CN117474353A

CN117474353A - Decision automatic generation method and device based on online education

Info

Publication number: CN117474353A
Application number: CN202311412141.XA
Authority: CN
Inventors: 宋丽哲; 刘兴国; 薛海峰; 陈守刚; 李静; 李雅静
Original assignee: OPEN UNIVERSITY OF CHINA
Current assignee: OPEN UNIVERSITY OF CHINA
Priority date: 2023-10-27
Filing date: 2023-10-27
Publication date: 2024-01-30

Abstract

The application provides an automatic decision generation method and device based on online education, which relate to the technical field of education decision, and the method comprises the following steps: determining a decision target according to the online education service problem input by the user; collecting decision support data according to the decision target; the decision support data represents personalized information of an object to be decided corresponding to the online education service problem, and the personalized information at least comprises: the online education platform, the educational administration management platform and the recruitment management platform are used for enabling behavior data, learning data and social data of the object to be decided; determining a proper data analysis method from a preset relation mapping table according to the relation between the decision target and the decision support data; analyzing the decision support data according to the data analysis method to generate a target decision; and visually displaying the target decision.

Description

Decision automatic generation method and device based on online education

Technical Field

The application relates to the technical field of education decision making, in particular to an automatic decision making generation method and device based on online education.

Background

In all education management activities, education decision is core and key place, which not only affects the efficiency and effect of education management work, but also concerns individual development of students and teachers, and even the decay of the whole education industry. At present, the technical means are applied to assist education decisions, so that more importance is placed on the education digitization transformation, and higher requirements are placed on scientific decisions.

However, in the process of digital transformation of education, the current scientific decision-making is not a mature and effective method, and an educational decision-maker cannot comprehensively and accurately know the learning condition and the demand of students directly through data, so that scientific and reliable educational decision-making is difficult to obtain. Therefore, there is a need to develop an automatic decision generation method and apparatus based on online education to realize scientific and effective education decisions.

Disclosure of Invention

In view of the above, embodiments of the present application provide a method and apparatus for automatically generating decisions based on online education, so as to overcome or at least partially solve the above-mentioned problems.

In a first aspect of an embodiment of the present application, there is provided an automatic decision generation method based on online education, including:

determining a decision target according to the online education service problem input by the user;

Collecting decision support data according to the decision target; the decision support data represents personalized information of an object to be decided corresponding to the online education service problem, and the personalized information at least comprises: the online education platform, the educational administration management platform and the recruitment management platform are used for enabling behavior data, learning data and social data of the object to be decided;

determining a proper data analysis method from a preset relation mapping table according to the relation between the decision target and the decision support data;

analyzing the decision support data according to the data analysis method to generate a target decision;

and visually displaying the target decision.

The second aspect of the embodiment of the application also provides an automatic decision generating device based on online education, which comprises:

the decision target determining module is used for determining a decision target according to the online education service problem input by the user;

the data acquisition module is used for acquiring decision support data according to the decision target; the decision support data represents personalized information of an object to be decided corresponding to the online education service problem, and the decision support data at least comprises: behavior data, learning data and social data of the object to be decided in the online education platform;

The data analysis method determining module is used for determining a proper data analysis method from a preset relation mapping table according to the relation between the decision target and the decision support data;

the data analysis module is used for analyzing the decision support data according to the data analysis method to generate a target decision;

and the visualization module is used for carrying out visual display on the target decision.

The third aspect of the embodiment of the application further provides an electronic device, which comprises a memory, a processor and a computer program stored on the memory, wherein the processor executes the computer program to realize the steps in the automatic decision generation method based on online education according to the first aspect of the embodiment of the application.

The fourth aspect of the embodiments of the present application further provides a computer readable storage medium, on which a computer program/instruction is stored, which when executed by a processor, implements the steps in the automatic online education-based decision generation method according to the first aspect of the embodiments of the present application.

A fifth aspect of the embodiments of the present application also provides a computer program product, which when run on an electronic device, causes a processor to perform the steps in the automatic online education-based decision generation method according to the first aspect of the embodiments of the present application.

The embodiment of the application provides an automatic decision generation method and device based on online education, wherein the method comprises the following steps: determining a decision target according to the online education service problem input by the user; collecting decision support data according to the decision target; the decision support data represents personalized information of an object to be decided corresponding to the online education service problem, and the personalized information at least comprises: the online education platform, the educational administration management platform and the recruitment management platform are used for enabling behavior data, learning data and social data of the object to be decided; determining a proper data analysis method from a preset relation mapping table according to the relation between the decision target and the decision support data; analyzing the decision support data according to the data analysis method to generate a target decision; and visually displaying the target decision.

The concrete beneficial effects are that:

according to the method, multiple data such as behavior data, learning data and social data of students are collected on multiple platforms (an online education platform, a educational administration management platform and a recruitment management platform), multidimensional analysis and mining are carried out on the data, so that target decisions are generated, and the accuracy and pertinence of the generated target decisions are improved. In addition, through analysis of student data (namely personalized information of an object to be decided), personalized decisions of each student, such as personalized course recommendation, personalized learning planning and the like, can be realized, so that learning effects and satisfaction of the students are improved. In addition, the method reduces the manual intervention and time cost and reduces the decision difficulty through automatic data acquisition and processing.

Drawings

In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings that are needed in the description of the embodiments of the present application will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present application, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.

FIG. 1 is a flow chart of steps of an automatic decision generation method based on online education provided in an embodiment of the present application;

FIG. 2 is a schematic diagram of a decision-making flow of online education provided in an embodiment of the present application;

FIG. 3 is a schematic diagram of a decision-making process for learning effect evaluation according to an embodiment of the present disclosure;

fig. 4 is a schematic structural diagram of an automatic decision generating device based on online education according to an embodiment of the present application;

fig. 5 is a schematic diagram of an electronic device according to an embodiment of the present application.

Detailed Description

Exemplary embodiments of the present application will be described in more detail below with reference to the accompanying drawings in the embodiments of the present application. While exemplary embodiments of the present application are shown in the drawings, it should be understood that the present application may be embodied in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the disclosure to those skilled in the art.

Digital transformation of education is a hotspot problem in current education reform and practice, and the dominant of national policies and the inherent needs of the education system drive the digital transformation of education together. The education is continuously changed from an imitation experience teaching paradigm of the industrialized age to a calculation-aided teaching paradigm of the informatization age to a data-driven teaching paradigm, but for education decision making, the decision making process is too dependent on experience and lacks of data support, and is a shortcoming of quite suffering from scaling in the education industry, big data penetrate into each industry and service field nowadays in the process of digital transformation, and education decision making based on the data brings hopes for solving the deficiency.

In recent years, some achievements are achieved in the research of applying big data to education decisions, but in the process of digitally transforming education, scientific decisions need innovative innovation, and no mature method is seen at present. Under the digital transformation background, advanced data analysis technology and artificial intelligence algorithm are needed to carry out fine analysis on student learning data so as to support more accurate teaching decision.

In view of the above problems, an embodiment of the present application provides a method and an apparatus for automatically generating decisions based on online education, so as to implement scientific and effective educational decisions. The method and the device for automatically generating the decisions based on the online education provided by the embodiment of the application are described in detail below through some embodiments and application scenes thereof with reference to the accompanying drawings.

An embodiment of the present application provides an automatic decision generation method based on online education in a first aspect, referring to fig. 1, fig. 1 is a step flowchart of the automatic decision generation method based on online education provided in the embodiment of the present application, as shown in fig. 1, where the method includes:

step S101, determining a decision target according to the online education service problem input by the user.

Decision targets represent target questions that the user needs to solve or decide. In the process of education decision generation, the primary task is decision targets, namely, the management of decision problems. Based on the questions posed by the user, the goal of the decision is established and then the practice of educational digitization transformation is driven in the question resolution orientation. The various problems in the online educational field are generally grouped together into what happens? "," why will occur? "," what will happen in the future? "," what action should be taken? "and the like. From the perspective of solving the key problems by using a digital analysis method, decision targets can be classified into four types of descriptive, diagnostic, predictive and normative, and the specific type of the decision target can be determined from the on-line education business problems input by a user. Different types of decision targets may choose different analysis methods. In particular, the descriptive analysis method answers "what happens in the past? "by summarizing past data, reporting business conditions and general trends; the diagnostic analysis method answers "why do it? "diagnostic problem, find the root cause of the problem by examining the data and traffic; predictive analysis methods answer "what will happen in the future? "predictive question, using historical data as training data, uses various regression analysis and machine learning techniques to construct a predictive model for predicting what will be a higher probability in the future. The normative analysis method answers "what action should be taken? "normalization problem, focusing on a method of recommending operability, normalization analysis methods collect data from a number of descriptive and predictive sources and apply it to the decision process.

Step S102, acquiring decision support data according to the decision target; the decision support data represents personalized information of an object to be decided corresponding to the online education service problem, and the personalized information at least comprises: and the online education platform, the educational administration management platform and the recruitment management platform are used for carrying out behavior data, learning data and social data on the objects to be decided.

In this embodiment, after a decision target is determined according to a business problem input by a user, the original data capable of supporting decision, i.e., decision support data, is integrated according to the decision target. Under the digital transformation background, determining a decision target and determining an object to be decided corresponding to an online education business problem, such as a student A or a class B, so that personalized information of the object to be decided can be acquired in a targeted manner when data are acquired, subsequent independent analysis of the data of the object to be decided is facilitated, personalized decisions of each individual, such as personalized course recommendation, personalized learning planning and the like, are realized, and learning effects and satisfaction of students are improved.

In this embodiment, the personalized information at least includes: and the online education platform, the educational administration management platform and the recruitment management platform are used for carrying out behavior data, learning data and social data on the objects to be decided. Specifically, the data acquisition of the embodiment in the field of online education is more comprehensive. In the traditional education field data acquisition mainly focuses on examination results, student status information and the like of students, but in the digital transformed online education provided by the embodiment, personalized information with more abundant data types can be acquired, and from a macroscopic point of view, talent demand data related to online education, national education resource allocation situation data and the like can be contained. From a microscopic perspective, data may also be included for educational teaching elements and teaching processes. These data types can provide more comprehensive information, helping educational organizations to better understand the overall profile of the online educational system and the learning situation of students.

Furthermore, the data sources of the present embodiment are more diversified. In the traditional education field, data mainly come from a management system inside a school, and in the digital transformed online education provided by the embodiment, data sources are more diversified, and the digital transformed online education field comprises a plurality of channels such as an online learning platform, a educational administration platform, an recruitment management platform, social media, mobile equipment and the like. The channels provide more data sources, and can more comprehensively know the information such as the learning condition of students.

In addition, the data acquisition of the present embodiment is more refined. In the digital transformation on-line education provided in this embodiment, more refined data collection modes may be adopted, such as collecting student learning behavior data, device state data, student physique data, student life data and the like by using an internet of things sensing technology, collecting campus safety data, classroom teaching data and the like by using a video recording technology, collecting student examination performance data, various homework exercise data, handwriting data such as student course notes and the like by using an image recognition technology, and collecting various on-line teaching and management data, mobile learning process data, operation and maintenance logs and user log data, network public opinion data and the like by using a platform collection technology.

In a possible implementation manner, the step S102, according to the decision target, collects decision support data, including:

and step S1021, making a data acquisition scheme according to the decision target.

In this embodiment, after determining the decision target, the decision target is analyzed, and a data acquisition scheme is formulated for the required decision support data, where the data acquisition scheme is used to divide the data range required to be acquired, i.e. the data acquisition range of the decision support data.

Step S1022, collecting data according to the data collection scheme, where the data at least includes: student examination performance data, student status information data, learning behavior data, talent demand data related to online education, national education resource allocation situation data, biogenic structure data, education teaching element data, and teaching process data.

In a possible implementation manner, the step S1022, according to the data acquisition scheme, acquires data, and further includes:

scheduling offline process flow data in the data acquisition scheme by using data lake factory services (Data Lake Factory, DLF), wherein the offline process flow data represents data which does not need real-time synchronization and real-time query;

For historical stock data in the data acquisition scheme, importing the historical stock data of a educational administration system, a learning space and a learning platform into an object storage service (Object Storage Service, OBS) through cloud data migration (Cloud Data Migration, CDM), processing the historical stock data through data lake exploration (Data Lake Insight, DLI) to generate student archive data, and loading the student archive data from the object storage service through a data warehouse service (Data Warehouse Service, DWS);

for real-time incremental data in the data acquisition scheme, extracting and importing the real-time incremental data of a educational administration system and a learning network based on a timestamp or a structured query language (Structured Query Language, SQL) into the object storage service OBS through cloud data migration CDM, processing the real-time incremental data through a data lake exploration DLI to generate new quality monitoring archive data, and loading the new quality monitoring archive data from the object storage service OBS by using the data warehouse service DWS;

the real-time delta data is combined with the historical stock data by the data warehouse service DWS.

In this embodiment, the collection data needs to collect various data from related systems such as a national opening university online learning platform, a educational administration platform, a recruitment management platform, and the like, and structured and unstructured data submitted by various branches in real time. Currently, various systems adopt different database architectures, an exemplary learning network uses Mysql, a recruitment system and a student space use SQL Server, a educational administration system uses Mysql and Oracle, and the historical data volume is huge. Therefore, the various processing strategies are executed on the Hua-Chen cloud according to different types and different structures of data when the cloud is concretely realized.

Step S1023, the collected data is probed and preprocessed to obtain the decision support data.

In an alternative embodiment, the step S1023 of probing and preprocessing the collected data to obtain the decision support data includes:

probing the acquired data to determine whether the acquired data is sufficient to support resolution of the decision target;

and under the condition that the acquired data is not enough to support solving the decision target, re-formulating the data acquisition scheme to re-acquire the data according to a new data acquisition scheme. In this embodiment, it is proposed that after data acquisition is completed, data needs to be probed and preprocessed, if the probed and preprocessed data is found to be insufficient to support the decision-making goal, the data acquisition scheme needs to be revised again, and the data acquisition and preprocessing are performed again until it is determined that the acquired data is sufficient to support the decision-making goal.

Preprocessing the acquired data to obtain the decision support data under the condition that the acquired data is determined to be enough to support solving the decision target; the pretreatment comprises the following steps: data cleaning, formatting, data conversion, missing value processing, data unbalance problem processing, data deviation problem processing, data distribution processing and abnormal value processing. The multi-dimensional data are acquired from a plurality of platforms and systems, so that the data can be effectively cleaned and integrated through various preprocessing in the face of the acquired complex and diverse data, and finally decision support data required by analysis are formed, so that the data quality is ensured.

Step S103, determining a proper data analysis method from a preset relation mapping table according to the relation between the decision target and the decision support data.

Specifically, in this embodiment, an appropriate analysis method is selected according to the relationship between data and decision targets, so as to make an intelligent judgment on the business decision problem. In the context of digital transformation, cloud computing provides computing and storage capabilities; big data technology provides the ability to acquire, store, process, and analyze data of any capacity, speed, and type; the artificial intelligence utilizes data learning experience to solve the problem, and by way of example, a machine learning algorithm is used for carrying out mathematical analysis on collected decision support data (such as images, texts, audios, time sequences and the like) in various forms, finding out the interrelationship among the data and the relation with a decision target, and further deducing an algorithm corresponding to the data analysis so as to analyze the decision support data by utilizing the corresponding algorithm to generate a specific decision. In this embodiment, the mapping relationship between the decision target and the decision support data and the data analysis method may be represented by a preset relationship mapping table. Thus, after determining the relationship between the decision target and the decision support data, the data analysis method to be used, including the one or more algorithms or models to be used, can be determined directly from the relationship mapping table. By way of example, the relationship mapping table may be expressed in the form of table 1 below:

TABLE 1

And step S104, analyzing the decision support data according to the data analysis method to generate a target decision.

In a possible implementation manner, the step S104 analyzes the decision support data according to the data analysis method to generate a target decision, including:

and under the condition that the data analysis method belongs to a simple data analysis method, generating the target decision by carrying out statistical analysis on the decision support data.

Under the condition that the data analysis method belongs to a complex data analysis method, taking historical data in the decision support data as training data, performing machine learning modeling to obtain a target learning model, inputting real-time data in the decision support data into the target learning model, and generating the target decision, wherein the target learning model is one or more of a regression model, a classification clustering model, a correlation rule analysis model, a time sequence analysis model, an emotion analysis model, an opinion mining model, a log analysis model, an anomaly detection model, a factor analysis model and a deep learning model.

In the present embodiment, the data analysis method can be roughly classified into a simple data analysis method and a complex data analysis method. Part of the decision targets can be concluded by a simple data analysis method (through basic description of data or simple statistical analysis), so as to generate required target decisions, and the other part of the decision targets can be concluded by a complex data analysis method (with more deep understanding of data and advanced data analysis). Because of the decision support data of multiple data sources acquired in this embodiment, there are multiple different types of data with different structures, and different manners are needed to analyze, process and integrate the data. Specifically, data such as images, voice, video and the like are recognized and analyzed through pattern recognition; exploring complex relationships between data through machine learning modeling techniques; learning process and environment are understood and optimized through learning analysis, so that student's academic achievement is estimated, future performance is predicted, and potential problems are found. For a decision goal, one or more data analysis methods may be required to address. The embodiment provides that through digital transformation, various intelligent algorithms and model combinations, such as machine learning, deep learning and the like, can be applied to realize prediction and optimization of student behaviors and learning data, so that more scientific and intelligent support is provided for decision making.

Illustratively, when the user-entered online education decision questions belong to questions that only need a simple data analysis method to be processed, such as "whether the professional catalog of the current online education meets the social requirement? For the problem, only the demands of enterprises in the socioeconomic industry for various post talents are collected to be used as decision support data, and answers, namely target decisions, can be obtained through statistical analysis and comparison with the current professional catalogue. For another example, when the question is "whether the teacher ratio of the organization meets the requirements? By the method, the conclusion can be drawn only by collecting the number of students and teachers and then calculating.

Illustratively, when the user-entered online educational decision questions pertain to questions that require complex data analysis methods to handle, such as "educational resource demand prediction in a place? The problem belongs to prediction problem, and the historical data in the decision support data including the regional economic development level, population flow characteristics, school age population scale and the like are required to be used for machine learning modeling (such as NSGA-II algorithm modeling, multi-level linear model or DEA-Tobit model is used), then the data of other indexes such as the current regional economic development level and the like are used as real-time data in the decision support data, and the real-time data are substituted into the model to calculate, so that the demand prediction of educational resources is finally obtained as a target decision.

The same is true of the problem between the learner and the environment where the learning occurs, which requires a complex data analysis method to handle, such as "whether the learning behavior of the online learner can predict the learning effect? "this question, confirm the question through the relation mapping table set up in advance needs to get the answer through data acquisition and analysis of student's learning process. Various data of online learning of online learners, such as test, interaction, homework and the like, are collected as decision support data, the relation between learning behaviors and learning effects is calculated through various data mining algorithms (such as logistic regression, hidden Markov model, regression analysis, factor analysis and the like), current learning behavior data of the students are used as real-time data in the decision support data, and the learning effects of the students are predicted to be used as output target decisions.

In a possible implementation manner, in a case that the decision target is student learning effect evaluation, the data analysis method belongs to the complex data analysis method, and the machine learning modeling is performed by using historical data in the decision support data as training data, so as to obtain a target learning model, which includes:

Performing data standardization on historical data in the decision support data to obtain a plurality of groups of index values, wherein each group of index values comprises a plurality of historical learning behavior index values and corresponding historical learning effect index values; the history data includes: historical learning behavior data and historical learning effect data of the object to be decided;

using factor analysis to reduce the dimension, analyzing each group of index values to obtain k principal components, wherein each principal component carries out feature description on the object to be decided from different aspects, and represents the historical learning behavior index value;

the factor number is fixed as k, a regression method is used for calculating factor scores, a factor score coefficient matrix is obtained, coefficients in the factor score coefficient matrix represent the contribution degree of each historical learning behavior index value to the main component, and the larger the coefficient is, the larger the influence of the historical learning behavior index value to the main component is;

and obtaining the target learning model by using principal component regression.

In this embodiment, under the condition that the decision target is student learning effect evaluation (i.e., predicting the learning effect of the student according to the online learning behavior of the student), basic information, learning behavior data and stage results in the learning process of the student are collected as decision support data, and the determined data analysis method is an analysis method combining principal component analysis and principal component regression, which belongs to a complex data analysis method and is used for constructing an online learner learning effect evaluation model (i.e., a target learning model), and the model is used for obtaining an overall judgment of the learning effect of the student by analyzing multiple aspects of the learning process of the student. The input is the respective feature data (evaluation index) generated in the learning process, and the output is the quantized final evaluation result.

And carrying out data standardization on the historical data in the decision support data to obtain a plurality of groups of index values. The historical data in the decision support data comprises the basic information of students, learning behavior data, stage results in the learning process and other historical information, and the characteristics of the learning process of the students are as follows: firstly, indexes describing various aspects of online learning are included; the index dimensions are different, and the index dimensions comprise the number of browsing, posting, examination and other actions, the duration data of physical examination, video watching and the like, the proportion data of various participation rates, completion rates and the like and the data of examination results; it is necessary to obtain a final comprehensive evaluation (i.e., prediction of the learning effect of the student) through various feature analyses of the student. This process is a multi-index evaluation process, and analysis is performed by using a multi-index evaluation method.

And (3) using factor analysis to reduce the dimension, and analyzing each group of index values to obtain k main components, wherein k is a constant larger than 1. Because the sample size is larger, various indexes are related to each other to reduce the influence of human factors, the embodiment provides that the index weight is not defined in advance, and a principal component analysis method is selected to reduce the dimension. The SPSS25 is used for processing, the first 30 indexes are used for constructing an online learner learning effect evaluation model, and the 31 st index is used for verifying the accuracy of the evaluation model. For example, 2 ten thousand pieces of data acquired are analyzed, and for each piece of data, 6 principal components are obtained to characterize the original 30 index values.

The factor number is fixed as k (for example, 6), a regression method is used for calculating factor scores to obtain factor score coefficient matrixes, coefficients in each principal component coefficient matrix represent the contribution degree of original features to principal components, the larger the coefficients are, the larger the influence of the original features on the principal components is, the obtained k principal components respectively carry out feature description on students from multiple sides, and the whole learning process of the students can be basically summarized; finally, the principal component regression is used to obtain an online learner learning effect evaluation model (namely a target learning model).

In a possible implementation manner, in a case that the decision target is student learning effect evaluation, the inputting real-time data in the decision support data into the target learning model generates the target decision, including:

performing data standardization on real-time data in the decision support data to obtain a plurality of current learning behavior index values; the real-time data represents current learning behavior data of the object to be decided;

inputting the plurality of current learning behavior index values into the target learning model to generate the target decision, wherein the target decision is the current learning effect index value of the object to be decided;

The method further comprises the steps of: and sending a learning early warning message to a user under the condition that the current learning effect index value is lower than a preset threshold value.

In this embodiment, after a target learning model (online learner learning effect evaluation model) is constructed, real-time data (real-time learning behavior data of the object to be decided) in decision support data is input into the model, so that data analysis is performed by using the model, a learning effect evaluation result (i.e., a target decision) is output, and evaluation of a learning effect of a specific object to be decided is completed.

For the learner, when the online learner logs in the platform to learn, the system automatically collects related information, so that the learning state of the learner is automatically evaluated. When the generated target decision indicates that the learning effect index value of the learner is lower than a preset threshold value, the learning of the learner is problematic, and a system automatically sends a learning early warning message. The learning effect evaluation of the students can enable the students to better know own learning results and know own advantages and disadvantages, so that the learning method and the learning strategy are better adjusted, and the learning effect is improved.

In addition, in the actual platform construction, the learning effect evaluation model can be used in the design development of some core modules including intelligent learning platforms, quality monitoring platforms and the like, and teachers, students and managers adjust decisions according to real-time data. For teachers, the learning process and effect evaluation of students are calculated in real time, a channel for knowing the states of the students in real time is provided for the teachers, the teachers can accurately position the states of online learners, conduct guidance and correction on behaviors of the online learners in time, the teaching content and rhythm are dynamically adjusted, and therefore more personalized services are provided for the online learners. For a manager, the learning effect evaluation of the students can help a teaching management department to better know the learning condition of the students and know the teaching effect of teachers, so that the teaching plan and the resource allocation are better adjusted, and the teaching quality is improved. The embodiment provides that the learning effect evaluation model is applied to teaching, so that teachers and students can be helped to better know the learning condition, teaching and learning strategies can be better adjusted, and teaching effects and learning effects can be improved.

Step S105, the target decision is visually displayed.

In this embodiment, after the conclusion, i.e. the target decision, is reached according to the data analysis method, it is presented to the decision maker by means of a suitable visualization scheme. Illustratively, in the case that the decision target is a student learning effect evaluation (i.e., the learning effect of the student is predicted according to the online learning behavior of the student), after the target decision (the prediction result of the learning effect of the student) is obtained, the learning effect of the student is displayed in various visual charts from various dimensions (such as the student, class, course, specialty, etc.).

In a possible implementation manner, the step S105 is to visually display the target decision, including:

performing result verification on the target decision;

and determining a proper visualization scheme according to the data relationship and the application scene of the target decision, wherein the visualization scheme is one or more of a histogram, a line graph, a pie chart, a bar graph, an area graph, a scatter graph, a stock price graph, a curved surface graph, a radar graph, a tree graph, a sunglass graph, a histogram, a box graph, a waterfall graph, a text cloud, a geographic information graph and a social network graph.

And displaying the target decision on a visual interface according to the visual scheme.

The graphical chart can more easily convey abstract information and promote the decision maker to understand the data, so that the decision result is displayed in a proper visual chart, so that a user can observe the data from different dimensions, and further knowledge and understanding are provided for the target decision. Because the visualization schemes are of various types, in the practical application process, a proper visualization scheme needs to be selected according to the data relationship and the practical application scene of the target decision, and by way of example, a visualization scheme selection table can be preset, and the visualization scheme selection table is used for storing the data relationship, the application scene and the mapping relationship between the visualization schemes, so that after the target decision is obtained, the proper visualization scheme can be determined from the visualization scheme selection table directly according to the data relationship and the application scene of the target decision. Illustratively, the visualization scheme selection table may be expressed in the form of table 2 below:

TABLE 2

In one possible implementation, after visually presenting the target decision, the method further comprises:

Receiving a decision effect evaluation result fed back by a user, wherein the decision effect evaluation result is used for indicating whether the target decision can solve the online education service problem;

ending the decision process under the condition that the decision effect evaluation result indicates that the target decision can solve the online education service problem;

and under the condition that the decision effect evaluation result indicates that the target decision cannot solve the online education service problem, adjusting a data collection range, and regenerating a new target decision to perform visual display until the decision is completed.

In this embodiment, after generating a target decision according to an online education service problem input by a user and visually displaying the target decision to a decision maker, the user may feed back a decision effect evaluation result, and illustratively, show "whether the target decision can solve the online education service problem" to the user, where "yes" or "no" is selected by the user. If the decision effect evaluation result indicates that the target decision can assist in solving the target problem (i.e. the online education service problem input by the user), the decision process is ended, if the decision effect evaluation result indicates that the target decision cannot help the user to make a decision, or the support strength of the target decision on the decision is insufficient, the problem needs to be re-analyzed, the data collection range is adjusted, the data analysis is performed again, and the visualization is performed (i.e. steps S101-S105 are re-executed) until the decision is completed, and the decision effect evaluation result fed back by the user indicates that the target decision can solve the online education service problem.

Referring to fig. 2, fig. 2 shows a schematic diagram of a decision generation flow of online education, and as shown in fig. 2, in this embodiment, decision support data is collected according to a decision target, data exploration and preprocessing are performed on the collected data, intelligent processing (including learning analysis, pattern recognition and machine learning modeling) is performed on the data, a target decision is output, result verification and communication are performed on the target decision, and visual display is performed. The decision mechanism provided by the embodiment of the application is oriented to typical problems in the online education process, the multi-objective system is comprehensively analyzed by using new technologies such as big data, machine learning and the like, finally a visual report is formed and provided for a decision maker to refer, and the decision level is improved while the complexity of decision is reduced. According to the method and the device for solving the problems, the problems are put forward, analyzed and solved, firstly, the problems needing to be decided are combed, then a decision mechanism is established, and finally, the example verification is passed. Specifically, the embodiment of the application carries out intelligent functional remodeling and process reconstruction on the process of educational decision by establishing a decision target and driving the thought of the decision by data. In the longitudinal view from data to conclusion, including three stages of data acquisition and integration, data mining and analysis and data visualization, which is a data driven decision making process, each stage of decision making needs to consider automation of the flow, diversification of dimensions, individualization of decision making and complexity of the model in the context of digital transformation, as shown in fig. 2. In particular, automation of the process, i.e., automation of data collection and processing, is a huge amount of data collected by the online educational platform, which needs to be cleaned, integrated, and processed for subsequent decision making. The embodiment provides digital transformation to require a data processing tool to be more intelligent, so that automatic data acquisition and processing can be realized, and the manual intervention and time cost are reduced. The embodiment provides that the online education platform can collect various data such as behavior data, learning data, social data and the like of students, and the digitalized transformation requires that the data can be subjected to multidimensional analysis and mining, so that the accuracy and pertinence of decision making are improved. The implementation of personalized decision means that in the digital transformation background of the embodiment, personalized decision for each student (target to be decided) such as personalized course recommendation, personalized learning plan making and the like can be realized through analysis of student data, so that the learning effect and satisfaction degree of the students are improved. The complexity of the model means that the embodiment can apply various intelligent algorithms and combinations of models, such as machine learning, deep learning and the like, through digital transformation, so as to realize prediction and optimization of student behaviors and learning data, thereby providing more scientific and intelligent support for decision making.

The technical solution of the embodiment of the present application is described below by taking the problem of "predicting student learning effect according to student online learning behavior" as an example.

Step 1, determining a decision target according to an online education service problem input by a user.

According to the online education business problem of 'predicting the learning effect of the student according to the online learning behavior of the student' input by a user, determining a decision target as 'predicting the learning effect of the student', wherein the target to be decided is the student, and the decision target belongs to the predictive problem. Referring to fig. 3, fig. 3 shows a decision generation flow chart of learning effect evaluation, and as shown in fig. 3, for students, determining evaluation content by decision targets and evaluating the problem by analyzing learning effects of the students, setting the learning effects of the students to be reflected by daily learning behaviors and short-term learning results of the students, and considering basic information, learning behavior data and stage results in the learning process of the students to be capable of basically reflecting the learning effects of the students by analyzing the learning process of the students. So that it is used as evaluation content for subsequent data acquisition.

Step 2, acquiring decision support data according to the decision target; the decision support data represents personalized information of the student A, and at least comprises: behavior data, learning data and social data of students in the online education platform, the educational administration platform and the recruitment management platform.

By way of example, the data may come from a national division 2020 spring school, containing 12 ten thousand pieces of student base information and learning outcome data, and 1.8 hundred million pieces of learning log data. As shown in fig. 3, student basic information is collected from the recruitment system, learning result information, which is the performance of each stage, is collected from the educational administration system, and learning behavior data is collected from the learning network. These data contain structured data as well as textual information of the forum. And then cleaning and integrating the data to finally form a subject data warehouse to obtain decision support data.

The step 2 comprises the following steps:

step 2-1, collecting original data, wherein the original data mainly comprises three parts: basic information of students, stage achievements of students, and log data. The basic information of the students comprises student ID, course ID, gender, age, school level and final examination result; the stage achievements of the students comprise student IDs, course IDs, along with hall test achievements and form examination achievements; the student behavior log contains student ID, course ID, behavior category, behavior time, behavior content, and the like.

And 2-2, preprocessing the collected data, and cleaning the data from the aspects of integrity and accuracy. And (3) setting strict filtering conditions, deleting missing, noisy and conflicting data, filtering out users with inactive periods, deleting records lacking important information in logs, judging a limited data range according to business rules and experience, and judging abnormal outlier data to delete.

And 2-3, converting the data. Calculating repetition rate of the posting text by a related technology of natural language processing, and averaging a plurality of along-with-hall test results of the learner according to courses; counting the number of various behaviors of each user in the log, and then performing row-column conversion to form a statistical sum of the data of each module stored according to the user; and calculating the total number of behaviors of the students, the online time, the online days, the active days, various participation rates and completion rates by using a statistical analysis method. Finally, table 3 was formed as follows.

Table 3 evaluation index of learning effect of student

/>

And 2-4, combining three results, namely basic information, learning behavior data and learning results of students to form a record form of a learner, wherein each learner has 30 indexes for representing learning behaviors and 1 index for representing learning effects, and thus data acquisition and preprocessing are completed. The experimental data is processed to finally determine 2 ten thousand pieces of data as historical data in the decision support data for modeling and testing.

And step 3, determining a proper data analysis method from a preset relation mapping table according to the relation between the decision target and the decision support data. According to the relation mapping table, an analysis method capable of selecting a combination of principal component analysis and principal component regression is determined for data analysis.

And step 4, analyzing the decision support data according to the data analysis method to generate a target decision. Step 4 comprises:

and step 4-1, the collected decision support data can be intuitively obtained by a simple data analysis method, namely basic description and simple statistical analysis, such as the distribution of the gender and age of students, the distribution of the online time length of the students, the number of times of submitting the students and the like.

And 4-2, constructing a model. For learning effect evaluation of students, historical data in decision support data can be used as training data, machine learning modeling is carried out through an analysis method combining principal component analysis and principal component regression, and a target learning model (namely an online learner learning effect evaluation model) is obtained. The inputs are the individual feature data (evaluation index) generated during learning, and the outputs are the quantized final evaluation results (target decisions).

And 4-3, verifying the model by using a test data set, comparing the evaluation value calculated by the model with the final examination result in the data, calculating the correlation by using the SPSS25, and if the two data are closely related, accurately reflecting the relation between the learning behavior and the learning effect of the student by using the model. If the accuracy of the model does not meet the requirements of a decision maker, the model is verified after the machine learning modeling algorithm is adjusted, if the model has problems, the data collection stage is returned again, the data index is adjusted, and the preprocessing process is checked until the calculation result of the model reaches the standard.

And 4-4, analyzing data. And inputting real-time data (real-time learning behavior data of the object to be decided) in the decision support data into the model, so that the model is utilized to perform data analysis, a learning effect evaluation result (namely a target decision) is output, and the evaluation of the learning effect of the specific object to be decided is completed. As shown in fig. 3, the evaluation result is obtained by evaluation analysis, and decision advice is generated.

And 5, visually displaying the target decision. After the target decision (prediction result of learning effect of the student) is obtained, the learning effect of the student is displayed in various visual charts, from various dimensions such as student, class, course, specialty, and the like.

The decision maker knows the whole class learning condition through data statistics analysis, and also can calculate the individual learning condition of students through an online learner learning effect evaluation model, and for students who cannot pass the evaluation, early warning and prompting are carried out on the students, and personalized teaching is designed according to the result. As shown in fig. 3, in the platform application process, a teacher, a manager or a parent can execute corresponding decisions according to decision suggestions, which is helpful for accurate teaching, improves learning quality and carries out comprehensive evaluation on students. If the teacher considers that the calculated data can not accurately reflect the learning effect of the students, the teacher needs to analyze the student behavior indexes concerned by the teacher again, adjust the collected data range, analyze and mine the data again, and verify the effect of the model again. This is a cyclic reciprocation, constantly tuning process, with the model being able to assist in decision making as the final goal.

The second aspect of the embodiment of the present application further provides an automatic decision generating device based on online education, referring to fig. 4, fig. 4 shows a schematic structural diagram of an automatic decision generating device based on online education, as shown in fig. 4, where the device includes:

In one possible embodiment, the data acquisition module includes:

The scheme making sub-module is used for making a data acquisition scheme according to the decision target;

and the acquisition sub-module is used for acquiring data according to the data acquisition scheme, and the data at least comprises: one or more of student examination performance data, student status information data, learning behavior data, talent demand data related to online education, national education resource allocation situation data, biogenic structure data, education and teaching element data, and teaching process data;

and the preprocessing sub-module is used for exploring and preprocessing the acquired data to obtain the decision support data.

In one possible embodiment, the preprocessing sub-module includes:

a probing unit configured to probe the collected data, and determine whether the collected data is sufficient to support resolution of the decision target;

a re-acquisition unit for re-formulating the data acquisition scheme to re-acquire data according to a new data acquisition scheme in case it is determined that the acquired data is insufficient to support resolution of the decision target;

the preprocessing unit is used for preprocessing the acquired data to obtain the decision support data under the condition that the acquired data is determined to be enough to support solving the decision target; the pretreatment comprises the following steps: data cleaning, formatting, data conversion, missing value processing, data unbalance problem processing, data deviation problem processing, data distribution processing and abnormal value processing.

In one possible embodiment, the data analysis module includes:

the simple analysis sub-module is used for generating the target decision by carrying out statistical analysis on the decision support data under the condition that the data analysis method belongs to a simple data analysis method;

the complex analysis sub-module is used for taking historical data in the decision support data as training data to carry out machine learning modeling to obtain a target learning model under the condition that the data analysis method belongs to the complex data analysis method, inputting real-time data in the decision support data into the target learning model to generate the target decision, wherein the target learning model is one or more of a regression model, a classification clustering model, a correlation rule analysis model, a time sequence analysis model, an emotion analysis model, an opinion mining model, a log analysis model, an anomaly detection model, a factor analysis model and a deep learning model.

In one possible implementation, the complex analysis sub-module includes:

the index value generation unit is used for carrying out data standardization on the historical data in the decision support data to obtain a plurality of groups of index values, wherein each group of index values comprises a plurality of historical learning behavior index values and corresponding historical learning effect index values; the history data includes: historical learning behavior data and historical learning effect data of the object to be decided;

The index value analysis unit is used for analyzing each group of index values by using factor analysis dimension reduction to obtain k principal components, each principal component carries out feature description on the object to be decided from different aspects, and the historical learning behavior index value is represented;

a factor score calculating unit, configured to fix the number of factors to k, calculate factor scores by using a regression method, and obtain a factor score coefficient matrix, where coefficients in the factor score coefficient matrix represent the contribution degree of each historical learning behavior index value to the principal component, and the greater the coefficient, the greater the influence of the historical learning behavior index value to the principal component;

and the regression unit is used for obtaining the target learning model by using principal component regression.

In one possible implementation, the complex analysis sub-module includes:

the normalization unit is used for carrying out data normalization on the real-time data in the decision support data to obtain a plurality of current learning behavior index values; the real-time data represents current learning behavior data of the object to be decided;

the target decision generation unit is used for inputting the plurality of current learning behavior index values into the target learning model to generate the target decision, wherein the target decision is the current learning effect index value of the object to be decided;

The apparatus further comprises: and the early warning module is used for sending a learning early warning message to a user under the condition that the current learning effect index value is lower than a preset threshold value.

In one possible embodiment, the visualization module includes:

the verification sub-module is used for verifying the result of the target decision;

the scheme determining submodule is used for determining a proper visual scheme according to the data relation and the application scene of the target decision, wherein the visual scheme is one or more of a histogram, a line graph, a pie chart, a bar graph, an area graph, a scatter graph, a stock price graph, a curved surface graph, a radar graph, a tree graph, a sunburst graph, a histogram, a box graph, a waterfall graph, a text cloud, a geographic information graph and a social network graph;

and the display sub-module is used for displaying the target decision on a visual interface according to the visual scheme.

In one possible embodiment, the apparatus further comprises:

the feedback receiving module is used for receiving a decision effect evaluation result fed back by a user, and the decision effect evaluation result is used for indicating whether the target decision can solve the online education service problem;

the ending module is used for ending the decision process under the condition that the decision effect evaluation result indicates that the target decision can solve the online education service problem;

And the re-execution module is used for adjusting the data collection range and re-generating a new target decision to carry out visual display until the decision is completed under the condition that the decision effect evaluation result indicates that the target decision cannot solve the online education service problem.

In a possible implementation manner, the collecting submodule further includes:

the first acquisition unit is used for scheduling offline processing flow data in the data acquisition scheme by using DLF, wherein the offline processing flow data represents data which does not need real-time synchronization and real-time inquiry;

the second acquisition unit is used for guiding the historical stock data of the educational administration system, the learning space and the learning platform into the OBS at one time by using CDM (code division multiple access) for the historical stock data in the data acquisition scheme, processing the historical stock data by using DLI, generating student archive data, and loading the student archive data from the OBS by using DWS;

the third acquisition unit is used for extracting and importing the real-time incremental data of the educational administration system and the learning network into the OBS based on a time stamp or a structured query language by using CDM (code division multiple access) for the real-time incremental data in the data acquisition scheme, processing the real-time incremental data by using DLI (digital versatile interface) to generate new quality monitoring archive data, and loading the new quality monitoring archive data from the OBS by using DWS;

And the merging unit is used for merging the real-time increment data with the historical stock data through DWS.

The embodiment of the application also provides an electronic device, and referring to fig. 5, fig. 5 is a schematic diagram of the electronic device according to the embodiment of the application. As shown in fig. 5, the electronic device 100 includes: the system comprises a memory 110 and a processor 120, wherein the memory 110 is in communication connection with the processor 120 through a bus, and a computer program is stored in the memory 110 and can run on the processor 120, so that the steps in the online education-based decision automatic generation method disclosed by the embodiment of the application are realized.

Embodiments of the present application also provide a computer readable storage medium having stored thereon a computer program/instruction which, when executed by a processor, implements steps in an online education-based decision automatic generation method as disclosed in embodiments of the present application.

Embodiments of the present application also provide a computer program product which, when run on an electronic device, causes a processor to perform the steps of an online education-based decision auto-generation method as disclosed in embodiments of the present application.

In this specification, each embodiment is described in a progressive manner, and each embodiment is mainly described by differences from other embodiments, and identical and similar parts between the embodiments are all enough to be referred to each other.

Embodiments of the present application are described with reference to flowchart illustrations and/or block diagrams of methods, apparatus, electronic devices, and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing terminal device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal device, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

While preferred embodiments of the present embodiments have been described, additional variations and modifications in those embodiments may occur to those skilled in the art once they learn of the basic inventive concepts. It is therefore intended that the following claims be interpreted as including the preferred embodiments and all such alterations and modifications as fall within the scope of the embodiments of the present application.

Finally, it is further noted that relational terms such as first and second, and the like are 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. Moreover, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or terminal that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or terminal. Without further limitation, an element defined by the phrase "comprising one … …" does not exclude the presence of other like elements in a process, method, article or terminal device comprising the element.

The above detailed description of the method and apparatus for automatically generating decisions based on online education provided in the present application applies specific examples to illustrate the principles and embodiments of the present application, where the above examples are only used to help understand the method and core ideas of the present application; meanwhile, as those skilled in the art will have modifications in the specific embodiments and application scope in accordance with the ideas of the present application, the present description should not be construed as limiting the present application in view of the above.

Claims

1. An automatic decision generation method based on online education, which is characterized by comprising the following steps:

and visually displaying the target decision.

2. The automatic decision generation method based on online education according to claim 1, wherein the collecting decision support data according to the decision target comprises:

according to the decision target, a data acquisition scheme is formulated;

according to the data acquisition scheme, acquiring data, wherein the data at least comprises: one or more of student examination performance data, student status information data, learning behavior data, talent demand data related to online education, national education resource allocation situation data, biogenic structure data, education and teaching element data, and teaching process data;

and probing and preprocessing the acquired data to obtain the decision support data.

3. The automatic decision generation method based on online education according to claim 2, wherein the probing and preprocessing the collected data to obtain the decision support data comprises:

Under the condition that the collected data is not enough to support solving the decision target, re-formulating the data collection scheme to re-collect the data according to a new data collection scheme;

preprocessing the acquired data to obtain the decision support data under the condition that the acquired data is determined to be enough to support solving the decision target; the pretreatment comprises the following steps: data cleaning, formatting, data conversion, missing value processing, data unbalance problem processing, data deviation problem processing, data distribution processing and abnormal value processing.

4. The automatic generation method of online education-based decision according to claim 1, wherein the analyzing the decision support data according to the data analysis method to generate a target decision comprises:

under the condition that the data analysis method belongs to a simple data analysis method, generating the target decision by carrying out statistical analysis on the decision support data;

5. The automatic decision generation method based on online education according to claim 4, wherein the data analysis method belongs to the complex data analysis method when the decision target is student learning effect evaluation, and the machine learning modeling is performed by using historical data in the decision support data as training data to obtain a target learning model, comprising:

6. The automatic generation method of online education-based decision according to claim 5, wherein in case the decision target is a student learning effect evaluation, the inputting real-time data of the decision support data into the target learning model generates the target decision, comprising:

7. The automatic online education-based decision generation method according to claim 1, wherein the visually displaying the target decision comprises:

performing result verification on the target decision;

According to the data relation and the application scene of the target decision, determining a proper visualization scheme, wherein the visualization scheme is one or more of a histogram, a line graph, a pie chart, a bar graph, an area graph, a scatter graph, a stock price graph, a curved surface graph, a radar graph, a tree graph, a sunglass graph, a histogram, a box graph, a waterfall graph, a text cloud, a geographic information graph and a social network graph;

8. The automatic online education-based decision generation method according to claim 1, wherein after visually displaying the target decision, the method further comprises:

9. The automatic online education-based decision generation method according to claim 2, wherein the collecting data according to the data collection scheme further comprises:

the offline processing flow data in the data acquisition scheme is arranged and scheduled by using data lake factory service, and the offline processing flow data represents data which does not need real-time synchronization and real-time query;

for the historical stock data in the data acquisition scheme, importing the historical stock data of a educational administration system, a learning space and a learning platform into an object storage service through cloud data migration at one time, processing the historical stock data through data lake exploration to generate student archive data, and loading the student archive data by using a data warehouse according to the object storage service;

for the real-time incremental data in the data acquisition scheme, extracting and importing the real-time incremental data of a educational administration system and a learning network based on a time stamp or a structured query language into the object storage service through cloud data migration, processing the real-time incremental data through data lake exploration to generate new quality monitoring archive data, and loading the new quality monitoring archive data by using the data warehouse according to the object storage service;

And merging the real-time incremental data with the historical stock data through the data warehouse service.

10. An automatic decision-making device based on online education, the device comprising: