Research on Ideological Course Recommendation System based on

Support Vector Machine

He Zhang

Ideological and Political Department, Suzhou Vocational University, Suzhou, Zhejiang, China

Keywords: Support Vector Machine, Ideological Course Recommendation System, Ideological Education.

Abstract: With the deepening of students' Ideological education in Colleges and universities, as the main channel to

cultivate students' world outlook, values, outlook on life and socialist core values, ideological courses occupy

an important position in College Ideological education. In view of the problems existing in the curriculum of

Ideological education in Colleges and universities, such as single form, weak pertinence, lack of synergy and

unable to form a personalized collaborative education mechanism, a college ideological Curriculum

recommendation system based on improved collaborative filtering technology is developed, which adopts an

improved collaborative filtering algorithm based on hybrid, By introducing the gradual forgetting curve based

on the timeliness change of user interest, the disadvantages of traditional collaborative filtering algorithms

such as low efficiency, weak adaptability and novelty elimination are better solved. In order to solve the

problems of long server response time and low user preference for the system recommendation results in the

practical application of the recommendation system based on knowledge map, the research on the curriculum

ideological Teaching Resource Recommendation System Based on big data is carried out. In the hardware

part, the selection of underlying physical host and database server are designed; In the software part, this

paper designs the reading and search of curriculum ideological teaching resources based on big data.

1 INTRODUCTION

With the development of modern science and

technology, informatization and networking have

become the development trend in China .At the same

time, with the development of China's higher

education from elite education to popular education,

the expansion and expansion of colleges and

universities and the running of multi campus schools,

ideological teachers have the reality of "mobile

teaching" in teaching practice, ideological teachers

have less and less time to contact students outside

class, and students' consolidation and review of

knowledge completely rely on their own efforts

(

Chen, 2017). Such a way is more likely to result in a

lack of in-depth understanding of the ideological

course, and students' learning has a strong utilitarian

mentality, resulting in students learning for the

examination in the ideological course. Therefore,

how to make efficient use of extracurricular time,

strengthen students' guidance on Ideological courses,

deepen college students' understanding of Ideological

course knowledge, cultivate students' logical thinking

ability, and strengthen the teaching effect of

Ideological courses has become another thinking

focus for the innovation of teaching methods of

Ideological courses in Colleges and universities

(

Chen, 2016). With the help of the current popular idea

of curriculum network, this paper constructs a

network platform that can be specially used for the

learning and communication of College Ideological

courses. In the Internet era, the data of curriculum

ideological teaching resources are increasingly rich,

and their forms are gradually diversified, which puts

forward higher requirements for the storage,

processing and analysis of curriculum teaching

resources (Chang,

2017).

The traditional stand-alone architecture server has

been unable to meet the needs of unstructured data

and diversified processing of current curriculum

ideological education resources. There is an urgent

need for a virtualization technical means to build the

server cluster. Relevant researchers have conducted

in-depth research on the above problems, and (Liu,

2021) put forward some storage, processing and

analysis methods of curriculum ideological

teaching resources from many aspects. However, the

Zhang, H.

Research on Ideological Course Recommendation System based on Support Vector Machine.

DOI: 10.5220/0011289900003440

In Proceedings of the International Conference on Big Data Economy and Digital Management (BDEDM 2022), pages 757-761

ISBN: 978-989-758-593-7

757

above methods can not realize the centralized

processing of curriculum ideological teaching

resource data in the process of application. Moreover,

the application of various methods also increases the

operation pressure of traditional servers, which

seriously limits the storage capacity and computing

power of the system. Moreover, some teaching

resources are stored in the system in the form of dark

information, which can not be obtained by users, and

it is difficult to make full use of all data. In order to

improve the utilization rate and value of curriculum

ideological teaching resources, this paper studies the

curriculum ideological teaching resources

recommendation system based on big data, and

verifies the effectiveness of the design system

through experimental comparison, in order to provide

users with higher precision and faster

recommendation services (Li,

2018).

2 SUPPORT VECTOR MACHINE

2.1 The Concept of Support Vector

Machine

Support vector machine is a machine learning method

based on statistical learning theory, dimension theory

and structural risk minimization principle. It shows

many unique advantages in solving the problems of

small sample, nonlinear and high-dimensional pattern

recognition, and overcomes the problems of

"dimension disaster" and "over learning" to a great

extent. In addition, it has a solid theoretical

foundation and a simple mathematical model.

Therefore, it has made great progress in the fields of

pattern recognition, regression analysis, function

estimation and time series prediction, and is widely

used in text recognition, handwritten font

recognition, face image recognition, gene

classification and time series prediction (Li,

2017).

The standard support vector machine learning

algorithm problem can be reduced to solving a

constrained quadratic programming problem. For

small-scale quadratic optimization problems, mature

classical optimization algorithms such as Newton

method and interior point method can be well solved.

However, when the training set is large, there will be

some problems, such as slow training speed, complex

algorithm, low efficiency and so on. Data based

machine learning is an important research content and

direction in modern artificial intelligence technology.

Its main research is to find laws from observed data,

and use these laws to predict future data or

unobservable data. Data based machine learning can

be roughly divided into three implementation

methods, as shown in Figure 1:

Figure 1: Data machine learning method.

The first is the classical statistical estimation

method. The second is empirical nonlinear methods,

such as artificial neural networks. The third method

is statistical learning theory. At present, some

mainstream training algorithms decompose the

original large-scale problem into a series of small

problems, solve the small problems repeatedly

according to some iterative strategy, construct the

approximate solution of the original large-scale

problem, and make the approximate solution

gradually converge to the optimal solution. However,

how to decompose the large-scale problems and how

to select the appropriate working set are the main

problems faced by the current training algorithms,

and they are also the performance of the advantages

and disadvantages of each algorithm. In addition, the

existing large-scale problem training algorithms can

not completely solve the problems faced. Therefore,

it is imperative to make reasonable improvement on

the original algorithm or study new training

algorithms. Firstly, this paper systematically

introduces the theory of support vector machine, then

summarizes the current training algorithms, and looks

forward to the future research direction.

2.2 Support Vector Machine Algorithm

The training of support vector machine needs to solve

a problem. Traditional optimization algorithms, such

as interior point algorithm, can not be directly used to

solve the problem of SVM because of the limitation

of computer memory capacity. So far, in order to

solve the training problem of SVM, several

algorithms have been developed to overcome the

difficulties, so as to better train SVM. This chapter

introduces a new SVM training method, which uses

particle swarm optimization algorithm to optimize

BDEDM 2022 - The International Conference on Big Data Economy and Digital Management

758

the quadratic programming problem of SVM, and

tests and compares it through experiments. The idea

of support vector machine classification algorithm is

to find the hyperplane with the largest classification

interval in its high-dimensional feature space and

separate the two types of samples. SVM is developed

on a solid theoretical basis and has better

generalization performance. However, in the training

of SVM, especially for large-scale sample data sets,

the quadratic programming problem that must be

solved is the restriction of the development and

application of SVM. For the sample data set, the

quadratic programming includes optimization

variables, linear inequality constraints and equality

constraints, and involves the calculation of

dimensional kernel function matrix and the

multiplication of matrix and vector. Figure 2 shows

the learning system model:

Figure 2: Learning system model.

In optimization theory, there are many mature

algorithms, such as quasi Newton method and general

software toolkit, which can be used as the training

algorithm of SVM. It is extended to the case with soft

boundary. However, the learning speed of this

method is not as fast as that of traditional QP

software, especially in the case of small data sets.

Therefore, this kind of method has rarely been

applied to the training of SVM. The second kind of

algorithm is the most studied and has good

application effect. The basic idea of the second

method is to decompose the large-scale QP problem

into a series of small-scale QP problems. In each

iteration, the traditional optimization algorithm is

used to solve a sub QP problem and update it α A

fractional quantum set of, that is, the working set.

These methods are collectively referred to as

decomposition algorithms. The main differences are

the size of working set, the principle of generating

working set and the solution method of sub QP

problem of working set. As for the third kind of

incremental online training algorithm, its

particularity is that the number of training samples is

unknown. In the process of use, training samples will

be added continuously, which is mainly used for

online system identification.

3 IDEOLOGICAL COURSE

RECOMMENDATION SYSTEM

3.1 System Hardware Design

In order to effectively recommend curriculum

ideological teaching resources, big data technology is

introduced, HDFS distributed is taken as the basic

structure of this system, the data files of curriculum

ideological teaching resources are stored and read by

calling hdfsapi interface, and the basic information of

system users is stored by using resource sharing

platform. The hardware structure of the

recommended system is shown in Figure 1: combined

with the structural composition of the system

hardware, this paper mainly designs the underlying

physical host and database server. The underlying

physical host is the core hardware structure of the

system recommended in this paper, and Inspur server

is selected as the underlying physical host. The server

of this model has CPU and adopts Xeon Series CPU.

The main frequency of CPU is 1066MHz, the

memory is 32GB, the capacity of hard disk is 1000gb,

the speed of hard disk is 12000 rpm, and the memory

type is DDR3. At the same time, the underlying

physical host of this model has a continuous data

protection mechanism, which can reduce the

probability of system downtime to a certain extent,

and further improve the stability and data fault

tolerance of the system, so as to cope with the

increasing curriculum ideological teaching resources

in the future. And build a hardware resource pool

including various computer resources, storage

resources and curriculum ideological teaching

resources, manage the hardware resource pool of

Inspur server through the client, and then virtualize

all hardware resources in the resource pool to

generate several independent virtual machines, so as

to reduce the burden of system operation. Figure 3

shows the system hardware design structure:

Figure 3: System hardware design structure.

Research on Ideological Course Recommendation System based on Support Vector Machine

759

The database server is mainly used to store the

operation parameters generated during the operation

of the recommendation system and the ideological

teaching resources of various courses. Considering

the capacity of current curriculum ideological

teaching resources and the growth trend of future

resources. In order to realize the effective

recommendation of teaching resource data, it is

necessary to store enough teaching resource data in

the system. This paper uses big data technology to

read the ideological teaching resources of various

courses. After reading the ideological teaching

resources of all courses, the distributed search

method is used to search them. Therefore, the

distributed search engine can be deployed on the

distributed cluster, and an index database for

curriculum ideological teaching resources can be

constructed to accurately retrieve the curriculum

ideological teaching resources in the big data

platform, so as to realize the retrieval function of the

recommendation system in this paper. After receiving

the corresponding request information in the system

background, search the relevant data information in

the resource database by calling. If the relevant index

information is queried, the searched curriculum

ideological teaching resource data will be displayed

to the system user.

3.2 Improvement of Recommendation

System of Ideological Course

Through the actual investigation of front-line

teachers and students, the functional requirements

analysis of College Ideological course

recommendation system based on improved

collaborative filtering algorithm is formed. On this

basis, the functional logic of the system is designed,

and the overall functional framework of the system is

designed according to the general process of software

engineering design (Mangeli,

2019). The system uses

B / S architecture to build the implementation

framework. The functional module design of the

system follows the principles of practicality,

modularity and scalability. The core module of the

system mainly includes the course selection sub

module of College Ideological course, the student

evaluation sub module of College Ideological course,

the recommendation sub module of College

Ideological course, and the system maintenance and

update sub module. Each sub module works together

under the control of the system workflow, Build an

efficient and practical closed-loop dynamic

recommendation mechanism for ideological courses

in Colleges and universities, form a virtuous circle,

and provide a basic guarantee for the development of

Ideological education in Colleges and universities.

After determining the application student group,

carry out the system initialization operation, mainly

complete the input of the ideological course

information currently opened by the school and the

students' preference information for each course in

the past historical cycle, and write the initial value to

the system data warehouse as the initial cold start data

set of the improved collaborative filtering algorithm;

Start the personalized scoring mechanism of

Ideological education curriculum, and evaluate the

personalized ideological education plan formulated

by colleges and universities from multi-dimensional

effect evaluation; Start the personalized

recommendation sub module of Ideological

education courses, recommend personalized and

accurate ideological courses for different students,

improve students' points of interest, and ensure the

formation of a three-dimensional education situation

of "watering flowers and roots, teaching people and

teaching heart" of Ideological education in

Colleges and universities.(Mohammadmehdi,

2018).

The learning problem can be generally expressed as

an unknown dependency between the output variable

y and the input variable x, that is, it follows an

unknown probability measure. The machine learning

problem is based on an independent and identically

distributed observation sample, namely formula (1):

),(),...,,(),,

2211 tt

yxyxyx（

(1)

In a set of functions, find an optimal function and

estimate the dependency, so that the expected risk is

as shown in formula (2):



= ),()),(,()( yxdFwxfyLwR

(2)

For pattern recognition problems, the output can

be expressed as y = {0,1} or {1, − 1} respectively.

The predicted function is called the indicator

function,

and the loss function can be defined as formula (3):

),(,0)),(,( wxfifywxfyL ==

(3)

In order to improve the portability of the system,

a college ideological course recommendation system

based on improved collaborative filtering algorithm

is developed by using the idea of modular design and

calling the form of dynamic link library

files(Saranjam

2016). The system can realize the

course selection and course evaluation of College

Ideological courses, the preference data statistics of

students for specific ideological courses Personalized

and accurate ideological course recommendation for

different students. The main reason for this

phenomenon is that the design system automatically

filters the curriculum ideological teaching resources

BDEDM 2022 - The International Conference on Big Data Economy and Digital Management

760

that do not meet the requirements, and adds more

operating conditions in the process of system design,

so as to realize the good recommendation of

curriculum ideological teaching resources. The

course Ideological teaching resource

recommendation system designed by using big data

technology can shorten the response time of the

system server and provide users with more accurate

recommendation services on the basis of fully

meeting user preferences (TingLong,

2018).

4 CONCLUSIONS

Based on big data technology, a new curriculum

ideological teaching resource recommendation

system is designed, and the feasibility and advantages

of the design system are proved by comparative

experiments. (Wang,

2020). However, due to the

limited research ability, the system designed in this

paper still has some deficiencies. For example, users

can not avoid repeatedly uploading the same

curriculum ideological teaching resources, resulting

in repeated storage of resource data. Therefore, in the

next research process, we also need to carry out

research on the above deficiencies, so as to further

improve the Teaching Resource Recommendation

effect of the design system. In view of the problems

existing in the current curriculum of Ideological

education in Colleges and universities, such as

mechanical rigidity, weak pertinence, lack of synergy

and inability to form a personalized collaborative

education mechanism (Zhao,

2017), a college

ideological Curriculum recommendation system

based on improved collaborative filtering technology

is developed, which adopts an improved collaborative

filtering algorithm based on hybrid, By introducing

the gradual forgetting curve based on the timeliness

change of user interest, an optimized College

Ideological course recommendation model is

designed. The simulation shows that the model solves

the disadvantages of low efficiency, weak

adaptability and novelty of the traditional

collaborative filtering algorithm. Personalized and

accurate ideological course recommendation for

different students. The system design logic is clear,

the internal working logic meets the general

requirements of software engineering, the division

between functional modules is reasonable, the

expected design purpose is well completed, and the

conditions for popularization and use in Colleges and

universities in China are preliminarily met.

REFERENCES

Chang F. (2017). High-performance Chinese multiclass

traffic sign detection via coarse-to-fine cascade and

parallel support vector machine detectors. Journal of

Electronic Imaging, vol.26, no.5.

Chen C. (2017). Effective computational detection of

piRNAs using n-gram models and support vector

machine. BMC Bioinformatics, vol.18, no.14, pp.17.

Chen W. (2016). Landslide susceptibility mapping based

on GIS and support vector machine models for the

Qianyang County, China. Environmental Earth

Sciences, vol.75, no.6, pp.1-13.

Li J. (2018). An Analysis on the Factors Affecting the

Satisfaction Degree of the System of the Ideological

Curriculum in General Education in Chinese Junior

Colleges of Science and Technology. Matec Web of

Conferences, no.2, pp.28.

Li K. (2017). Hierarchical multi-class classification in

multimodal spacecraft data using DNN and weighted

support vector machine. Neurocomputing, vol.25,

no.11, pp.55-65.

Liu R. (2021). Design of Ideological Multimedia Network

Teaching Resources Integration System Based on

Wireless Network. Scientific Programming, vol.21,

no.3, pp.1-15.

Mangeli M. (2019). Improvement of risk assessment in the

FMEA using nonlinear model, revised fuzzy TOPSIS,

and support vector machine. International Journal of

Industrial Ergonomics, no.69, pp.20-26.

Mohammadmehdi S (2018). Comparative Performance

Analysis of Support Vector Machine, Random Forest,

Logistic Regression and k-Nearest Neighbours in

Rainbow Trout Classification Using Image-Based

Features. Sensors, vol.18, no.4, pp.27.

Saranjam K. (2016). Analysis of dengue infection based on

Raman spectroscopy and support vector machine

(SVM). Biomedical Optics Express, vol.7, no.6,

pp.249.

Ting-Long T. (2018). Support vector machine incremental

learning triggered by wrongly predicted samples.

Optoelectronics Letters, vol.14, no.3, pp.23-25.

Wang Y. (2020). Analysis on the construction of

ideological education system for college students based

on mobile artificial intelligence terminal. Soft

Computing, vol.24, no.20.

Zhao J. (2017). A design of college ideological education

management system based on data mining technology.

Revista de la Facultad de Ingenieria, vol.32, no.14,

pp.73-78.

Research on Ideological Course Recommendation System based on Support Vector Machine

761