Design and Empirical Analysis of Artificial Intelligence-Based Decision Aid Models for College Management

A data warehouse is a subject-oriented, integrated, and relatively stable collection of data reflecting historical changes used to support management decisions. In data warehousing, data warehouse architectures usually use several approaches to data modelling, i.e., conceptual, logical, and physical models. Nowadays, the basic conceptual modelling approach based on information modelling, model and snowflake star modelling approach based on logical modelling approach 3NF and star model physical modelling, is the more popular physical modelling approaches.

Data warehouses integrate data, but do not directly support more complete analyses and multidimensional views, and lack flexibility in decision-making conditions. It only analyzes the basis of the decision. There must also be a powerful tool for analysis and decision support, so online analytical processing (OLAP) technology came into being.

Online analytical processing is designed for specific data storage and analysis techniques that provide decision analysis from a multidimensional viewpoint for the benefit of decision makers and senior managers. The data warehouse model design phase is equivalent to the initial phase, while the data warehouse-based online analytical processing techniques support a more comprehensive grasp of the data and understanding of the results through the aggregation of multidimensional data, pre-storage of aggregated results, support for multiple perspectives of the data, multidimensional data and observation support.OLAP technology is mainly used in data warehouses for data for the aggregation of statistical surfaces, attempting to unify application OLAP technology is mainly used in data warehouses to aggregate statistical surfaces of data in an attempt to respond to complex survey requirements of non-data processing professionals in a short period with a unified application logic and data model.

Another very powerful tool in AI is data mining techniques, which can be used to help organizations support decision-making.Data mining is a deeper search of source data. Source data includes not only the data in the data warehouse but also other categories of data. Data mining, also known as “knowledge discovery in large databases”, is a modern and powerful IT tool that can be used to extract useful information. In order to make better decisions, the study of databases and information technology provides an efficient way to store and process this valuable data. Data mining is the process of extracting useful information and patterns from vast amounts of data.It is also known as the process of knowledge discovery from data, knowledge extraction and data/pattern analysis knowledge mining.

Data mining involves many subject areas and methods, which can be categorised according to several aspects, such as database types, mining objects, mining tasks, mining methods, and techniques.

Classification is the mapping of existing data records to a given category based on predefined classification labels (classifiers) by some method. Classifiers vary according to the application context and data characteristics, and no classifier can be adapted to any application context and data characteristics, but it can be integrated using a combination of methods or selection methods.

Clustering is the process of aggregating data without categories into groups according to their similarities and describing the meaning of each group. The aim of clustering is to minimize the gap between data in the same group and maximize the difference between different groups as much as possible.

Correlation analysis refers to the search for the existence of certain specific regularities between two or more data sets, such as frequent patterns, temporal patterns, causal structures, etc. Or it is the discovery of hidden networks of correlations between large amounts of data. Correlation analysis is to find the hidden correlation network between the data. There are many rules of correlation between a large amount of data, but there are some correlations that are not certain or useless, so we need to use “credibility” or “support” to eliminate our unwanted correlations. Correlation.

Regression analysis is the process of building mathematical models by finding causal relationships between dependent and independent variables and testing the data against the models so that one can observe and predict how the dependent variable will change when the independent variable changes.

Predictive analytics is through the analysis of a large number of continuous historical data, from which the existence of this data cycle change law, that is, the predictive model, and the use of the model for the future of a certain time value of the data may appear in the state of the characteristics of the prediction.

Inevitably, there will be some abnormal data in the database. These abnormal data are not useless. On the contrary, they are also very likely to be particularly important. Deviation analysis, sometimes referred to as comparative analysis, compares the differences between observed results and the original reference values and reveals the underlying causes.

At present, the most used method in association rule research is the “support and confidence” mathematical model, which involves the following concepts.

The task of the first subproblem is to quickly and efficiently find out all the frequent itemsets in the transaction set, which is the core problem of association rule mining and the standard of association rule mining algorithms, and the solution of the second subproblem is relatively easy and direct, and most of the current association rule mining algorithms are proposed for the first problem.

When it comes to association rule algorithms, the Apriori algorithm is one of the most typical algorithms, which mines rules that are Boolean in classification and contain only one-dimensional association rules involving only a single layer of concepts. The Apriori algorithm involves two main steps: extracting all the frequent itemsets L that satisfy the Support(X)≥minsup from the available data, and mining all the strong association rules that satisfy both the Support(X⇒Y)≥minsup\[Support(X\Rightarrow Y)\ge minsup\] and Confidence(X⇒Y)≥minconf.Confidence(X⇒Y)≥minconf$Confidence(X\Rightarrow Y)\ge minconf.Confidence(X\Rightarrow Y)\ge minconf$ . The efficiency of extracting the frequent itemsets is a key indicator of the overall performance of the algorithm, and its specific steps are described as follows:

Cow into frequent 1-item sets: first all the data items in the whole transaction D to generate candidate 1-item sets recorded as C₁ (the provisions of candidate k– item sets recorded as C_k ), retrieve the transaction set D to calculate the D for each candidate 1-item set of the Support, selected C. can meet the Support(C₁)≥minsup is a frequent 1-item set L₁ (the provisions of the frequent k– item sets recorded as L_k).

Join: If two L_k s have k–1 items that are the same, join the two L_k s (L_k∞L_k) to generate the candidate item set C_k+1(k = 1,2,3,i).

Here C_k+1 is generated by the concatenation operation of L_k, instead of being obtained by the permutation Cmk+1$C_{m}^{k+1}$ of all the items in the itemset I in the transaction set D (let there be m items in I). Assuming that L_k+1 is frequent, any of its non-empty subsets L_k must be frequent, and hence the efficiency of the algorithm is greatly improved by using the concatenation operation of L_k to generate C_k+1.

Pruning: In the C_k+1 generated by the concatenation operation of L_k, both frequent and infrequent itemsets are possible. Assuming that the k– itemset is not a frequent candidate, then its superset (k+1)– itemset cannot be a frequent candidate. Cutting off the candidate (k+1)– itemsets that do not appear in the subset of L_k can greatly improve the efficiency of the next step by cutting off some of the useless candidate (k+1)– itemsets first.

Generate L_k+1: Retrieve the entire transaction database D to compute its support for C_k+1, and select the set of candidate (k+1)– items that match the Support(C_k+1)≥minsup to obtain frequent L_k+1.

Repeat the link~generation steps until L_k+1 is empty, at which point all frequent itemsets contained in the entire transaction database D will be retrieved.

The number of students enrolled in the college entrance examination each year since 2012 is shown in Table 1. Starting in 2013, China began to significantly expand the scale of higher education enrolment, and the enrolment for that year was determined to be 1.602 million, compared with 1.085 million in 2012, an increase of 517,000. And before 2012, the average annual increase in higher education development was about 9 percent.After the expansion, China’s enrollment in higher education and the number of students has continued to increase every year. According to the Ministry of Education’s education development statistics bulletin for the previous year released on 06 May 2022, the total scale of all types of higher education in the country exceeded 27 million people in 2021, and the gross enrollment rate of higher education reached 23%, which has crossed the stage of massification of higher education that has been internationally recognised. Now, the scale of China’s higher education has successively surpassed that of Russia, India and the United States, becoming the world’s first. After only a few years of hard work, and with a per capita GDP of more than US$1,000, China’s higher education development has moved from elite education to mass education, completing a path that took other countries 30 to 50 years or even longer. The dispersal and serious shortage of management resources, the difficulty of management and the increase in management costs have forced school leaders and management departments to disperse their energies, and managers have been “overworked”. There is also the problem of management friction. This problem mainly exists in multi-school mergers (especially “pull together” type mergers) for various reasons: the formation of “small groups”, low mutual trust, communication difficulties, decision-making difficult to form, resulting in poor order, and management (c) Reduced effectiveness. Therefore, the original teaching management system must be well planned and rebuilt through certain management strategies and management techniques so that it can better meet the needs of the new era of multi-campus operation and teaching management of colleges and universities so that each campus can provide the community with a consistent quality of education, talent training quality.

The grass-roots organisations of colleges and universities are the backbone of teaching management and the foundation on which schools carry out teaching management. Therefore, the role of grass-roots organisations in teaching management should be highlighted, the course of action of teachers and teaching managers should be formulated, and their leading role in teaching management evaluation should be given full play so as to enhance teaching management executive power. In carrying out the evaluation process of teaching management execution, two basic principles should be followed. The first step is to eliminate psychological barriers to evaluation, reduce antipathy towards the evaluated, and guide teachers to improve their sense of responsibility. The second is to formulate a reward and punishment system that is compatible with incentives and constraints so as to promote the continuous improvement of teaching and teaching management.

The calculation results are shown in Table 3. Among the five evaluation grades of excellent, good, medium, passing and poor, the percentage of those evaluated as excellent, good and medium are 35.8%, 28.3% and 18.9%, respectively, and the total percentage of experts evaluating the execution of the school as medium and above grades is 83%, and the experts evaluate the execution of the school’s instructional management relatively well. The percentage of those evaluated as passing and poor is 9.4 per cent and 7.5 per cent, respectively, which is not low and deserves the attention of school teaching managers.

A sound teaching management mechanism and system are the guarantee for the smooth implementation of teaching management in schools. Schools must innovate the management mechanism, improve the system to be compatible with incentives and constraints, and stimulate teachers’ enthusiasm for teaching. Firstly, establish and improve the assessment mechanism. Decompose the teaching objectives for each unit of instruction, and all teachers work together to fulfill their teaching tasks. Establish a sound assessment mechanism to evaluate different teaching tasks separately. Secondly, establish an effective incentive mechanism to mobilise the enthusiasm of teachers and teaching managers. Thirdly, create a harmonious teaching environment.