Вы находитесь на странице: 1из 5

UIC-ATC-ScalCom-CBDCom-IoP 2015

Research on E-learning in College Education


A model on personalized resource recommendation based on learning situation

Jinghua Zhang
Department of Computer
North China Electric Power University (Baoding), NCEPU
Baoding, China
zjh_bd@tom.com

Abstract—This paper pays great attention to the specific e- and evaluation of learning achievement, the learning path
learning for college education and tries to get better and process are controlled and different learning materials
combination with classroom teaching and provides better are provided. The correlational researches involved the
service for college course learning. The research on e-learning whole application model, the student model and evaluation
is Based on the thought of learning resource recommendation, strategy of learning achievement[6], the adaptive control
and does more work on linking up with college course and strategy[3,4,5] and so on. Another category is personalized
instructing learning by more suitable recommendation under resource recommendation. This category researches
college course learning context. The paper presents a solution recommendation. It gives more freedom to the learner
of generating a virtual self-learning space named learning
without the control of learning process. The correlational
situation which represents the learning demands linking up
researches were mainly for recommendation strategies. Some
with college course. At last, a generic model for the solution is
given to support the personalized resource recommendation
researches do recommendation through learner
based on learning situation. characteristics[7,8] such as learning interests, learning style,
learning preferences, learner’s knowledge level. Some papers
Keywords-e-learning; college education; learning situation; used collaborative filtering techniques[9]. Some Papers
personalized recommendation; learning instruction; researched the context model to guide the
recommendation[10,11].
I. INTRODUCTION This paper focuses on the e-learning design for the
purpose of complementary to college education. Under this
With the rapid development of internet, e-learning purpose, we think resource recommendation application is
generally refers to the web-based education applications. E- helpful. As a complement of classroom teaching we do not
learning has a wide range in application, such as formal need the control of learning process, but we need to link up
distance education, training of company or enterprise, with the classroom teaching which many researches on
learning resources sharing web site and so on. Different learning resource recommendation didn’t pay more attention
application has different characteristic and leads to different to. So also from the application layer perspective, this paper
research emphasis in e-learning. This paper pays great gives the solution, model and strategy for the e-learning in
attention to the specific e-learning for college education. college education and try to connect it to the college courses
What should e-learning do with the traditional classroom and make the recommendations to fit the course demands.
education in college? In next section, we give an conceive of the e-learning on
Form the perspective of application layer, we think e- the problem of combination with classroom teaching. Then
learning can be summarized as two categories. One is for in section III, we present an idea to add a special instruction
providing learning resources, the other is for supporting the layer linking up to the college course and a solution is
specific teaching method, such as collaborative learning[11], designed to instruct learning based on adaptive learning
learning by game[1], inquiry learning[2], just-in-time situation. In section Č , we give the specific model for
learning, virtual laboratory and other special methods for personalized resource recommendation application based on
specific courses. The two research fields are both meaningful. learning situation. The last section summarizes the paper.
This paper chooses the research field on learning resources,
for learning resources are needed by most of the courses and II. E-LEARNING IN COLLEGE EDUCATION
abundant resources are all along the merits of e-learning
Our discussion is based on the computer science subject
while the weakness of classroom teaching. On the other hand,
which the author is familiar with, for other subject may have
learning through resources by oneself is also an important
different demands.
ability for college education.
The researches of e-learning based on learning resources A. Merits and Shortcomings
can also be seemed as two categories from the application Classroom teaching in college education has its merits
layer perspective. One is adaptive learning control. This for teacher’s knowledge and experience. So the classroom
category researches the learning control. Through the test

978-1-4673-7211-4/15 $31.00 © 2015 IEEE 1265


1264
DOI 10.1109/UIC-ATC-ScalCom-CBDCom-IoP.2015.229
teaching by teacher can make the knowledge easy to III. A SOLUTION FOR E-LEARNING
understand and brings students vivid and deep perception by Nowadays the education thought has turned to the
some classroom teaching methods such as discussion, viewpoint that teaching and learning both have the same
question and so on. On the other hand, classroom teaching emphasis. At the same time, blended learning is being paid
has some shortcomings coming from the limits by its form. more and more attention which just represents the
The lacking of time limits the contents in classroom. As a combination of teaching and learning.
centralized learning mode, the individual characteristic and E-learning adopts the autonomous learning mode by just
ability level can not be taken good care of. Classroom providing resources but not controlling the learning process.
teaching is controlled by teacher, and gives little freedom to But without teaching, it is just a resource platform, not a
students. learning system. In this paper, the combination of teaching
The shortcomings of classroom teaching are just the and learning lies in the learning instruction.
merits of e-learning. So using the e-learning merits to So for the e-learning in college education, we have an
compensate for the classroom teaching is a good idea. First, the infrastructure is resource library. Then we
combination direction. should link up with the college courses. At last, we should
B. Conceives of E-learning instruct the student’s autonomous learning in resource library.
The e-learning should be the complement of classroom According to the idea, we give a solution. We present a
teaching. We have four conceives: thought of learning situation to express the personalized
1) Divide the college course into two parts: classroom learning demands linking up with the classroom teaching.
part and e-learning part. The teacher should give a detailed Then the learning instruction can be done under the learning
course plan. Some knowledge points should be taught in situation.
class, and the others should be learned by self in e-learning A. Link up with the College Courses
system. The classroom contents should be the basic and The basis of e-learning system is rich resources. But as a
necessary knowledge using examples, exercises, questions, supplement of college education, the resources should be
discussions as teaching strategies. E-learning fulfils the tasks related to the course demands. So e-learning needs a course
which classroom teaching can not do such as remedying or and knowledge structure in whole subject. The teacher is
deepening the knowledge of classroom, strengthening the responsible for building and updating the knowledge
practical ability, expanding knowledge and so on. structure of his course. The teacher also submits the
2) The basis of e-learning is rich resources. There are resources for his course, and builds the feature vector of
many resources in internet, but many of them are repeated, resource for resource recommendation. Some resource
mixed by good and bad, and irrelevant to the college course. systems in web build the resource feature vector by feature
Searching in internet is always lack of efficiency for extraction algorithm, but in e-learning, the resources are
information overload. So resources provided by teacher are related to the course objectives the teacher designed, so the
necessary for meeting the learning demands. A resource feature vector built by teacher is more accurate and
library with teacher’s resources for his course is worthwhile. necessary. The system supports the course process in every
The resources should be diverse and fit the demands of term. The teacher marks the course process along with
students in different levels, different purposes and different classroom teaching. Every student enrolls in the system
characteristics. during his four years college time. The system should record
3) A resource library for a whole subject is more useful the student information. Students enroll his course every
than for a single course. For courses of a whole subject are term. At last, teacher gives the result of the course for every
relevant, so the resources can be used repeatedly. Some tasks student as the students’ knowledge structure. The system
of a course may involve several courses also. records the learning process and actions of every student.
4) The e-learning system as a complement of classroom
doesn’t need learning control for more freedom and B. Instruct based on Learning Situation
autonomous ability. Its purpose is remedying, strengthening, The teacher and the system cooperate to do the
expending, exploring after class according to student’s instruction. The teacher is responsible for the pedagogical
learning state. So under this purpose e-learning is a resource design on course plan, course objectives, course tasks and
system at first. Some learning resource systems are only for course knowledge demands. The system is responsible for
search. Some can give recommendation mostly by user perception of student status and gives advices according to
information or by collaborative filtering technique. These the teacher’s pedagogical design and other information. In
may seem not enough for college course. Researches should our idea, the teacher’s design for course is the guiding
pay attention to the “teaching and learning” pedagogy in ideology, the system does the actual instruction to students
internet circumstance in addition to the technical problems to through advice, the student still has great autonomy for the
provide better service. So as an e-learning in college instruction style is only advice.
education, it should be an actual complement of classroom A learning situation idea is provided. The learning
teaching. It should consider the course plan, learning situation in our e-learning is not for specific knowledge, it’s
objectives and demands of students to give guidance. a learning circumstance for student individuality. E-learning
system is a huge resource circumstance. Learning situation is
just a small room separating from the whole resource space

1265
1266
for a student with specific demands. In the proper learning describes the relationship of two knowledge points to reflect
situation, student’s learning may be of better efficiency and the knowledge structure of a course.
may fit the needs of student. 2) Resource model
So the solution is personalized resources The resource model is defined as follow:
recommendation based on learning situation. The learning RInfo = (resourceId, resourceName, path, media).
situation represents the learning demands or objectives, the RandK = (resourceId, courseId, knowledgeId).
instruction is hidden behind it. When entering the system, RandR = (resourceId1, resourceId2, relation).
students can learn resources by fundamental search method, RFeature = (resourceId, category, contentType,
or by help of system. The system can build several suitable difficulty).
learning situation. If student chooses a learning situation, Rinfo describes the basic information of a resource,
there are the suitable resources for specific learning demands RandK describes the relationship of resource and knowledge
hinted by the learning situation. point, RandR describes the relation of resource and resource,
The learning situation is designed with three pattern at RFeature describes the feature of a resource. category is
present: valued by the range of {knowledge, practice}. contentType is
z pattern one: remedy and improve learning mode. valued by the range of {classroomContent, depthContent,
This pattern is used for a supplement to classroom example, problemAnalysis, projectDesign, expandContent,
teaching. It offers several adaptive situation to extraKnowledge}.
choose through referring to the information about 3) Student model
course states, course objective, student’s learning The student model is defined as follow:
states and so on. The situation implies the demands SInfo = (studentId, studentName, class, specialty ).
such as remedy of the knowledge points in SandC = ( studentId, courseId, state, result).
classroom, strengthening the practical skills and so SandR = ( studentId, resourceId, time, feedback,
on. evaluation).
z pattern two: problem based exploration learning SFeatureOnAbility = ( studentId, score, cognitiveAbility).
mode. This pattern is for solving large problem such SFeatureOnPre = ( studentId, contentTypePreference,
as project, task and so on. It is used to instruct the contentPreference, situationPreference).
learning direction on a specific problem. SInfo describes the basic information of a student. SandC
z pattern three: no purpose learning mode. This pattern describes course selection information of student. SandR
is for the learning without clear learning objectives. describes the resources learning record of student.
It can be used to meet the student interests. SfeatureOnAbility describes the student features on learning
ability. score is a value as a evaluation of learning ability.
IV. A MODEL ON PERSONALIZED RESOURCE cognitiveAbility is also evaluation of learning ability but by a
RECOMMENDATION BASED ON LEARNING SITUATION value vector as (memorization, comprehension, analysis,
According to the solution in section III, we design a application, design, synthesis) to describe the cognitive
concept model for e-learning system. The model including ability of a student. SFeatureOnPre describes the student
the information structure model and the learning instruction feature on preference. contentTypePreference is a model
model. Information structure model expresses the describing the student preference to the resource content type.
information design for supporting the e-learning’s demand. contentPreference is also a model for the knowledge student
Learning instruction model is the strategy design for the preferred. situationPreference is a model describing the
solution of personalized resources recommendation based on preference to learning situation.
learning situation. 4) Pedagogy design model
The pedagogy design model is defined as follow:
A. Information Structure Model
KFeature = (knowledgeId, demand, type,
The information structure model describes the cognitiveDemand).
information the system needed. CTeaching = (knowledgeId, state).
1) Subject structure model CTask = (taskId, taskName, taskType, relatedK).
The subject structure model is defended as follow: CProject = (projectId, projectName, discription, taskType,
CInfo = (courseId, couseName, describe). relatedCK).
CRe = (courseId1, courseId2, relation). KFeature describes the knowledge point feature. demand
KInfo = (knowledgeId, knowledgeName, keyWords). is valued in the set{inClass, outClass}. type is the knowledge
KRe = (knowledgeId1,knowledgeId2,relation). type valued in the set{knowledge, algorithmAndApplication,
CInfo is a three tuple. It describes the course information. methodAndProcedure}. congnitiveDemand is a value vector
courseId, couseName, describe are the three elements of as (memorization, comprehension, analysis, application,
CInfo which express course identification, course name, design, synthesize). CTask is the course task teacher
course description. We use the same grammar to describe designed from the perspective of skill students should
other information units. CRd describes the relationship of master. CProject is the actual project teacher designed for
two courses to reflect the course structure of a subject, KInfo training the skills of students. taskType describes the type of
describes course knowledge point information, and KRe task or project valued in a set {memorization,
comprehension, analysis, application, design, synthesize}.

1266
1267
The model expresses the teacher’s pedagogical design. As an For pattern one, first get the course, then get the objective,
example of programming course, the task can be designed as at last get the knowledge point of a course. Since the
“Exhaustive algorithm application”, “program design using combination of the three parameters lead to large amount of
procedure oriented method”. learning situations. Then a strategy is needed to choose the
learning situations. The strategy is used to choose the
B. Learning Instruction Model for Personalized Resource learning situations which most fit the student’s information.
Recommendation based on Learning situation. The relevant student information including
1) Learning situation expression SFeatureOnAbility, SFeatureOnPre. At present, we just
In section III, we give three patterns for learning situation. designed a simplified strategy on course and knowledge
We designed the parameter vector to represent a specific point: choose the one recent changed. The personalized
learning situation. The three patterns have different strategy only are designed for the objective. We use rule as
parameter vectors. the method to implement the strategy. The front part of the
a) Pattern one: remedy and improve learning mode rule is student information element, the consequent part of
the rule is the objective and a value as degree. After
This pattern is used for complement of classroom
reasoning of the rule set, an accumulative degree for the
teaching. After class, the student may have some feedbacks
different objectives can be acquired. So the result is to sort
on his own learning states, so he would use the e-learning
the degree as the recommendation sequence. At last, the
system to learn according to his state. So the parameter
information of preference on learning situation
vector is designed to describe the learning demand or
(SFeatureOnPre.situationPreference) is also a reference for
objective.
adjusting the sequence.
situationParam1 = (course, knowledgePoint, objective)
For pattern two, first check the task or project of the
objective is used to describe the demand or objective of
courses the student takes part in, then get the parameters
learning. We designed the possible learning demands after
from the task or project information the teacher gives.
class: remedy the knowledge in class, strengthen the practice
For pattern three, get the parameters from the student
ability, deepen the knowledge, expand the knowledge out
model SFeatureOnAbility, SFeatureOnPre, SandR. And the
class, strengthen synthesis application ability.
userGroup comes from a user clustering algorithm by user
b) Pattern two: problem based exploration learning model information.
mode 3) Personalized recommendation of resources
This pattern is used for solving the tasks or projects the When student enters a learning situation, a room is
teacher assigned. It tries to instruct the exploration direction. established. In the room, the adaptive resources is
Any task or project is relevant to the knowledge and skill. So recommended for the demands the situation implied.
we designed a parameter vector, and try to instruct the According to the pattern of learning situation, the strategy is
exploration through knowledge point and problem nature. designed.
situationParam2 = (problem, courseKnowledgeSet, a) Pattern one
nature).
nature describes the question’s type such as summary In this pattern, the import principle is to accomplish the
and overview, deep research, problem analysis, design. learning objective. In this pattern, the recommendation
Different type needs different exploration method, needs method we choose is rule. The rule is used to acquire the
different resources to read. resource feature fitting the objective and user. To simplify
the rule and inference, the method is divided into four steps.
c) Pattern three: no purpose learning mode First, system filters the resources according to course and
This pattern is used for student without clear purpose. So knowledge points. The query method is enough in this step.
we need some other methods such as through interests to Second, system filters the resources according to
build his learning situation. So the parameter vector we objective. The rule is a deduction from objective value to
designed is: resource feature value (RFeature.category,
situationParam3 = (userPreference, userAbility, RFeature.contentType). So system can acquire the resource
histroyRecord, userGroup ). feature to filter the resource.
The parameters involves student’s preference, student’s Third, system filters the resources according to the
learning ability, student’s learning records, and student’s student learning ability level. This step is used to fit the
companions. student’s ability. In this step, system do the reasoning to
2) Dynamic generation of the learning situation proper difficulty of resource ( RFeature.difficult ).
When student enters the system, the learning situations Fourth, system adjusts the resource sets according to the
can be built for him dynamically. We use the parameter feedback about the resource on learning. This step is used to
vector to express a learning situation. adjust the difficulty (RFeature.difficult) and content
First, a basic method is filling in the parameters by (RFeature.contentType) according to the feedback on a
student himself. resource from two aspects: difficulty and satisfaction degree
Also, a strategy of auto filling in the parameters is (SandR.feedback).
provided. b) Pattern two

1267
1268
In this pattern, the important principle is to instruct on There are still some works left for the future. To validate
direction of exploration. So the knowledge points should be the high layer model, we did some simplification on details.
enough, and the resources should fit the problem type. In this The learning situation and recommendation strategies can be
pattern, a hybrid method is adopted. researched further. On the other hand, technology is also a
First, system filters the resources according to course and research directions for performance problems. The e-
knowledge points. The query method is enough in this step. learning’s application in college education is a rather
Second, system filters the resources according to the complex problem, we just present an idea from the direction
problem nature. This step uses rule to acquire the resource of improving e-learning to generate a virtual self-learning
feature, just like pattern one. room with learning demands, but other pedagogy also should
Third, system does the recommendation according to the be considered to resolve the problem.
user group. This step uses a different method to rule. It uses a
user-based recommendation. The method is a simplified user REFERENCES
based collaborative filtering algorithm for the neighbors are [1] Jui-Hung Chen, Shih T.K. and Jui-Yi Chen, "To develop the
easy to find. We can use the task or project as a nature ubiquitous adventure RPG (role play game) game-based learning
clustering. system," 2012 IEEE International Conference on Systems, Man, and
Cybernetics(SMC), IEEE Press, 2012, P: 2973 - 2978, DOI: 10.1109/
c) Pattern three ICSMC.2012.6378247.
In this pattern, we cannot do the recommendation [2] Guan Chen Zhi and Liu Zhen Y, “a An inquiry-based blended
according learning objective. So the recommendation can be learning system for computer network curriculum,” 2013 8th
International Conference on Computer Science & Education
done according to the preference, action and user group. In (ICCSE), IEEE Press, April 2013, pp. 1340 – 1345, DOI:
this pattern, a hybrid method is also adopted. 10.1109/ICCSE.2013.6554130.
First, system do the recommendation according to the [3] Shih-Ming Bai and Shyi-Ming Chen, “Automatically constructing
preference (SFeatureOnPre.contentTypePreference, concept maps based on fuzzy rules for adapting learning systems,”
SFeatureOnPre.contentPreference). The preference Expert Systems with Applications, vol.35, 2008, pp.41-49.
information is extracted from the learning actions. The [4] Ana Iglesias, Paloma Martínez, Ricardo Aler and Fernando
contentTypePreferece is a vector of value to the resource Fernández, “Learning teaching strategies in an Adaptive and
content type set. The contentPreference is three sets for Intelligent Educational System through Reinforcement Learning.”
Applied Intelligence, vol. 31, August 2009, pp. 89-106.
courses, knowledge points, keywords. In this step, the
[5] Antonija Mitrovic, Stellan Ohlsson and Devon K. Barrowa, “The
content filtering method is adopted. The similarity degree of effect of positive feedback in a constraint-based intelligent tutoring
student preference vector and the resource feature vector is system,” Computers & Education, vol.60, 2013, pp. 264-272.
used for recommendation. [6] Antony J., Ramakrishnan S., Joy M.T.and Thottupuram R., “SMS
Second, system do the recommendation according to the Based E-assessments Enabling better Student Engagement,
neighbor users. This step computes the similar degree of user Evaluation and Recommendation Services in E-learning Making Use
to get neighbor sets according to the user feature of of Fuzzy Rules and Course Ontologies,” 2012 IEEE Fourth
International Conference on Technology for Education, IEEE Press,
preference and ability, then do the user based collaborative 2012, pp.119-125, DOI: 10.1109/T4E.2012.9.
filtering. [7] Wei Kuang, Nianlong Luo and Zilei Sun, “Resource recommendation
Third, system adjust the resources sets according to the based on topic model for educational system” 2011 6th IEEE Joint
feedback about the resource on learning. The thoughts of International Information Technology and Artificial Intelligence
item based collaborative filtering is adopted in this step. This Conference (ITAIC) (Volume:2 ), IEEE Press, Aug. 2011, pp.370-
step is used for sorting the resources in resource set already 374, DOI: 10.1109/ITAIC.2011.6030352.
established by similar degree of resource to resource. [8] ZhiHong Ruan, Guiyang, Yi Jin and Zhuying Lin, “Research of
Personalized Learning Resource Recommendation Based on Learner's
V. FDI,” Education International Workshop on Technology and
CONCLUSION
Training, IEEE Press, Dec. 2008, pp. 84- 87, DOI:
The paper used some popular techniques to support the 10.1109/ETTandGRS.2008.114.
learning instruction model, such as rule based reasoning, [9] Wang Yonggu, Qiu Feiyue, Zhao Jialong and Liu Hui, “Research on
content based filtering, user based collaborative filtering and Personalized Recommendation of Learning Resources Based on
Collaborative Filtering Recommandation Technology,” Distance
item based collaborative filtering. But problems of technique education journal, vol. 3, 2011, pp.66-71,
layer are not the paper’s focus. The paper’s work is a generic
[10] Junzhou Luo, Fang Dong, Jiuxin Cao, Aibo Song and Bo Liu, “A
model on high layer in order to get more proper service for Multicontext-aware Resource Recommendation Mechanism for
college course teaching. The model constructs the Service-oriented Ubiquitous Learning Environment,” Third
connection with classroom teaching, uses learning situation International Conference on Pervasive Computing and Applications,
to represents the learning demands under the specific context (ICPCA 2008), IEEE Press, Oct. 2008, pp. 792 – 797, DOI:
10.1109/ICPCA.2008.4783717.
of college course. So the resource recommendation strategies
[11] Akbari, F.; Taghiyareh, F., E-SoRS: A personalized and social
is first based on the course plan, objective and task. Though recommender service for E-learning environments, 2014 8th National
with the popular recommendation algorithms, the high layer and 5th International Conference on e-Learning and e-Teaching
model and synthesis strategies play important roles to (ICeLeT), IEEE Press, 2014, pp.1-12, DOI:
acquire a learning system than a resource recommendation 10.1109/ICELET.2014.7040615.
system only.

1268
1269

Вам также может понравиться