Professional Documents
Culture Documents
1 RV
1 RV
1 RV
Corresponding Author:
Dede Anjar Permana
School of Computer Science, Faculty of Informatics, Telkom University
Bandung, Indonesia
Email: dedeanjarpermana@gmail.com
1. INTRODUCTION
Algebra is an essential section of mathematics concept in science, statistics, business, or current
technology. The proper comprehension of algebra is a prerequisite of mastery and advance in higher
mathematics and life [5]. Preparing students to understand algebra easier is a challenge for the teachers. A
research show that junior high school pupils possess fallacy in the foundation of algebra concept, variables,
algebraic expressions, equations and word problems [5]. In the initial formative, the average of student’s
achievement was 55.78, plenty lessened from the least possible grade which is 75 [40]. Thus, it needs an
evaluation to improve both teaching methods and treatments for students who made misconceptions in
solving algebra problems.
Beside the low score issue, another issue in teaching in a classroom with a varied group of students
is very challenging, since appropriate teaching style that works for some group students is unlikely can be
applied for entire students in the class. Although the most teacher are aware that students have different rate
pace and different improvement. During the learning process, usually teacher only show limited spotlight to
every pupil, because the school conducts the “one-size fits all” procedure [24]. Another reason is the usage
of e-learning. According to [23] that e-learning alters traditional teaching, but these methods are unable to
give the learners with adaptive and personalized experienced. The system just provides the same materials for
all students without personalization.
In the real life, it is really hard for teachers how to give help one on one student when they face a
problem. It is really challenging how to give support personalized when students working with different
question and different level difficulty. Usually, a teacher will give explanation about a question in front of
class, for entire students not to one-on-one student. This is in line with [24] that teachers still conduct “one
size fits all” for all students even when teachers give a help to student in one classroom. Regards to interview
with some mathematics teachers in secondary level, some activities that would help in solving a problem are
direct feedback, hint and giving worked example to students.
Regard to the result of numbers of research in ITS [16], interactive problem-solving support and
example-based problem-solving support are the least items which has been researched if compares to
adaptive presentation and navigation support [16]. Even though, much of the present research in cognitive
psychology has shown that students learn better when utilized in solving problems and really supported by
interactive problem solving [16].
In comparison to unsupported problem solving, numerous studies have demonstrated that learning
from worked examples is beneficial for beginners [3, 44, 45, 47, 55]. In [44] worked examples have been
proved to be effective in the classroom. According to research published in [45], using worked examples can
improve students' performance. According to research published in [45], using worked examples can improve
students' performance. [3, 47]. Based on [45] student who were given worked example finished more of the
needed problem and were less likely to drop out of tutoring than those who just did in problem solving. The
result has demonstrated that worked examples are really useful [45].
Applied ITS in math has shown in activeMath [58]. ActiveMath is used in adaptive presentation.
The result has demonstrated that use activeMath students learn mathematics more effective [58]. However,
the usage of adaptive help in other subject such as in Fundamentals of artificial intelligence [1] and Physics
[42]. In Mathema [42] the result of student’s average has increased. The difference average score between
student who do not used hint and feedback was 63.14, while students who used hint and feedback the average
score was 71.04. This result has shown the difference improvement in learning result.
This study focuses on interactive problem-solving support in learning polynomial for high school students.
According to previous studies, interactive problem-solving support, in the forms of hints and feedback, have
been applied [1, 16, 17, 18, 19, 21] and worked examples have been applied in [44, 45, 46].
To improve student result in math especially in polynomial algebra, then build an ITS based on
adaptive learning. How to support students when students solve problem in polynomial algebra. By providing
support adaptive help it will increase student result of students.
If it seen based on the delivering by time, feedback is divide become two parts, first immediate
feedback and second delay feedback. Immediate feedback is giving signal or feedback to students after each
step or action in the problem-solving process has been done. While delay feedback is giving signal or
feedback to students after students submit whole their answers of problem solving [1]. Furthermore, on
immediate feedback itself the division become two parts, first positive feedback which is if the student
answers correctly, second negative feedback which is if students answer incorrectly [1]. Delay feedback is
giving feedback to students for a whole correction to the problems which student was being solved. It could
be general or specific hints. Feedback and hints were also proposed in [16, 17] and applied to group
collaboration. During learning in a group collaboration, students will get hints is shown in the dashboard if
they made a mistake. This research applied rule modeling, constraint-based modeling, and knowledge
tracing.
Another study [18] proposed interactive problem-solving support, not only in a cognitive domain,
but also in metacognitive and affective skills. This research used an association rule mining for behavior
discovery. Different with the aforementioned research, an ITS research in [19] applied hints to computer
programming, especially for beginner students in computer science. This research used two kinds of hints,
Int J Inf & Commun Technol, Vol. 99, No. 1, Month 2099: 1-1x
3Int J Inf & Commun Technol ISSN: 2252-8776
which are explanation text and self-explained hints. The research shows that hints with textual explanations
could significantly improve the students’ programming performance
2. RESEARCH METHOD
We propose an ITS that offers adaptive help for students, consist of feedback, hint and worked-
examples. The topic implemented is polynomial algebra for grade 8 in secondary level. The proposed ITS
applies the general architecture of ITS, but the focus is on the pedagogical model, domain model and student
model.
A. Pedagogical Model
Pedagogic model is a process which is adapted in the classroom. In this study the pedagogical model
is set by providing help for student when the students face difficulties in solving a problem. The help would
be provided are feedback, giving hint and giving worked example for students.
Paper’s should be the fewest possible that accurately describe … (First Author)
4 ISSN: 2252-8776
The general overview of learning scenario is shown in figure 1. The first time, student register to the
system and try to answer pre-test. In this study the score of pre-tests do not used for initial learning, pre-test
would be used for data analyzing. After student answer pre-test, afterward student look at to the exercise.
When student answer the question, students would get support or help if they cannot answer the question,
vice versa student do not get help if they can answer question and could continue to the next question. Kind
of help would be given to students if they cannot answer the questions are direct feedback, hint and giving
worked examples.
In the first time the students face the no 1 and related to competence 1. If students answer correctly
the system will update student model directly and students will not gain help. While for student who could
not answer the question will gain help. The first-time kind of helps would appear after submitting the answer
is feedback in the form of signals of correct and wrong. In this moment the student would be given button to
tried answer question again.
Then, after click button to try the answer again, the procedure was same as above, if student could
answer correctly the system will update student model, for those who cannot answer again, the system will
not only give the feedback, but also provides hints. These hints based on students’ errors. Then button tried
answer again will appear. The procedure was same as mentions before. If students could answer the question
the system would update the student’s model, for those who cannot answer will get feedback that the answer
is wrong. In addition, worked examples are also given. A worked example contains a problem that has been
solved by an expert. It has a similar targeted knowledge with a problem being solved by student.
After getting feedback, hints and worked examples, student re-try the question for the last time, if in the last
time student could not answer correctly, the system will provide the correct answer. In this session student
just put the correct answer that already provided by the system.Domain Model.
The domain knowledge is the knowledge where the system is teaching. In this study the domain of
knowledge is about algebra for grade 8 of the secondary level. The domain knowledge is organized in a
hierarchy, in which topics are the top entity which includes some units and each unit consists of several
chapters.
B. Student Model
The student model is the core component of an ITS. Ideally, it should contain as much knowledge.
In this study the student model using Performance Factor Analysis (FPA). PFA is used to measure student
performance when students answer questions. Student performance consist of accumulative score from
students answer even from true answer or false answer.
Int J Inf & Commun Technol, Vol. 99, No. 1, Month 2099: 1-1x
5Int J Inf & Commun Technol ISSN: 2252-8776
In PFA, student mastery is expressed as the probability that a student will produce the correct
answer to a problem. PFA models student performance using a logistic regression equation with two
variables - the cumulative number of correct and incorrect answers that a student has produce the current
skill. It also uses three parameters:
gamma (γ), the degree for a correct answer is associated with better future performance.
rho (ρ), the degree for an incorrect answer is associated with better future performance;
beta (β), the overall ease or difficulty of the KC to be learned.
❑
m=β q + ∑ ❑ ( γ j si , j + ρ j f i , j ) (1)
j ∈KCs
Where:
mi = score of student’s performances on question i
i = question for i = 1,2,3 …. n
j = skill for j = 1,2,3 … m
s = item success
f = item failure
β i = coefficient of easiness or difficulty of question i
γ j = history of item corrects
ρ j = history of item incorrect
Si , j and ρ j f i , j are two observable variables in the equations, representing the number of prior successful and
unsuccessful practices performed by student on skill j
In order value of m become the probability value (range 0-1), value of m would be changed become
probability form by equation 2:
1
p ( m )= −m (2)
1+ e
We propose the PFA approach to perform student model in this study. The graphical for PFA is like figure 4.
TABLE 1. THE AVERAGE OF WEIGHTED OF DIFFICULTY QUESTION, HISTORY OF SUCCESSFUL, HISTORY OF FAILURE.
Paper’s should be the fewest possible that accurately describe … (First Author)
6 ISSN: 2252-8776
Table 1 shows relationship among question, competence, taxonomy bloom, beta ( β ¿, gamma ( γ ¿ and
rho ( ρ ¿. This table shows that every question connects to suitable competence and taxonomy bloom. Beta,
gamma and rho is coefficient which is needed in computational process which were collected from three
expert teachers and was taken the average of them. This table also shows information about the previous
successful and failure of students for each competence. The number of succeeding in competence C3.2 is 0.74
which is the highest amount among all 3 competencies. In other hand, the failure history has the biggest
amount in competence C3.3 as much as 0.53. By those numbers, can be concluded that most of students are
able to solve questions in second competence, but found some difficulties to answer competence C3.3
questions correctly. In line with that, can be written that the difficulty’s intentions of competence C3.3 is
higher than 2 other competencies.
After filtering for question and worked example, another way to get recommendation is to count
similarity between question and worked example by using Jaccard similarity. Jaccard Similarity use weighted
to counted similarity between question and worked example. The formulation for Jaccard like formulation
below.
Int J Inf & Commun Technol, Vol. 99, No. 1, Month 2099: 1-1x
7Int J Inf & Commun Technol ISSN: 2252-8776
After counting similarity every question and every worked example, all of the result from Jaccard will
be put in webbing matrix like table 3.
If is seen in table 3.8 that the value of pair WE1, Q1 has the highest value. Pair of WE1, Q1 is the best
candidate to be recommended to a student, if finds difficulty in question 1.
The evaluation of webbing matrix is using Normalized Discounted Cumulative Gain (NDCG). Rather
than explicitly estimating ratings, several recommender systems provide estimations of the underlying ranks
[57]. In practice, the recommender system creates a rank item for a user and then recommends the top-k
items. The value of k varies depending on the system, item, and user. In general, highly rated goods should
be listed higher than items that are not highly rated [58]. Based on the result, has been gotten that NDCG for
this webbing matrix is 93%, it means the effectiveness of using webbing matrix is 93%.
Paper’s should be the fewest possible that accurately describe … (First Author)
8 ISSN: 2252-8776
Fig. 5 Comparation N-gain score between experimental group and control group.
By comparing N-gain scores between control group and experimental group we can conclude that
students who use adaptive help in polynomial algebra have five-time differences. This can be concluded
students who used adaptive help have better learning outcome than student who do not use adaptive help
REFERENCES
[1] A. Anohina, “Advances in Intelligent Tutoring Systems: Problem-solving Modes and Model of Hints,” International Journal of
Computers, Communications & Control, pp. 48-55, 2007.
[2] N.N. AbuEloun, S. S. A. Naser, “Mathematics intelligent tutoring system,” International Journal of Advanced Scientific Research,
pp 11-16, 2017.
[3] R. J. C. M. Salden, K. R. Koedinger, A. Renkl,V. Aleven & B. M. McLaren, “Accounting for Beneficial Effects of Worked
Examples in Tutored Problem Solving ,“ Article in Educational Psychology Review, 2010.
[4] GÜNEL, R. AŞLIYA, “Determining Difficulty Of Questions In Intelligent Tutoring Systems”, The Turkish Online Journal of
Educational Technology (TOJET), 2009.
[5] E. S. Fumador, D. D. Agyei, “Students’ Errors And Misconceptions In Algebra: Exploring The Impact Of Remedy Using
Diagnostic Conflict And Conventional Teaching Approaches,” International Journal of Education, Learning and Development,
pp.1-15, 2018.
[6] https://www.mathunion.org/icmi/role-mathematics-overall-curriculum accessed 13 July 2021 09:00
[7] F. S. Gharehchopogh, Z. A. Khalifelu, “Using Intelligent Tutoring Systems in Instruction and Education,“ 2nd International
Conference on Education and Management Technology, 2011.
[8] A. O. Mahdi, M. I. Alhabbash, S. S. A. Naser, “An intelligent tutoring system for teaching advanced topics in information
security,” 2016.
[9] A. Karaci, H. I. Akyuz, G. Bilgici, N. Arici, “Effects of Web-based Intelligent Tutoring Systems on Academic Achievement and
Retention,” International Journal of Computer Applications, 2018.
[10] S. Niwattanakul, J. Singthongchai, E. Naenudorn and S. Wanapu, “Using of Jaccard Coefficient for Keywords Similarity,”
Proceedings of the International Multi Conference of Engineers and Computer Scientists 2013.
[11] P. D. Reddy and M. Sasikumar, "Student Model for an Intelligent Language Tutoring System," 2014 IEEE 14th International
Conference on Advanced Learning Technologies, 2014, pp. 441-443.
[12] N. M. Noha, A. Ahmada, Shamimi, Halima and A. M. Alia, “Intelligent Tutoring System Using Rule-Based And Case Based: A
Comparison,” Procedia - Social and Behavioral Sciences, 2012.
[13] P. Phobun, J. Vicheanpanya, “Adaptive intelligent tutoring systems for e-learning systems,” Procedia Social and Behavioral
Sciences, pp 4064–4069, 2010.
Int J Inf & Commun Technol, Vol. 99, No. 1, Month 2099: 1-1x
9Int J Inf & Commun Technol ISSN: 2252-8776
[14] H. Gamboa and A. L. N. Fred, “Designing Intelligent Tutoring Systems: A Bayesian Approach,” 2001.
[15] P. I. Pavlik, H. Cen and K. R. Koedinger, “Performance Factors Analysis-A New Alternative to Knowledge Tracing,“ Proceedings
of the conference on Artificial Intelligence in Education: Building Learning Systems that Care: From Knowledge Representation
to Affective Modelling, pp 531–538, 2009.
[16] A. Papadimitriou, M. Grigoriadou, G. Gyftodimos, “Interactive Problem Solving Support in the Adaptive Educational Hypermedia
System MATHEMA,” IEEE Transactions On Learning Technologies, 2009.
[17] E. Walker, N. Rummel, K. R. Koedinger, “Adaptive Intelligent Support to Improve Peer Tutoring in Algebra,” International
Artificial Intelligence in Education Society, 2013.
[18] C. Conati, S. Karda “Student Modeling: Supporting Personalized Instruction, from Problem Solving to Exploratory Open-Ended
Activities,” Association for the Advancement of Artificial Intelligence, 2013.
[19] S. Marwan, J. J. Williams, T. Price “An Evaluation of the Impact of Automated Programming Hints on Performance and
Learning,” ICER, 2019.
[20] P. Brusilovsky and C.Peylo, “Adaptive and Intelligent Technologies for Web-based Education,” International Journal of Artificial
Intelligence in Education, 2003.
[21] R. J. C. M. Salden, V. Aleven, R. Schwonke A. Renkl “The expertise reversal effect and worked examples in tutored problem
solving,” Springer Science Business Media, 2009.
[22] H. Cen, R. K. Koedinger and B Junker, “Learning Factors Analysis-A General Method for Cognitive Model Evaluation and
Improvement,” Intelligent Tutoring Systems, Lecture Notes in Computer Science, 2006
[23] S. Duo and Z. C. Ying “Personalized E- learning System Based on Intelligent Agent,” International Conference on Applied Physics
and Industrial Engineering, 2012.
[24] S. Meliana, D. Nurjanah “Adopting Good-Learners’ Paths in an Intelligent Tutoring System,“ IEEE International Conference on
Teaching, Assessment, and Learning for Engineering (TALE), 2018.
[25] R. S. J. d. Baker, A. T. Corbett, and V. Aleven, “More Accurate Student Modeling through Contextual Estimation of Slip and
Guess Probabilities in Bayesian Knowledge Tracing,” International Conference on Intelligent Tutoring Systems, 2008.
[26] R. Hosseinia, P. Brusilovsky, “A Study of Concept-Based Similarity Approaches for Recommending Program Examples,” New
Review of Hypermedia and Multimedia, 2017.
[27] R. Nkambou, J. Bourdeau and R. Mizoguchi, “Introduction: What Are Intelligent Tutoring Systems, and Why This Book?,”
Advances in Intelligent Tutoring Systems, 2010.
[28] J. Reye “Student Modelling based on Belief Networks,” International Journal of Artificial Intelligence in Education, 2004.
[29] M. Yudelson, K. R. Koedinger, “Individualized Bayesian Knowledge Tracing Model,” International Conference on Artificial
Intelligence in Education, 2013.
[30] Z. A. Pardos and N. T. Heffernan, “Modeling Individualization in a Bayesian Networks Implementation of Knowledge Tracing,”
Conference: User Modeling, Adaptation, and Personalization, 2010.
[31] T. Käser, S. Klingler, A. G. Schwing and M. Gross, "Dynamic Bayesian Networks for Student Modeling," in IEEE Transactions on
Learning Technologies, pp. 450-462, 2017.
[32] A. Díaz and P. Gervás, “Adaptive User Modeling for Personalization of Web Contents,” Adaptive Hypermedia and Adaptive Web-
Based Systems, 2004.
[33] V. Aleven, E. A. McLaughlin, R. A. Glenn and K. R. Koedinger, “Instruction based on adaptive learning technologies, ” Handbook
of research on learning and instruction, pp. 522-560. 2017.
[34] J. Hattie and H. Timperley, “The Power of Feedback,” Review of Educational Research pp. 81–112, 2007.
[35] P. Brusilovsky, “Adaptive Hypermedia,” User Modeling and User-Adapted, 2001.
[36] P. Brusilovsky, “Methods and techniques of adaptive hypermedia,” User Modeling and User Adapted Interaction, 1996.
[37] H. M. Seffrin, G. L. Rubi, P. A. Jaques, “A Dynamic Bayesian Network for Inference of Learners’ Algebraic Knowledge,” ACM
Symposium on Applied Computing, 2014.
[38] C. Conati, “Bayesian Student Modeling” In: Nkambou R., Bourdeau J., Mizoguchi R. (eds) Advances in Intelligent Tutoring
Systems. Studies in Computational Intelligence, pp.281-299, 2010.
[39] Paul De Bra “AHAM: A Dexter-Based Reference Model for Adaptive Hypermedia,” Proceedings of the 10th ACM Conference on
Hypertext and Hypermedia: Returning to Our Diverse Roots, 1999.
[40] E. Syahputra, “Increasing of Students’ Achievement in Polynomial by Using Jigsaw Method,” Journal of Education and Practice,
2014.
[41] J. Dāboliņš and J. Grundspenkis, “The Role of Feedback in Intelligent Tutoring System,” Applied Computer Systems, pp. 88-93,
2013.
[42] K. Vanlehn, J. A. Shapiro and C. Lynch, “The Andes physics tutoring system: Lessons learned,” International Journal of Artificial
Intelligence in Education, 2005.
[43] G. Hume, J. Michael and A. Rovic, “Hinting as a Tactic in One-on-One Tutoring,” The Journal of Learning Science, pp. 23-47,
1996.
[44] N. Green, B. D. Eugenio and D. Fossati, “Towards Adaptive Worked-Out Examples in an Intelligent Tutoring System,” Artificial
Intelligence in Education, pp.94-99, 2019.
[45] B. Mostafavi, G. Zhou, C. Lynch, M. Chi, T. M. Barnes, “Data-Driven Worked Examples Improve Retention and Completion in a
Logic Tutor,” International Conference on Artificial Intelligence in Education, 2015.
[46] R. K. Atkinson, S. Derry, A. Renk, D. Wortham, ”Learning from Examples: Instructional Principles from the Worked Examples
Research,” Review of Educational Research, pp.181-214, 2000.
[47] A. S. Najar, A. Mitrovic, K. Neshatian, “Using Eye Tracking to Identify Learner Differences in Example Processing,” International
Conference on Artificial Intelligence in Education, 2015.
[48] N. Sumarna, Wahyudin and T. Herman, “The Increase of Critical Thinking Skills through Mathematical Investigation Approach”
IOP Conf. Series: Journal of Physics: Conf. Series 812, 2017.
Paper’s should be the fewest possible that accurately describe … (First Author)
10 ISSN: 2252-8776
[49] A. Shemshack and J. M. Spector, ”A systematic literature review of personalized learning terms,” Smart Learning Environments,
2020.
[50] V. Esichaikul, S. Lamnoi, C. Bechter, ”Student Modelling In Adaptive E-learning Systems,” Knowledge Management and E-
Learning, pp. 342-355, 2011.
[51] M. Wenting, Olusola O. Adesope, John. Nesbit and Qing Liu, “Intelligent Tutoring Systems and Learning Outcomes: A Meta-
Analysis” Journal of Educational Psychology, 2014.
[52] Y. Chen, X. Li, J. Liu, Z. Ying, “Recommendation System for Adaptive Learning,” Applied Psychological Measurement, 2017.
[53] J. Aguilar, P Valdiviezo-Dıaz and G. Riofrio, “A general framework for intelligent recommender systems,” Applied Computing
and Informatics, 2016.
[54] J. Booth, K. McGinn, C. A. Barbieri, L. K. Young “Misconceptions and Learning Algebra,” In book: And the Rest is Just Algebra,
pp.63-78, 2017.
[55] A. Renkl and R. K. Atkinson, “Learning from worked-out examples and problem solving,” Cognitive load theory and research in
educational psychology, pp. 91–108, Cambridge University Press. 2010.
[56] Y. Gong, J. E. Beck, N. T. Heffernan, “Comparing Knowledge Tracing and Performance Factor Analysis by Using Multiple Model
Fitting,” Intelligent Tutoring Systems, 2010.
[57] Silabus Mata Pelajaran Matematika kelas IX Sekolah Menengah Pertama/ Madrasah Tsanawiyah Kurikulum 2013
[58] Melis and J. Siekmann, “ActiveMath: An Intelligent Tutoring System for Mathematic,” Artificial Intelligence and Soft Computing
- ICAISC 2004.
Int J Inf & Commun Technol, Vol. 99, No. 1, Month 2099: 1-1x
Int J Inf & Commun Technol
11 ISSN: 2252-8776
Deepika Koundal received her Ph.D. and M.E. Degrees in Computer Science
and Engineering from UIET, Panjab University, Chandigarh, India and B.Tech. Degree in
Computer Science & Engineering from Kurukshetra University, Kurukshetra, Haryana,
India. She is currently working as an Assistant Professor in the Department of Computer
Science and Engineering at Chitkara University Institute of Engineering & Technology
(CUIET), Chitkara University, INDIA since March, 2016. She is having 10 years of
teaching and research experience at various reputed Universities of India. She was
previously associated with NIT Hamirpur as an Assistant Professor. Prior to that she
worked as a Research Associate at UIET, Panjab University, Chandigarh. Her Ph.D. thesis
focused on Automated delineation of thyroid nodules in Ultrasound Images. She is actively
pursuing research in Medical Image Processing. She is the awardee of TEQIP-II fellowship
during Ph.D. Her areas of expertise are Computer Vision, Image & Video Processing,
Pattern Recognition, Machine Learning, Soft Computing, Artificial Intelligence,
Information Retrieval and related fields. from Neutrosophic Science Association from
University of Mexico for her outstanding publication in the Applied Soft Computing
Journal (SCI Journal). She has published 22 research articles in reputed SCI and scopus
indexed journals and conferences. She can be contacted at email: koundal@gmail.com.
Paper’s should be the fewest possible that accurately describe … (First Author)