Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

0

Share

Download to read offline

Evaluate adaptive learning model at ICL December 05 2014

Download to read offline

Evaluate adaptive learning model at ICL December 05 2014

Related Books

Free with a 30 day trial from Scribd

See all
  • Be the first to like this

Evaluate adaptive learning model at ICL December 05 2014

  1. 1. Evaluating Adaptive Learning Model I. Introduction II. Evaluation criterions III. An evaluation scenario IV. Conclusion ICL WEEF 2014 : Evaluating Adaptive Learning Model (December 05 2014) Author: Loc Nguyen Sponsor: Prof. Dr. Dong Thi Bich Thuy Affiliation: Department of IS, Faculty of IT, University of Science ICL WEEF 2014 1 11/28/14
  2. 2. I. Introduction ICL WEEF 2014 2 Adaptive System Selection Rules User Modeling System User Model TARGET: Adaptive System changes its action to provide learning materials for every student in accordance with her/his model Learning Materials 11/28/14
  3. 3. I. Introduction This research has two goals 1. Firstly, research proposes criterions to evaluate adaptive learning model. 2. Secondly, research gives some scenarios as an example that applies criterions above into performing evaluation task in concrete situations. 11/28/14 ICL WEEF 2014 3
  4. 4. II. Evaluation criterions This research proposes 3 criterions of evaluation 1. Criterion α so-called system criterion tells us how adaptive learning system works with/without user modeling system 2. Criterion β so-called academic criterion tells us how well modeling server helps users to study. 3. Criterion γ so-called adaptation criterion or satisfaction criterion measures the quality of adaptation function of learning system with the support of modeling server. 11/28/14 ICL WEEF 2014 4
  5. 5. II. Evaluation criterions Calculating criterion α User knowledge Ki = {ki U, ki U,…, ki U} has sample variance si 2. User knowledge Kj = {kj U, kj U,…, kj U} has sample variance sj 2. F-distribution is used to test two variances: F= si 2 / sj 2. If F < f0.95, n-1, n-1 then the null hypothesis is rejected, criterion α get Boolean value true, indicating the preeminence of user modeling system 11/28/14 ICL WEEF 2014 5
  6. 6. II. Evaluation criterions Calculating criterion α 1. The linear regression function is constructed by least square method. 2. Estimated knowledge is computed based on regression function. 3. Knowledge error is the difference between estimated knowledge and user’s real knowledge. 4. Measure α is the inverse of knowledge error. 11/28/14 ICL WEEF 2014 6
  7. 7. II. Evaluation criterions KA = (k1 A, k2 Calculating criterion β A) has sample variance sA A,…, kn 2 and sample mean A . K= (kB, kB 1 2 B) has sample variance sB B,…, kn 2 and B sample mean The measure β for each group is computed as accumulative probability of assumption user in such group has mastered over course. 11/28/14 ICL WEEF 2014 7
  8. 8. II. Evaluation criterions Calculating criterion γ • Suppose a questionnaire is built up by expert and it is composed of n questions Q = (q1, q2, …, qn) • By the simplest way, criterion γ is defined as the ratio of the number of satisfied users to the whole number of users 11/28/14 ICL WEEF 2014 8
  9. 9. II. Evaluation criterions Calculating criterion γ 1. Firstly, rating matrix is “shrunk” by projecting it onto its eigenvectors. The number of columns is much smaller than the number of questions. 2. Secondly, each column of matrix corresponding to each question is assumed as a statistical distribution Fi. Thus the mean of Fi is estimated by μi. 3. Finally, the mean vector of this matrix is composed of all estimates μi and the criterion γ is the module of such mean vector. 11/28/14 ICL WEEF 2014 9
  10. 10. III. An evaluation scenario This evaluation scenario is divided into 3 main acts in which students and teacher play the roles of actors 1. Study act: Teacher teaches and students learn in both face-to- face manner and e-learning manner via website. Suppose students are classified into three groups A, B and C. Groups A and B represent face-to-face manner and e-learning manner via website, respectively. Especially, group C represents e-learning manner with support of user model, namely Bayesian network. 2. Feedback act: Students give feedbacks to teacher and teacher collects and analyzes them. 3. Evaluation act is done by teacher; thus, criterions α, β and γ are calculated according to data collected from two above acts. The quality of adaptive learning in groups A, B and C are determined based on such criterions. 11/28/14 ICL WEEF 2014 10
  11. 11. III. An evaluation scenario Study act has 5 scenes: 1. Teacher builds up school’s curriculums and set up adaptive e-learning website with/without the support of user modeling system. 2. Teacher teaches and students in groups A, B and C learn by face-to-face manner. 3. Students in groups B and C go on website and study by themselves. Teacher monitors them and put up important notice. 4. Students in groups A, B and C do tests and exercises via website. 5. Teacher evaluates students based on their test results. 11/28/14 ICL WEEF 2014 11
  12. 12. III. An evaluation scenario Feedback act has 3 scenes: 1. Teacher creates the questionnaire to survey students’ feeling about both adaptive learning website and curriculum such as very satisfied, satisfied and not satisfied. 2. Students answer or rate on such questions online. 3. Teacher collects students’ feedbacks and analyzes them. 11/28/14 ICL WEEF 2014 12
  13. 13. III. An evaluation scenario Evaluation act has 2 scenes: 1. Teacher calculates three criterions based on students’ feedback and test results. 2. Teacher makes the decision about the quality of face-to-face teaching manner and e-learning manner with/without support of user modeling system. 11/28/14 ICL WEEF 2014 13
  14. 14. IV. Conclusion • As aforementioned, there are three criterions such as system criterion α, academy criterion β and adaptation criterion γ. • That two of three criterions, concretely α and β, assessing user knowledge implicates that evaluation of adaptive learning model focuses on the effect of education which is ability to help students to improve their knowledge although adaptation and personalization are significant topics in adaptive learning. • You can recognize that the education never goes beyond the main goal that increases amount of human knowledge. 11/28/14 ICL WEEF 2014 14
  15. 15. IV. Conclusion • Evaluation scenario, an example for demonstrating how to determine these criterions, indicates that study is lifelong process for everyone and so, classes and courses are short movies in this lifelong process. • Both students and teachers are actors and their roles can mutually interchange, for example, teaching is the best way to learn and student is the best teacher of teacher. 11/28/14 ICL WEEF 2014 15
  16. 16. Reference 1. Alfred Kobsa. Generic User Modeling Systems. User Modeling and User-Adapted Interaction 2006 (UMUAI-2006). 2. Peter Brusilovsky and Eva Millán. User Models for Adaptive Hypermedia and Adaptive Educational Systems. The Adaptive Web – Methods and Strategies of Web Personalization. Lecture Notes in Computer Science, Springer-Verlag Berlin Heidelberg 2007, ISBN-13: 978-3-540-72078-2. 11/28/14 ICL WEEF 2014 16
  17. 17. THANK FOR YOUR ATTENTION 11/28/14 ICL WEEF 2014 17

Evaluate adaptive learning model at ICL December 05 2014

Views

Total views

485

On Slideshare

0

From embeds

0

Number of embeds

9

Actions

Downloads

31

Shares

0

Comments

0

Likes

0

×