge mastering conditions of different students, so as to significantly improve the knowledge mastering level of the students.
In order to achieve the above object, embodiments of the present application are implemented as follows:
in a first aspect, an embodiment of the present application provides a learning content recommendation method based on artificial intelligence, where all students in a same course learn the course and participate in a test of the course on a same system, and the method is applied to the system and includes: obtaining a test answer sheet of a student, wherein the test answer sheet is an electronic answer sheet of the test paper; acquiring a knowledge point distribution network corresponding to the test answer sheet, counting the score condition of each knowledge point in the test answer sheet based on the knowledge point distribution network, and outputting a graph statistical result; determining knowledge points to be promoted of the student according to the graph statistical result; and determining the learning content for learning and/or testing the knowledge points to be improved according to the knowledge points to be improved, and pushing the learning content to the student for learning.
In the embodiment of the application, the score condition of each knowledge point in the test answer sheet is counted by acquiring the test answer sheet (electronic answer sheet) of the student and the knowledge point distribution network corresponding to the test answer sheet and based on the knowledge point distribution network, and the graph statistical result is output. The output of picture class statistical result can reflect for student itself (when the student is middle school student, can also reflect for mr) to make the student know the knowledge grasping condition of oneself (when the student is middle school student, can also be convenient for mr to know student's knowledge grasping condition, feed back to self, carry out the regulation of teaching mode and the supplementary teaching of course content). Then determining knowledge points to be promoted of the students according to the statistical results of the graphs; therefore, the learning content for learning and/or testing the knowledge points to be promoted is determined and pushed to the student for learning. Therefore, automatic and accurate pushing of the learning content can be achieved, manpower and material resources are saved, artificial deviation can be overcome (for example, whether a teacher is responsible for supervising learning of students, whether the teacher is sufficiently responsible for learning, whether the teacher has an eccentric problem and the like), and artificial careless omission (neglect of knowledge points needing to be reviewed due to carelessness) can be overcome in the mode, more valuable learning content is pushed to the students, the learning efficiency of the students is improved, and the mode is very practical and effective without the need of mental committal of the teacher.
With reference to the first aspect, in a first possible implementation manner of the first aspect, before the obtaining a test answer sheet of a student, the method further includes: for each set of test paper: processing the question stem of each test question of the test paper, and determining that the question stem of the test question contains a target single sentence for solving the requirement; performing entity extraction, attribute extraction and relationship extraction on the target single sentence to construct a solution point of the test question; carrying out entity extraction, attribute extraction and relation extraction on auxiliary single sentences except the target single sentence in the question stem of the test question to construct condition points of the test question; matching attributes in the solution points and the condition points of the test questions with attributes in a knowledge graph of the curriculum which is constructed in advance, and constructing knowledge points of the test questions based on matching results; constructing a knowledge point distribution network of the test paper according to the knowledge points of each test question of the test paper; correspondingly, the acquiring the knowledge point distribution network corresponding to the test answer sheet includes: and determining a knowledge point distribution network corresponding to the corresponding test paper according to the test answer sheet.
In the implementation mode, the question stem of each test question of the test paper is processed to determine a target single sentence containing a solving requirement in the question stem of the test question, entity extraction, attribute extraction and relationship extraction are carried out on the target single sentence to construct a solving point of the test question, and entity extraction, attribute extraction and relationship extraction are carried out on an auxiliary single sentence in the question stem of the test question to construct a condition point of the test question; and then matching the attributes in the solution points and the condition points of the test questions with the attributes in the pre-constructed knowledge graph of the course, and constructing the knowledge points of the test questions based on the matching result, thereby constructing a knowledge point distribution network of the test paper according to the knowledge points of each test question of the test paper. In such a way, all points (solution points, condition points and the like) involved in each test question can be constructed as knowledge points of the test question, and when students lose scores on the test question, the knowledge points causing the time division of the students can be determined, so that missing and filling can be found, and learning contents can be pushed in a targeted manner. The constructed knowledge point distribution network can visualize the distribution situation of the knowledge points of the whole set of test questions, thereby facilitating the subsequent additional processing (for example, outputting the visualized score statistical result through the knowledge point distribution network). And the entity extraction, the attribute extraction and the relation extraction are utilized to process the single sentences (target single sentences, auxiliary single sentences and the like) of the question stem, so that the effective information in the question stem can be efficiently and accurately extracted, the decomposition of the test questions, the determination of the knowledge points of the examination of the test questions and the like are facilitated.
With reference to the first possible implementation manner of the first aspect, in a second possible implementation manner of the first aspect, the processing a stem of each test question of the test paper includes: performing sentence breaking processing on the question stem of each test question to determine a plurality of single sentences to be defined; and inputting each single sentence to be defined into a preset sentence classification model, and determining the category of the single sentence to be defined, wherein the category of the single sentence to be defined is the target single sentence or the auxiliary single sentence.
In the implementation mode, sentence breaking processing is carried out on the question stem of each test question to determine a plurality of single sentences to be defined; and inputting each single sentence to be defined into a preset sentence classification model, and determining the category of the single sentence to be defined. The classification can be carried out after sentence breaking, relative independence among the single sentences is guaranteed, and then the corresponding single sentence is determined to be the target single sentence or the auxiliary single sentence, so that the method can help to determine the knowledge points for investigation (the solution point is the main investigation point, the condition point is the secondary investigation point, or the solution point and the condition point are the knowledge points needing investigation on the whole), and is favorable for constructing the corresponding knowledge points.
With reference to the second possible implementation manner of the first aspect, in a third possible implementation manner of the first aspect, the statement classification model is a random forest model, n _ estimators of the random forest model is 8, min _ samples _ leaf is 3, and max _ features, max _ depth, min _ samples _ split, min _ weight _ fraction _ leaf, and max _ leaf _ nodes are all set as default values.
In the implementation mode, the sentence classification model is mainly applied to classifying the single sentences, so that the random forest model is selected as the basis for training, and a satisfactory effect can be achieved. In addition, considering the application scenario of the sentence classification model, the sample size that can be obtained, and other factors, the training mode may adopt a supervised training mode, and the sample size that can be obtained is mainly limited by the test question size of the subject course, so for university courses with relatively few test question sizes (such as university physics, which mainly comes from post-school problems, years test questions, and the like), the random forest model has n _ estimators (the number of decision trees) of 8, min _ samples _ leaf (the minimum number of samples of leaf nodes) of 3, max _ features (the maximum number of features), max _ depth (the maximum depth of decision trees), min _ samples _ split (the number of samples of nodes is less than min _ samples _ split), and no attempt is made to select the optimal features again to divide, min _ weight _ free _ leaf (the minimum value of all sample weights and the leaf nodes), and the maximum number of leaf nodes _ sample nodes _ leaf _ segment (the maximum number of leaf nodes), can prevent overfitting) is set as a default value, the operation efficiency of the model can be effectively improved while the classification precision is ensured. For the courses with large examination problem amount (such as physical courses of middle school), the sample amount is large, n _ estimators can be mainly adjusted (for example, adjusted to 12), min _ samples _ leaf can be correspondingly adjusted (for example, adjusted to 4, and the like), and max _ depth can be correspondingly adjusted (set according to specific situations), and the maximum depth (for example, 4 or 6) is set, so that a short and fat decision tree can be obtained as far as possible, and the classification accuracy and the model operation efficiency are improved.
With reference to the first possible implementation manner of the first aspect, in a fourth possible implementation manner of the first aspect, the matching the attributes in the solution point and the condition point of the test question with the attributes in the pre-constructed knowledge graph of the course, and constructing the knowledge point of the test question based on the matching result includes: for each solution point of the test question, carrying out consistency matching on the attributes in the solution points of the test question and the attributes in the knowledge graph of the course, and determining the matched attributes as the main attributes of corresponding entities in the solution points; for each condition point of the test question, carrying out consistency matching on the attributes in the condition point and the attributes in the knowledge map of the course, if the entity in the condition point is consistent with any entity in the solution point, determining the matched attributes as the secondary attributes of the corresponding entities in the solution point, and if the entity in the condition point is not consistent with all the entities in the solution point, determining the matched attributes as the condition attributes of the corresponding entities in the condition point; and constructing the knowledge points of the mutual correlation between the solution points and the condition points of the test question based on the main attribute and the auxiliary attribute of the entity in each solution point and the condition attribute of the entity in each condition point.
In the implementation manner, for each solution point of the test question, consistency matching is performed between attributes in the solution points of the test question and attributes in the knowledge graph of the course, the matched attributes are determined to be main attributes of corresponding entities in the solution points, consistency matching is performed between attributes in the condition points and attributes in the knowledge graph of the course, if an entity in the condition points is consistent with any entity in the solution points, the matched attributes are determined to be secondary attributes of the corresponding entities in the solution points, and if the entity in the condition points is inconsistent with all entities in the solution points, the matched attributes are determined to be conditional attributes of the corresponding entities in the condition points. Therefore, the entities of each solution point and each condition point can be found, the corresponding attributes are integrated, the same attributes of the corresponding entities are integrated under the same entity (the entity of the solution point is preferred), the main attribute and the auxiliary attribute of the corresponding entity in the solution point and the condition attribute of the corresponding entity in the condition point are obtained, and a knowledge point which is mutually related and has clear veins between the solution point and the condition point is constructed. Therefore, the accuracy of the knowledge points in the test questions can be ensured, and the accuracy of recommending the subsequent learning content is facilitated.
With reference to the fourth possible implementation manner of the first aspect, in a fifth possible implementation manner of the first aspect, the constructing the knowledge points of the test question, which are associated with each other between the solution points and the condition points, based on the primary attribute and the secondary attribute of the entity in each solution point and the condition attribute of the entity in each condition point includes: establishing the relation between the entities with different solution points, the relation between the entities with different condition points and the relation between the entities with solution points and the entities with condition points; and establishing a relation between the main attribute and the auxiliary attribute of the entity in each solution point and the solution point, and establishing a relation between the condition attribute of the entity in each condition point and the condition point to construct a knowledge point of the test question.
In the implementation mode, the relation between the entities with different solution points is constructed, the relation between the entities with different condition points is constructed, and the relation between the entity with the solution point and the entity with the condition point is constructed; and establishing a relation between the main attribute and the auxiliary attribute of the entity in each solution point and the solution point, and establishing a relation between the condition attribute of the entity in each condition point and the condition point to construct a knowledge point of the test question. This way, the relationship between the entities in the solution point and the condition point can be established, so that the established knowledge point of the test question can have a very clear correlation between knowledge (for example, the fixed axis and moment of the rigid body a, the fixed axis of the rigid body B, the rotational kinetic energy of the moment of the rigid body B, and the like, and the rigid body a and the rigid body B are coaxial).
With reference to the first possible implementation manner of the first aspect, in a sixth possible implementation manner of the first aspect, the constructing a knowledge point distribution network of the test paper according to knowledge points of each test question of the test paper includes: aiming at the knowledge points of each test question of the test paper, determining the belonged points, the belonged sections and the belonged chapters of the knowledge points; and constructing a tree network according to the affiliated dots, the affiliated sections and the affiliated chapters of each knowledge point to obtain a knowledge point distribution network of the test paper, wherein each branch node of the tree network is provided with a text box, the display content of the text box comprises question amount, and the question amount represents the number of the test questions of the branch node.
In the implementation manner, for the knowledge points of each test question of the test paper, the belonged dots, the belonged sections and the belonged chapters of the knowledge points are determined, and then a tree network is constructed by using the belonged dots, the belonged sections and the belonged chapters of each knowledge point, so as to obtain the knowledge point distribution network of the test paper (each branch node has a text box, and the display content of the text box includes the quantity of questions). The knowledge point distribution network of the test paper constructed in the way has a very clear and clear knowledge point relation network, so that the knowledge points of each test question of the test paper can be clearly disclosed, the distribution of the knowledge points inspected by the test paper can be reflected, and the output of the statistical results of subsequent pictures is facilitated.
With reference to the sixth possible implementation manner of the first aspect, in a seventh possible implementation manner of the first aspect, the performing statistics on the score condition of each knowledge point in the test answer sheet based on the knowledge point distribution network, and outputting a graph statistics result includes: and determining the total score and the score of each test question in the test answer sheet, recording the total score and the score of each test question in a corresponding branch node of the knowledge point distribution network, so that the corresponding branch node can perform statistics based on all the total scores and the score of the branch node, and outputting a visual knowledge point distribution network displaying the total score and the score through a text box on the branch node, wherein the visual knowledge point distribution network is the graph statistical result.
In this implementation manner, for each test question in the test answer sheet, the total score and the score of the test question are determined, and the total score and the score of the test question are recorded in a corresponding branch node of the knowledge point distribution network, so that the corresponding branch node can perform statistics based on all the total scores and the scores on the branch node, and output a visual knowledge point distribution network (i.e., a graph-like statistical result) in which the total score and the score are displayed through a text box on the branch node. The method can accurately count the answer condition of the student, correspondingly count and output the score condition of the knowledge point, and obtain the knowledge point mastering condition of the student, so that the learning content is recommended in a targeted manner, the accuracy of pushing the learning content is ensured, and the learning efficiency and the knowledge mastering level of the student are effectively improved.
With reference to the seventh possible implementation manner of the first aspect, in an eighth possible implementation manner of the first aspect, the determining, according to the graph statistics result, a knowledge point to be improved of the student includes: determining a first class of branch nodes, a second class of branch nodes and a third class of branch nodes according to a score ratio between a score value and a total score value of each branch node at the bottommost layer in the visual knowledge point distribution network, wherein the branch node at the bottommost layer represents the branch node corresponding to the small point of the knowledge point, the first class of branch nodes represent that the knowledge point on the branch node does not need to be subjected to additional study, the second class of branch nodes represent that the knowledge point on the branch node needs to be subjected to advanced promotion, and the third class of branch nodes represent that the knowledge point on the branch node needs to be subjected to basic compaction; and determining the knowledge points corresponding to the second class of branch nodes and the third class of branch nodes as the knowledge points to be improved.
In the implementation mode, the score ratio between the score value and the total score value of each branch node at the bottommost layer in the visual knowledge point distribution network is utilized to classify the branch nodes at the bottommost layer (the branch nodes corresponding to the small points to which the knowledge points belong) (first-class branch nodes, second-class branch nodes and third-class branch nodes), the knowledge points corresponding to the second-class branch nodes and the third-class branch nodes are determined as the knowledge points to be promoted, and the knowledge points are used as conditions for determining the learning content to be pushed, so that the learning content which is most beneficial to the learning of the student is screened out.
In a second aspect, an embodiment of the present application provides an artificial intelligence-based learning content recommendation system, on which all students of a same course learn the course and participate in a test of the course, the system includes: the test answer sheet acquisition unit is used for acquiring test answer sheets of students, wherein the test answer sheets are electronic answer sheets of the test answer sheets; the statistical result output unit is used for acquiring a knowledge point distribution network corresponding to the test answer sheet, performing statistics on the score condition of each knowledge point in the test answer sheet based on the knowledge point distribution network, and outputting a graph statistical result; the to-be-promoted knowledge point determining unit is used for determining the to-be-promoted knowledge points of the students according to the graph statistical result; and the learning content recommending unit is used for determining the learning content for learning and/or testing the knowledge points to be improved according to the knowledge points to be improved and pushing the learning content to the student for learning.
In order to make the aforementioned objects, features and advantages of the present application more comprehensible, preferred embodiments accompanied with figures are described in detail below.
Drawings
To more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are required to be used in the embodiments of the present application will be briefly described below, it should be understood that the following drawings only illustrate some embodiments of the present application and therefore should not be considered as limiting the scope, and that those skilled in the art can also obtain other related drawings based on the drawings without inventive efforts.
Fig. 1 is a flowchart of a learning content recommendation method based on artificial intelligence according to an embodiment of the present application.
Fig. 2 is a block diagram of a learning content recommendation system based on artificial intelligence according to an embodiment of the present application.
An icon: 10-artificial intelligence based learning content recommendation system; 11-a test answer sheet acquisition unit; 12-a statistical result output unit; 13-a knowledge point determination unit to be promoted; 14-learning content recommendation unit.
Detailed Description
The technical solutions in the embodiments of the present application will be described below with reference to the drawings in the embodiments of the present application.
In this embodiment, the learning content recommendation method based on artificial intelligence may be applied to an electronic device, where the electronic device may be a terminal having a display function and an input function (e.g., a smart phone, a tablet computer, a notebook computer, a personal computer having a peripheral input device, etc.); or a server connected to a display terminal (also having an input function), where the server operates a learning content recommendation method based on artificial intelligence, and sends the output result and the learning content to the display terminal for display, so that students can check and learn the learning content, which is not limited herein.
Therefore, the artificial intelligence based learning content recommendation method will be described below with an electronic device as an execution subject.
Referring to fig. 1, fig. 1 is a flowchart of a learning content recommendation method based on artificial intelligence according to an embodiment of the present application.
In the present embodiment, the artificial intelligence based learning content recommendation method may include step S10, step S20, step S30, and step S40.
To facilitate understanding of the present solution, before describing step S10, a process of constructing a knowledge point distribution network for each test paper set is described here.
In this embodiment, for each set of test paper, the electronic device may perform the following processing on each test question in the test paper:
firstly, the electronic device can process the question stem of the test question and determine that the question stem of the test question contains the target clause of the solving requirement.
In this embodiment, for each test question, the electronic device may perform sentence-breaking processing on a question stem of the test question to determine a plurality of single sentences to be defined. The sentence-breaking mode can intelligently break sentences by using the existing mode to obtain a plurality of single sentences. Of course, the existing sentence-punctuating model may be adjusted as needed so that the single sentence output after the punctuation is a single sentence without commas or colons (i.e., the terminal of the sentence is not used as the punctuation basis, and commas may also be used as the punctuation basis), which is not limited herein.
After a plurality of to-be-defined sentences are determined, the electronic device can input each to-be-defined sentence into a preset sentence classification model, determine the category of the to-be-defined sentence, and determine the category of the to-be-defined sentence, wherein the category of the to-be-defined sentence is a target sentence or an auxiliary sentence, the target sentence represents a sentence containing a solving requirement, and the auxiliary sentence is a sentence not containing the solving requirement and mainly serves as a sentence providing the solving condition.
Performing sentence breaking processing on the question stem of each test question to determine a plurality of single sentences to be defined; and inputting each single sentence to be defined into a preset sentence classification model, and determining the category of the single sentence to be defined. Therefore, classification can be carried out after sentence breaking, relative independence among the single sentences is guaranteed, and then whether the corresponding single sentence is the target single sentence or the auxiliary single sentence is determined, so that the method can help to determine the knowledge points to be investigated (solving points are main investigation points, condition points are secondary investigation points, or solving points and condition points are knowledge points which need to be investigated on the whole), and is favorable for constructing the corresponding knowledge points.
Illustratively, the sentence classification model herein may be a random forest model. The random forest model has n _ estimators (number of decision trees) of 8, min _ samples _ leaf (number of samples with minimum leaf node) of 3, max _ features _ maximum (option of maximum feature number), max _ depth (maximum depth of decision tree), min _ samples _ split (limiting condition, if the number of samples of the node is less than min _ samples _ split, then no further attempt is made to select the optimal feature for partitioning), min _ weight _ fraction _ leaf (minimum of all sample weight sums of leaf nodes), max _ leaf _ nodes (maximum number of leaf nodes, fitting can be prevented) all set as default values.
Because the sentence classification model is mainly applied to the classification of single sentences, a random forest model is selected as a basis for training, and a satisfactory effect can be achieved. Considering the application scenario of the sentence classification model, the sample size that can be obtained, and other factors, the training mode may adopt a supervised training mode, and the available sample size is mainly limited by the test question size of the subject course, so for university courses with relatively small test question size (for example, university physics, mainly derived from post-school exercises, years testing test questions, and the like), n _ estimators of the random forest model is 8, min _ samples _ leaf is 3, and max _ features, max _ depth, min _ samples _ split, min _ weight _ fraction _ leaf, and max _ leaf _ nodes are all set as default values, which can effectively improve the operation efficiency of the model while ensuring the classification accuracy.
For the lessons with large examination subject quantity (such as the physical lessons of middle school), the sample quantity is large, n _ estimators (for example, adjusted to 12) and min _ samples _ leaf (for example, adjusted to 4) can be mainly adjusted, and correspondingly, max _ depth can be correspondingly adjusted (set according to specific situations), and the maximum depth (for example, 4 or 6) is set. Other parameters, such as max _ features, min _ samples _ split, min _ weight _ fraction _ leaf, max _ leaf _ nodes, etc., can be set as default values, so as to obtain a short and fat decision tree as much as possible, thereby improving classification accuracy and model operation efficiency.
After determining that the question stem of the test question contains the target single sentence and the auxiliary single sentence for solving the requirement, the electronic equipment can perform entity extraction, attribute extraction and relation extraction on the target single sentence to construct a solution point of the test question. And for the auxiliary clause of the question stem of the test question, entity extraction, attribute extraction and relationship extraction can be carried out to construct the condition point of the test question.
The entity extraction, the attribute extraction and the relation extraction can realize the extraction of a target single sentence or an auxiliary single sentence by utilizing the existing construction model of the knowledge graph, and as for the construction of the ontology model, the entity extraction, the attribute extraction and the relation extraction can be constructed according to the actual situation: for example, a rigid body as an entity; the rotating shaft of the rigid body is used as an attribute of the entity, the moment is an attribute of the rigid body, and the rotational kinetic energy of the moment is also an attribute of the rigid body; the relationship may be expressed by a simple dependency relationship, which is only an example and is not limited herein.
The solution points and condition points of the test question can be constructed by using the ontology model in the knowledge graph to obtain solution points and condition points with similar structures (having entities, attributes and relationships).
After the solution points and the condition points of the test question are constructed, the electronic equipment can match the attributes of the solution points and the condition points of the test question with the attributes of the knowledge graph of the curriculum constructed in advance, and construct the knowledge points of the test question based on the matching result.
In this embodiment, for each solution point of the test question, the electronic device may perform consistency matching on the attribute in the solution point of the test question and the attribute in the knowledge graph of the course, and determine the matched attribute as the main attribute of the corresponding entity in the solution point.
That is, the electronic device may match the attributes of the solution point with the attributes in the knowledge-graph of the course (consistency matching, i.e., determining whether the attributes of the solution point are in the attributes of the knowledge-graph), and determine the matched attributes as the primary attributes of the corresponding entity in the solution point.
For each condition point of the test question, the electronic device may consistency-match attributes in the condition point with attributes in the knowledge-graph of the course. If the entity in the condition point is consistent with any entity in the solution point, determining the matched attribute as the secondary attribute of the corresponding entity in the solution point; and if the entity in the condition point is inconsistent with all the entities in the solution point, determining the matched attribute as the condition attribute of the corresponding entity in the condition point.
It should be noted that the knowledge graph of the course herein may be constructed after the ontology model is established, the construction method may be constructed by using an existing relatively mature knowledge graph construction method, for example, a top-down construction method, and the specifically extracted content may be stored by using the graph database Neo4j to obtain the knowledge graph of the course, which is not limited herein.
Of course, if some entities and attributes are extracted as irrelevant (e.g. for the purpose of introducing the background of the subject, e.g. the subject begins with … …. the subject is followed by the body of the subject), then the attributes of these entities are difficult to match with the attributes in the knowledge-graph, and can be discarded to eliminate the interference.
And for each solution point of the test question, carrying out consistency matching on the attribute in the solution point of the test question and the attribute in the knowledge graph of the course, determining that the matched attribute is the main attribute of the corresponding entity in the solution point, carrying out consistency matching on the attribute in the condition point and the attribute in the knowledge graph of the course, if the entity in the condition point is consistent with any entity in the solution point, determining that the matched attribute is the secondary attribute of the corresponding entity in the solution point, and if the entity in the condition point is inconsistent with all entities in the solution point, determining that the matched attribute is the condition attribute of the corresponding entity in the condition point. Therefore, the entities of each solution point and each condition point can be found, the corresponding attributes are integrated, the same attributes of the corresponding entities are integrated under the same entity (the entity of the solution point is preferred), the main attribute and the auxiliary attribute of the corresponding entity in the solution point and the condition attribute of the corresponding entity in the condition point are obtained, and a knowledge point which is mutually related and has clear veins between the solution point and the condition point is constructed. Therefore, the accuracy of the knowledge points in the test questions can be ensured, and the accuracy of recommending the subsequent learning content is facilitated.
After determining the primary attribute and the secondary attribute of the entity corresponding to the solution point and the condition attribute of the entity corresponding to the condition point, the electronic device may construct the knowledge point of the solution point and the condition point of the test question, which are related to each other, based on the primary attribute and the secondary attribute of the entity in each solution point and the condition attribute of the entity in each condition point.
For example, the electronic device may build a connection between entities of different solution points, build a connection between entities of different condition points, and build a connection between an entity of a solution point and an entity of a condition point; and establishing a relation between the main attribute and the auxiliary attribute of the entity in each solution point and the solution point, and establishing a relation between the condition attribute of the entity in each condition point and the condition point to construct a knowledge point of the test question.
This allows the relationship between the entities in the solution points and condition points to be established such that the established knowledge points of the test question can have a very clear correlation between knowledge (e.g., the fixed axis and moment of the rigid body a, the fixed axis of the rigid body B, the rotational kinetic energy of the moment of the rigid body B, etc., and the rigid body a and the rigid body B are coaxial).
After the knowledge points of each test question of the test paper are established, the electronic equipment can establish a knowledge point distribution network of the test paper according to the knowledge points of each test question of the test paper.
In this embodiment, for a knowledge point of each test question of the test paper, the electronic device may determine a subordinate small point, a subordinate small section, and a subordinate chapter of the knowledge point. For example, the first point (rigid body motion) in chapter 4 (rigid body's dead axle rotation) section 1 (rigid body motion description) of college physical course. Still alternatively, the person teaches the 2 nd point (measurement of mass) of section 1 (mass) of chapter 6 (mass and density) of the eight-year-scale physical book of the edition.
Then, the electronic device may construct a tree network according to the affiliated dots, affiliated sections, and affiliated chapters of each knowledge point, so as to obtain the knowledge point distribution network of the test paper, where each branch node of the tree network has a text box, and the display content of the text box includes an item quantity, and the item quantity represents the number of test items of the branch node.
The knowledge point distribution network of the test paper constructed in the way has a very clear and clear knowledge point relation network, so that the knowledge points of each test question of the test paper can be clearly disclosed, the distribution of the knowledge points inspected by the test paper can be reflected, and the output of the statistical results of subsequent pictures is facilitated.
In the embodiment, the question stem may not be distinguished from the choice of the question stem (i.e., the choice may be matched with the question stem). Of course, in order to express the examination points of the selection questions more accurately, the options can be processed separately, and the processed results can be integrated into the knowledge points of the test questions, so that on one hand, the knowledge points of the test questions can be expressed more completely, and on the other hand, the statistics of the options can be performed, so that the students (or the whole students, such as students in a class, students in a year, and the like) can be better analyzed to select a certain option to reflect the knowledge mastering level of the selected option.
Based on this, in order to accurately recommend learning content to students, the electronic device may perform step S10.
Step S10: and acquiring a test answer sheet of the student, wherein the test answer sheet is an electronic answer sheet of the test paper.
In this embodiment, the electronic device may obtain a test answer sheet generated by a student learning a course on a system (an artificial intelligence-based learning content recommendation system) and taking a test of the course. The test answer sheet is an electronic answer sheet of the test paper.
After obtaining the test answer sheet, the electronic device may perform step S20.
Step S20: and acquiring a knowledge point distribution network corresponding to the test answer sheet, counting the score condition of each knowledge point in the test answer sheet based on the knowledge point distribution network, and outputting a graph statistical result.
In this embodiment, the manner for the electronic device to obtain the knowledge point distribution network corresponding to the test answer sheet may be as follows: the electronic equipment determines the knowledge point distribution network corresponding to the corresponding test paper according to the test answer sheet, so that the knowledge point distribution network corresponding to the test answer sheet can be obtained.
Then, the electronic device can count the score condition of each knowledge point in the test answer sheet based on the knowledge point distribution network, and output the graph statistical result.
For each test question in the test answer sheet, the electronic device may determine a total score and a score of the test question, record the total score and the score of the test question in a corresponding branch node of the knowledge point distribution network, so that the corresponding branch node may perform statistics based on all the total scores and the scores on the branch node, and output a visual knowledge point distribution network showing the total score and the score through a text box on the branch node, where the visual knowledge point distribution network is a graph-like statistical result.
The method can accurately count the answer condition of the student, correspondingly count and output the score condition of the knowledge points, and obtain the knowledge point mastering condition of the student, so that the learning content is recommended in a targeted manner, the accuracy of pushing the learning content is guaranteed, and the learning efficiency and the knowledge mastering level of the student are effectively improved.
After obtaining the visualized knowledge point distribution network (i.e., the graph-like statistics), the electronic device may execute step S30.
Step S30: and determining knowledge points to be promoted of the student according to the statistical result of the graph.
In this embodiment, the electronic device may determine a first-class branch node, a second-class branch node, and a third-class branch node according to a score ratio between a score value and a total score value of each branch node at the bottom layer in the visual knowledge point distribution network, where the branch node at the bottom layer represents a branch node corresponding to a corresponding small point of a knowledge point, the first-class branch node represents that the knowledge point on the branch node does not need to be complemented, the second-class branch node represents that the knowledge point on the branch node needs to be advanced, and the third-class branch node represents that the knowledge point on the branch node needs to be basic-tamped.
For example, if the score ratio between the score value of the branch node and the total score value is more than 85% (or 90%), the branch node may be determined to be a type of branch node; if the score ratio between the score value and the total score value of the branch node is between 60 and 84 percent (or between 60 and 90 percent), the branch node can be determined to be a second type of branch node; if the score ratio between the score value of the branch node and the total score value is below 60%, the branch node can be determined to be a three-class branch node.
After the first-class branch node, the second-class branch node and the third-class branch node are determined, the electronic device can determine the knowledge points corresponding to the second-class branch node and the third-class branch node as the knowledge points to be improved.
Classifying the branch nodes (branch nodes corresponding to the small points of the knowledge points) at the bottommost layer (first class branch nodes, second class branch nodes and third class branch nodes) by utilizing the score ratio between the score value and the total score of each branch node at the bottommost layer in the visual knowledge point distribution network, determining the knowledge points corresponding to the second class branch nodes and the third class branch nodes as the knowledge points to be promoted, and using the knowledge points as conditions for determining the learning content needing to be pushed so as to screen out the learning content which is most beneficial to the learning of the student.
After determining the knowledge point to be improved, the electronic device may perform step S40.
Step S40: and determining the learning content for learning and/or testing the knowledge points to be promoted according to the knowledge points to be promoted, and pushing the learning content to the student for learning.
In this embodiment, the electronic device may determine, according to the knowledge point to be improved, learning content for learning and/or testing the knowledge point to be improved, and push the learning content to the student for learning.
For example, the electronic device may match the knowledge points to be promoted with the knowledge points of the test questions in the question bank, and may also match the knowledge points to be promoted with the knowledge point tags of the analysis content. And determining the matched test questions and/or the analysis content. Of course, the learning content may be one or both of the test question and the analysis content, and of course, other types of learning content may also be included, which is not limited herein. Then, the electronic equipment can recommend the matched test questions and/or the analysis content to the terminal of the student, so that the student can learn the part which is poorly mastered by the student in a targeted manner, and a better learning effect is achieved.
Referring to fig. 2, based on the same inventive concept, the present application further provides an artificial intelligence based learning content recommendation system 10, wherein all students of the same course learn the course and participate in the test of the course on the system 10, and the system 10 includes:
the test answer sheet obtaining unit 11 is configured to obtain a test answer sheet of a student, where the test answer sheet is an electronic answer sheet of the test paper.
And the statistical result output unit 12 is configured to acquire a knowledge point distribution network corresponding to the test answer sheet, perform statistics on the score condition of each knowledge point in the test answer sheet based on the knowledge point distribution network, and output a graph statistical result.
And the knowledge point to be improved determining unit 13 is used for determining the knowledge point to be improved of the student according to the graph statistical result.
And the learning content recommending unit 14 is configured to determine, according to the knowledge points to be improved, learning content for learning and/or testing the knowledge points to be improved, and push the learning content to the student for learning.
In this embodiment, the system further includes a knowledge point distribution network unit, configured to, before the test answer sheet obtaining unit 11 obtains the test answer sheets of the students, for each set of test answer sheets: processing the question stem of each test question of the test paper, and determining that the question stem of the test question contains a target single sentence for solving the requirement; performing entity extraction, attribute extraction and relation extraction on the target single sentence to construct a solution point of the test question; carrying out entity extraction, attribute extraction and relation extraction on auxiliary single sentences except the target single sentence in the question stem of the test question to construct condition points of the test question; matching attributes in the solution points and the condition points of the test questions with attributes in a knowledge graph of the curriculum which is constructed in advance, and constructing knowledge points of the test questions based on matching results; constructing a knowledge point distribution network of the test paper according to the knowledge points of each test question of the test paper; correspondingly, the statistical result output unit 12 is specifically configured to determine, according to the test answer sheet, a knowledge point distribution network corresponding to the test paper corresponding to the test answer sheet.
In this embodiment, the knowledge point distribution network unit is specifically configured to: for each test question, sentence breaking processing is carried out on the question stem of the test question to determine a plurality of single sentences to be defined; and inputting each single sentence to be defined into a preset sentence classification model, and determining the category of the single sentence to be defined, wherein the category of the single sentence to be defined is the target single sentence or the auxiliary single sentence.
In this embodiment, the sentence classification model is a random forest model, n _ estimators of the random forest model is 8, min _ samples _ leaf is 3, and max _ features × max _ depth, min _ samples _ split, min _ weight _ fraction _ leaf, and max _ leaf _ nodes are all set as default values.
In this embodiment, the knowledge point distribution network unit is specifically configured to: for each solution point of the test question, carrying out consistency matching on the attributes in the solution points of the test question and the attributes in the knowledge graph of the course, and determining the matched attributes as the main attributes of corresponding entities in the solution points; for each condition point of the test question, consistency matching is carried out on attributes in the condition point and attributes in a knowledge graph of the course, if an entity in the condition point is consistent with any entity in the solution point, the matched attributes are determined to be secondary attributes of corresponding entities in the solution point, and if the entity in the condition point is not consistent with all entities in the solution point, the matched attributes are determined to be conditional attributes of corresponding entities in the condition point; and constructing the correlative knowledge points between the solution points and the condition points of the test question based on the main attribute and the auxiliary attribute of the entity in each solution point and the condition attribute of the entity in each condition point.
In this embodiment, the knowledge point distribution network unit is specifically configured to: establishing the relation between the entities with different solution points, the relation between the entities with different condition points and the relation between the entities with solution points and the entities with condition points; and establishing a relation between the main attribute and the auxiliary attribute of the entity in each solution point and the solution point, and establishing a relation between the condition attribute of the entity in each condition point and the condition point to construct a knowledge point of the test question.
In this embodiment, the knowledge point distribution network unit is specifically configured to: determining the subordinate dots, the subordinate nodes and the subordinate chapters of the knowledge points according to the knowledge points of each test question of the test paper; and constructing a tree network according to the affiliated dots, the affiliated sections and the affiliated chapters of each knowledge point to obtain a knowledge point distribution network of the test paper, wherein each branch node of the tree network is provided with a text box, the display content of the text box comprises question amount, and the question amount represents the number of the test questions of the branch node.
In this embodiment, the statistical result output unit 12 is specifically configured to: and determining the total score and the score of each test question in the test answer sheet, recording the total score and the score of each test question in a corresponding branch node of the knowledge point distribution network, so that the corresponding branch node can perform statistics based on all the total scores and the score of the branch node, and outputting a visual knowledge point distribution network displaying the total score and the score through a text box on the branch node, wherein the visual knowledge point distribution network is the graph statistical result.
In this embodiment, the to-be-promoted knowledge point determining unit 13 is specifically configured to: determining a first class of branch nodes, a second class of branch nodes and a third class of branch nodes according to a score ratio between a score value and a total score value of each branch node at the bottommost layer in the visual knowledge point distribution network, wherein the branch node at the bottommost layer represents the branch node corresponding to the small point of the knowledge point, the first class of branch nodes represent that the knowledge point on the branch node does not need to be subjected to additional study, the second class of branch nodes represent that the knowledge point on the branch node needs to be subjected to advanced promotion, and the third class of branch nodes represent that the knowledge point on the branch node needs to be subjected to basic compaction; and determining the knowledge points corresponding to the second class of branch nodes and the third class of branch nodes as the knowledge points to be improved.
In summary, the embodiment of the present application provides a learning content recommendation method and system based on artificial intelligence, which output a graph statistical result by obtaining a test answer sheet (electronic answer sheet) of a student, a knowledge point distribution network corresponding to the test answer sheet, and performing statistics on the score condition of each knowledge point in the test answer sheet based on the knowledge point distribution network. The output of picture class statistical result can reflect for student itself (when the student is middle school student, can also reflect for mr) to make the student know the knowledge of oneself and master the condition (when the student is middle school student, can also be convenient for mr to know student's knowledge and master the condition, feed back to self, carry out the regulation of the mode of imparting knowledge to students and the supplementary teaching of course content). Then determining knowledge points to be promoted of the student according to the statistical result of the graphs; therefore, the learning content for learning and/or testing the knowledge points to be promoted is determined and pushed to the student for learning. Therefore, automatic and accurate pushing of the learning content can be achieved, manpower and material resources are saved, artificial deviation (such as whether a teacher has responsibility to supervise learning of students, whether the teacher has enough responsibility and whether the teacher has eccentricity) can be overcome, artificial careless omission (neglect of knowledge points needing to be reviewed due to carelessness) can be overcome in the mode, more valuable learning content is pushed to the students, the learning efficiency of the students is improved, and the mode does not need to be committed by the teacher, and is very practical and effective.
In the embodiments provided in the present application, it should be understood that the disclosed system and method may be implemented in other ways. The above-described device embodiments are merely illustrative, and for example, the division of the units is only one logical function division, and other division manners may be available in actual implementation.
In this document, relational terms such as first and second, and the like may be used solely to distinguish one entity or action from another entity or action without necessarily requiring or implying any actual such relationship or order between such entities or actions.
The above description is only an example of the present application and is not intended to limit the scope of the present application, and various modifications and changes may be made to the present application by those skilled in the art. Any modification, equivalent replacement, improvement and the like made within the spirit and principle of the present application shall be included in the protection scope of the present application.
Claims (9)
1. A learning content recommendation method based on artificial intelligence is characterized in that all students of the same course learn the course and take part in the test of the course on the same system, and the method is applied to the system and comprises the following steps:
acquiring a test answer sheet of a student, wherein the test answer sheet is an electronic answer sheet of the test paper;
acquiring a knowledge point distribution network corresponding to the test answer sheet, counting the score condition of each knowledge point in the test answer sheet based on the knowledge point distribution network, and outputting a graph statistical result;
determining knowledge points to be promoted of the student according to the graph statistical result;
determining learning contents for learning and/or testing the knowledge points to be promoted according to the knowledge points to be promoted, and pushing the learning contents to the students for learning;
prior to said obtaining a test answer sheet for a student, said method further comprises:
for each set of test paper:
processing the question stem of each test question of the test paper, and determining that the question stem of the test question contains a target single sentence for solving the requirement;
performing entity extraction, attribute extraction and relation extraction on the target single sentence to construct a solution point of the test question;
carrying out entity extraction, attribute extraction and relation extraction on auxiliary single sentences except the target single sentence in the question stem of the test question to construct condition points of the test question;
matching the attributes in the solution points and the condition points of the test questions with the attributes in the pre-constructed knowledge graph of the curriculum, and constructing the knowledge points of the test questions based on the matching results;
constructing a knowledge point distribution network of the test paper according to the knowledge points of each test question of the test paper;
correspondingly, the acquiring the knowledge point distribution network corresponding to the test answer sheet includes: and determining a knowledge point distribution network corresponding to the corresponding test paper according to the test answer sheet.
2. The artificial intelligence based learning content recommendation method according to claim 1, wherein said processing the stem of each test question of the test paper comprises:
for each test question, sentence breaking processing is carried out on the question stem of the test question to determine a plurality of single sentences to be defined;
and inputting each single sentence to be defined into a preset sentence classification model, and determining the category of the single sentence to be defined, wherein the category of the single sentence to be defined is the target single sentence or the auxiliary single sentence.
3. The artificial intelligence based learning content recommendation method according to claim 2, wherein the sentence classification model is a random forest model, n _ estimators of the random forest model is 8, min _ samples _ leaf is 3, max _ features, max _ depth, min _ samples _ split, min _ weight _ fraction _ leaf, and max _ leaf _ nodes are all set as default values.
4. The artificial intelligence based learning content recommendation method according to claim 1, wherein the matching of the attributes in the solution point and the condition point of the question with the attributes in the pre-constructed knowledge graph of the lesson and the construction of the knowledge point of the question based on the matching result comprises:
for each solution point of the test question, carrying out consistency matching on the attribute in the solution point of the test question and the attribute in the knowledge graph of the course, and determining the matched attribute as the main attribute of the corresponding entity in the solution point;
for each condition point of the test question, carrying out consistency matching on the attributes in the condition point and the attributes in the knowledge map of the course, if the entity in the condition point is consistent with any entity in the solution point, determining the matched attributes as the secondary attributes of the corresponding entities in the solution point, and if the entity in the condition point is not consistent with all the entities in the solution point, determining the matched attributes as the condition attributes of the corresponding entities in the condition point;
and constructing the knowledge points of the mutual correlation between the solution points and the condition points of the test question based on the main attribute and the auxiliary attribute of the entity in each solution point and the condition attribute of the entity in each condition point.
5. The artificial intelligence based learning content recommendation method according to claim 4, wherein the constructing of the knowledge points of the question that are correlated between the solution points and the condition points based on the primary attributes and the secondary attributes of the entities in each solution point and the condition attributes of the entities in each condition point comprises:
establishing the relation between the entities with different solution points, the relation between the entities with different condition points and the relation between the entities with solution points and the entities with condition points;
and establishing a relation between the main attribute and the auxiliary attribute of the entity in each solution point and the solution point, and establishing a relation between the condition attribute of the entity in each condition point and the condition point to construct a knowledge point of the test question.
6. The artificial intelligence based learning content recommendation method of claim 1, wherein the constructing a knowledge point distribution network of the test paper according to the knowledge points of each test question of the test paper comprises:
determining the subordinate dots, the subordinate nodes and the subordinate chapters of the knowledge points according to the knowledge points of each test question of the test paper;
and constructing a tree network according to the affiliated dots, the affiliated sections and the affiliated chapters of each knowledge point to obtain a knowledge point distribution network of the test paper, wherein each branch node of the tree network is provided with a text box, the display content of the text box comprises question amount, and the question amount represents the number of the test questions of the branch node.
7. The artificial intelligence based learning content recommendation method according to claim 6, wherein the counting score of each knowledge point in the test answer sheet based on the knowledge point distribution network and outputting graph statistics result comprises:
and determining the total score and the score of each test question in the test answer sheet, recording the total score and the score of each test question in a corresponding branch node of the knowledge point distribution network, so that the corresponding branch node can perform statistics based on all the total scores and the score of the branch node, and outputting a visual knowledge point distribution network displaying the total score and the score through a text box on the branch node, wherein the visual knowledge point distribution network is the graph statistical result.
8. The artificial intelligence based learning content recommendation method according to claim 7, wherein the determining the knowledge points to be improved of the student according to the graph statistics result comprises:
determining a first class of branch nodes, a second class of branch nodes and a third class of branch nodes according to a score ratio between a score value and a total score value of each branch node at the bottommost layer in the visual knowledge point distribution network, wherein the branch node at the bottommost layer represents the branch node corresponding to the corresponding small point of the knowledge point, the first class of branch nodes represent that the knowledge point on the branch node does not need to be subjected to learning, the second class of branch nodes represent that the knowledge point on the branch node needs to be subjected to advanced promotion, and the third class of branch nodes represent that the knowledge point on the branch node needs to be subjected to basic tamping;
and determining the knowledge points corresponding to the second class of branch nodes and the third class of branch nodes as the knowledge points to be improved.
9. An artificial intelligence based learning content recommendation system, wherein all students of a same course learn the course and participate in a test of the course on the system, the system comprising:
the system comprises a test answer sheet acquisition unit, a test answer sheet processing unit and a test result processing unit, wherein the test answer sheet acquisition unit is used for acquiring test answer sheets of students, and the test answer sheets are electronic answer sheets of the test answer sheets;
the statistical result output unit is used for acquiring a knowledge point distribution network corresponding to the test answer sheet, counting the score condition of each knowledge point in the test answer sheet based on the knowledge point distribution network, and outputting a graph statistical result;
the to-be-promoted knowledge point determining unit is used for determining the to-be-promoted knowledge points of the students according to the graph statistical result;
the learning content recommending unit is used for determining the learning content for learning and/or testing the knowledge points to be improved according to the knowledge points to be improved and pushing the learning content to the student for learning;
the system further comprises a knowledge point distribution network unit, which is used for aiming at each set of test paper before the test paper acquiring unit acquires the test paper of the student: processing the question stem of each test question of the test paper to determine a target single sentence containing a solving requirement in the question stem of the test question; performing entity extraction, attribute extraction and relationship extraction on the target single sentence to construct a solution point of the test question; carrying out entity extraction, attribute extraction and relationship extraction on auxiliary single sentences except the target single sentence in the question stem of the test question to construct condition points of the test question; matching the attributes in the solution points and the condition points of the test questions with the attributes in the pre-constructed knowledge graph of the curriculum, and constructing the knowledge points of the test questions based on the matching results; constructing a knowledge point distribution network of the test paper according to the knowledge points of each test question of the test paper; correspondingly, the statistical result output unit is specifically configured to determine, according to the test answer sheet, a knowledge point distribution network corresponding to the test paper.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110629607.6A CN113254629B (en) | 2021-06-07 | 2021-06-07 | Learning content recommendation method and system based on artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202110629607.6A CN113254629B (en) | 2021-06-07 | 2021-06-07 | Learning content recommendation method and system based on artificial intelligence |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN113254629A CN113254629A (en) | 2021-08-13 |
| CN113254629B true CN113254629B (en) | 2022-07-26 |
Family
ID=77186664
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202110629607.6A Active CN113254629B (en) | 2021-06-07 | 2021-06-07 | Learning content recommendation method and system based on artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN113254629B (en) |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106779166A (en) * | 2016-11-23 | 2017-05-31 | 北京师范大学 | Grasp the forecasting system and method for state in a kind of knowledge point based on data-driven |
| CN107273490A (en) * | 2017-06-14 | 2017-10-20 | 北京工业大学 | A kind of combination mistake topic recommendation method of knowledge based collection of illustrative plates |
| CN109255998A (en) * | 2018-10-25 | 2019-01-22 | 杭州数理大数据技术有限公司 | A kind of student performance diagnosis lifting system |
| CN110162713A (en) * | 2019-05-31 | 2019-08-23 | 成都鼎晟数智科技有限公司 | Adaptive learning content recommendation method and system based on convolutional neural networks |
| CN110929020A (en) * | 2019-12-18 | 2020-03-27 | 四川文轩教育科技有限公司 | Knowledge point mastery degree analysis method based on test achievement |
| CN111274411A (en) * | 2020-01-22 | 2020-06-12 | 文思海辉智科科技有限公司 | Course recommendation method and device, electronic equipment and readable storage medium |
| CN111753098A (en) * | 2020-06-23 | 2020-10-09 | 陕西师范大学 | Teaching method and system based on cross-media dynamic knowledge graph |
| CN112507140A (en) * | 2021-02-08 | 2021-03-16 | 深圳市阿卡索资讯股份有限公司 | Personalized intelligent learning recommendation method, device, equipment and storage medium |
| CN112765310A (en) * | 2020-12-11 | 2021-05-07 | 北京航天云路有限公司 | Knowledge graph question-answering method based on deep learning and similarity matching |
| CN112860983A (en) * | 2019-11-27 | 2021-05-28 | 上海流利说信息技术有限公司 | Learning content pushing method, system, equipment and readable storage medium |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105355111A (en) * | 2015-12-02 | 2016-02-24 | 华中师范大学 | After-class reinforced learning system based on learning situation analysis |
| CN108897828A (en) * | 2018-06-21 | 2018-11-27 | 广东小天才科技有限公司 | A kind of knowledge system construction method and electronic equipment |
-
2021
- 2021-06-07 CN CN202110629607.6A patent/CN113254629B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106779166A (en) * | 2016-11-23 | 2017-05-31 | 北京师范大学 | Grasp the forecasting system and method for state in a kind of knowledge point based on data-driven |
| CN107273490A (en) * | 2017-06-14 | 2017-10-20 | 北京工业大学 | A kind of combination mistake topic recommendation method of knowledge based collection of illustrative plates |
| CN109255998A (en) * | 2018-10-25 | 2019-01-22 | 杭州数理大数据技术有限公司 | A kind of student performance diagnosis lifting system |
| CN110162713A (en) * | 2019-05-31 | 2019-08-23 | 成都鼎晟数智科技有限公司 | Adaptive learning content recommendation method and system based on convolutional neural networks |
| CN112860983A (en) * | 2019-11-27 | 2021-05-28 | 上海流利说信息技术有限公司 | Learning content pushing method, system, equipment and readable storage medium |
| CN110929020A (en) * | 2019-12-18 | 2020-03-27 | 四川文轩教育科技有限公司 | Knowledge point mastery degree analysis method based on test achievement |
| CN111274411A (en) * | 2020-01-22 | 2020-06-12 | 文思海辉智科科技有限公司 | Course recommendation method and device, electronic equipment and readable storage medium |
| CN111753098A (en) * | 2020-06-23 | 2020-10-09 | 陕西师范大学 | Teaching method and system based on cross-media dynamic knowledge graph |
| CN112765310A (en) * | 2020-12-11 | 2021-05-07 | 北京航天云路有限公司 | Knowledge graph question-answering method based on deep learning and similarity matching |
| CN112507140A (en) * | 2021-02-08 | 2021-03-16 | 深圳市阿卡索资讯股份有限公司 | Personalized intelligent learning recommendation method, device, equipment and storage medium |
Non-Patent Citations (2)
| Title |
|---|
| "基于属性关系深度挖掘的试题知识点标注模型";何彬等;《南京信息工程大学学报：自然科学版》;20190630(第2019年第6期);727-734 * |
| "基于知识点的学习内容个性化推荐研究";胡晓楠;《中国优秀硕士学位论文全文数据库 (信息科技辑)》;20110315(第2011年03期);I138-1585 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN113254629A (en) | 2021-08-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Engeln et al. | Inquiry-based learning in mathematics and science: A comparative baseline study of teachers’ beliefs and practices across 12 European countries | |
| Csapó et al. | Computer-based assessment of school readiness and early reasoning. | |
| Goh et al. | The approaches to teaching inventory: A preliminary validation of the Malaysian translation | |
| Lee et al. | Teachers' scientific epistemological views, conceptions of teaching science, and their approaches to teaching science: An exploratory study of inservice science teachers in Taiwan | |
| Laukkanen | Comparative causal mapping and CMAP3 software in qualitative studies | |
| Kim | Literacy skills gaps: A cross-level analysis on international and intergenerational variations | |
| Akerlof et al. | The growth and disciplinary convergence of environmental communication: A bibliometric analysis of the field (1970–2019) | |
| Mühling | Investigating knowledge structures in computer science education | |
| Frischemeier | Primary school students’ reasoning when comparing groups using modal clumps, medians, and hatplots | |
| Song et al. | Closing the circle: Use of students’ responses for peer-assessment rubric improvement | |
| Cook-Wallace | Who Is Running Online Education Programs? | |
| Jayakody et al. | Enhancing competencies of less-able students to achieve learning outcomes: Learner aware tool support through Business intelligence | |
| CN113254629B (en) | Learning content recommendation method and system based on artificial intelligence | |
| de la Chica et al. | Computational foundations for personalizing instruction with digital libraries | |
| Sejati et al. | Digital native levels of Indonesia pre-service teachers | |
| Arlitt et al. | Feature engineering for design thinking assessment | |
| Lemantara et al. | An Innovative Strategy to Anticipate Students’ Cheating: The Development of Automatic Essay Assessment on the “MoLearn” Learning Management System | |
| Veeramanickam et al. | Smart education system to improve the learning system with CBR based recommendation system using IoT | |
| Ayeni et al. | Web-based student opinion mining system using sentiment analysis | |
| Altınırmak et al. | Implementation of the fuzzy DEMATEL method in higher education course selection: the case of Eskişehir vocational school | |
| Hassan et al. | UPSI Students’ Perceptions on the Use of ICT in Learning: Comparison Between Knowledge Test and Task-based Test on ICT Competency | |
| Muchlisin et al. | An analysis of Javanese language test characteristic using the Rasch model in R program | |
| Nuraeni et al. | Impact of quantitative literacy on student reasoning in plant anatomy course | |
| Lipson | 12. WHAT DO STUDENTS GAIN FROM SIMULATION EXERCISES? AN EVALUATION OF ACTIVITIES DESIGNED TO DEVELOP AN UNDERSTANDING OF THE SAMPLING DISTRIBUTION OF A PROPORTION | |
| JP2005331615A (en) | Sentence evaluation device and method |
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 |