A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences
3 likes750 views
The document presents a recommender system designed for students, utilizing social knowledge and assessment data to enhance learning experiences. It discusses the motivation behind the system, its processes, and the incorporation of collaborative algorithms to improve personalized recommendations. The authors also detail conclusions drawn from functional testing and outline directions for ongoing research to further validate the system in real course settings.
![Social Knowledge and
TEL Recommender Systems
• Adolphs in [10] describes social knowledge as knowledge of the
11 A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences – Presented at EC-TEL 2014 – Graz,
Austria
minds of others.
• According to [1], social knowledge is the one about the larger
community of users other than the target user. Furthermore, social
knowledge enables the use of collaborative algorithms in which
predictions about individuals are extrapolated from their community
opinions [1].
Input data Goal Technique
Recommender
System
Using typical
data
(Demographic,
preferences,
behavior)
Using other
type of data
Using time
parameter that
may change
decisions
Taking into
account
competence
development
Taking into
account social
knowledge
Finding good
items
Helping to
improve
learning
performance
Knowledge
based
Collaborative
Filtering
[6] x x x
[4] x x x
[5] x x x x
[8] x x x x x
[9] x x x x](https://image.slidesharecdn.com/2014-09-19ec-telpresentation-140922152442-phpapp02/85/A-Recommender-System-for-Students-Based-on-Social-Knowledge-and-Assessment-Data-of-Competences-11-320.jpg)
![Background
ONTO-EQF CC-DESIGN RUBRICS-360 WEB SOLAR
[15]
Florian-Gaviria, B. ; Glahn, C. ; Fabregat Gesa, R.
A Software Suite for Efficient Use of the European Qualifications
Framework in Online and Blended Courses
Published in:
Learning Technologies, IEEE Transactions on (Volume:6 , Issue: 3 )
12 A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences – Presented at EC-TEL 2014 – Graz,
Austria](https://image.slidesharecdn.com/2014-09-19ec-telpresentation-140922152442-phpapp02/85/A-Recommender-System-for-Students-Based-on-Social-Knowledge-and-Assessment-Data-of-Competences-12-320.jpg)
![Adding successful experiences of students
Adding ratings of students’ experiences
Mobile SOLAR
Background
ONTO-QF CC-DESIGN RUBRICS-360 Web SOLAR
13 A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences – Presented at EC-TEL 2014 – Graz,
Austria
TEL
RecSys for Students
Calculating
personalized
recommendations
Adding other
qualification frameworks
[14]](https://image.slidesharecdn.com/2014-09-19ec-telpresentation-140922152442-phpapp02/85/A-Recommender-System-for-Students-Based-on-Social-Knowledge-and-Assessment-Data-of-Competences-13-320.jpg)
A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences
- 1. EC-TEL 2014 A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences Oscar Chavarriaga, Beatriz Florian-Gaviria, and Oswaldo Solarte EISC, University of Valle, Cali - Colombia {oscar.chavarriaga, beatriz.florian, oswaldo.solarte}@correounivalle.edu.co
- 2. The Authors o Oscar Chavarriaga – Computer Science Engineering pre-grade student, EISC, University of Valle, Colombia o Beatriz Florian-Gaviria – Assisstant Professor at EISC, University of Valle, Colombia – Ph.D. Information Technology, University of Girona, Spain o Oswaldo Solarte – Assisstant Professor at University of Valle, Colombia – M.Sc., University of Valle, Colombia A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences – Presented at EC-TEL 2014 – Graz, Austria 2
- 3. Our University 3 A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences – Presented at EC-TEL 2014 – Graz, Austria CALI BUENAVENTURA CARTAGO PALMIRA S. QUILICHAO
- 4. Overview o Motivation o Background o Recommender System Process o Prototype and Functional Testing o Conclusions o On-going Research A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences – Presented at EC-TEL 2014 – Graz, Austria 4
- 5. Motivation How the idea comes? 5 A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences – Presented at EC-TEL 2014 – Graz, Austria
- 6. Motivation Recommender Systems Strategies 6 A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences – Presented at EC-TEL 2014 – Graz, Austria
- 7. Motivation TEL Recommender Systems A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences – Presented at EC-TEL 2014 – Graz, Austria 7
- 8. Motivation Retrival good experiences Like-minded students not necessarily are students with similar learning performance. Like-minded students not necessarily can give/or receive good learning advice to/from In TEL RecSys is not always about what you like, sometimes it is about something that is good for your learning, even if you do not like it (or your like-minded peers do not like it). having similar opinions and interests their friends. 8 A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences – Presented at EC-TEL 2014 – Graz, Austria
- 9. Motivation Social Assessment 9 A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences – Presented at EC-TEL 2014 – Graz, Austria Data of Competences Social Knowledge Management of Good Learning Experiences
- 10. Background What are the pillars of this recommender system? 10 A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences – Presented at EC-TEL 2014 – Graz, Austria
- 11. Social Knowledge and TEL Recommender Systems • Adolphs in [10] describes social knowledge as knowledge of the 11 A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences – Presented at EC-TEL 2014 – Graz, Austria minds of others. • According to [1], social knowledge is the one about the larger community of users other than the target user. Furthermore, social knowledge enables the use of collaborative algorithms in which predictions about individuals are extrapolated from their community opinions [1]. Input data Goal Technique Recommender System Using typical data (Demographic, preferences, behavior) Using other type of data Using time parameter that may change decisions Taking into account competence development Taking into account social knowledge Finding good items Helping to improve learning performance Knowledge based Collaborative Filtering [6] x x x [4] x x x [5] x x x x [8] x x x x x [9] x x x x
- 12. Background ONTO-EQF CC-DESIGN RUBRICS-360 WEB SOLAR [15] Florian-Gaviria, B. ; Glahn, C. ; Fabregat Gesa, R. A Software Suite for Efficient Use of the European Qualifications Framework in Online and Blended Courses Published in: Learning Technologies, IEEE Transactions on (Volume:6 , Issue: 3 ) 12 A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences – Presented at EC-TEL 2014 – Graz, Austria
- 13. Adding successful experiences of students Adding ratings of students’ experiences Mobile SOLAR Background ONTO-QF CC-DESIGN RUBRICS-360 Web SOLAR 13 A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences – Presented at EC-TEL 2014 – Graz, Austria TEL RecSys for Students Calculating personalized recommendations Adding other qualification frameworks [14]
- 14. Recommender System Process The algorithm, step by step 14 A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences – Presented at EC-TEL 2014 – Graz, Austria
- 15. Recommender System Process 15 A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences – Presented at EC-TEL 2014 – Graz, Austria
- 16. Preprocessing Data • The system collects: – students' qualifications linked to the course in the past – up-to-date qualifications of a current student 16 A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences – Presented at EC-TEL 2014 – Graz, Austria
- 17. Filtering Former Students Candidates Selecting former students who achieved next planned competence level 17 A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences – Presented at EC-TEL 2014 – Graz, Austria FILTER
- 18. Calculating similarity Qualification coincidences in their learning curve give the similarity. 18 A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences – Presented at EC-TEL 2014 – Graz, Austria The Jaccard’s coefficient AA1 AA2 AA3 AA4 Similarity Current 1 2 2 2 Andy 1 2 2 2 1 Bob 1 2 2 3 ¾ Alice 1 2 3 3 ½ RESULT
- 19. Ranking and retrieving recommendation items Student Andy Andy Bob Bob Alice Alice Alice Item I1 I2 I1 I2 I1 I3 I4 Rate 4 3 4 4 5 2 3 19 A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences – Presented at EC-TEL 2014 – Graz, Austria I1 = (1*4)+(¾*4)+(½*5) = 7,5 I2 = (1*3)+(¾*4) = 4 I3 = (½*2) = 1 I4 = (½*3) = 1,5 1 2 3 4 AA1 AA2 AA3 AA4 Similarity Current 1 2 2 2 Andy 1 2 2 2 1 Bob 1 2 2 3 ¾ Alice 1 2 3 3 ½ RESULT
- 20. Demo and Functional Testing How the prototype looks? 20 A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences – Presented at EC-TEL 2014 – Graz, Austria
- 21. Prototype 21 A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences – Presented at EC-TEL 2014 – Graz, Austria
- 22. Funtional Testing 22 A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences – Presented at EC-TEL 2014 – Graz, Austria Cold Start -> synthetically data 3 test cases using the limit value technique
- 23. Conclusions Preliminar findings 23 A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences – Presented at EC-TEL 2014 – Graz, Austria
- 24. Conclusions • It is possible to build a TEL RecSys based on documented historic experiences and competence assessments of former students • Collaborative filtering techniques and knowledge-based techniques allow taking advantage of social knowledge and competence-based assessment for TEL RecSys • TEL RecSys could encourage social knowledge, due to fact that the participants of a course can contribute ideas that can be used by other students in the future • Because many competence qualifications frameworks follow the same pattern of definitions this approach may be used to implement new recommenders for courses using some of those competence frameworks • It is necessary to complement this research in real courses. 1) More Data, 2) To evaluate the impact on students 24 A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences – Presented at EC-TEL 2014 – Graz, Austria
- 25. On-Going Research What we are doing right now? What is next? 25 A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences – Presented at EC-TEL 2014 – Graz, Austria
- 26. On-going Research 26 A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences – Presented at EC-TEL 2014 – Graz, Austria
- 27. On-going Research 27 A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences – Presented at EC-TEL 2014 – Graz, Austria
- 28. Acknowledgements o University of Valle – Thank for funding the research project NUBE-UV (CI 2756). o Professor Martha Millán – Thank for her valuable consulting on the Knowledge Discovery in Databases Area o EISC, University of Valle – Thank for inscription, plane tickets, and accommodation to assist to EC-TEL 2014 A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences – Presented at EC-TEL 2014 – Graz, Austria 28
- 29. EC-TEL 2014 Thanks for your attention Questions? Oscar Chavarriaga, Beatriz Florian-Gaviria, and Oswaldo Solarte EISC, University of Valle, Cali - Colombia {oscar.chavarriaga, beatriz.florian, oswaldo.solarte}@correounivalle.edu.co A Recommender System for Students Based on Social Knowledge and Assessment Data of Competences – Presented at EC-TEL 2014 – Graz, Austria 29
Editor's Notes
- #2: Good afternoon. My name is Beatriz Florián-Gaviria. As second author, and on behalf of my co-authors, today I am going to present our paper “…..”. The three of us are affiliated to the University of Valle, Colombia. We are very honor to be here representing Colombia at this International Conference. We are grateful to have the opportunity to share our research with this research community.
- #3: The first author of this paper is an enthusiastic pre-grade student, Oscar Chavarriaga. Unfortunately, due to financial issues in funding students’ mobility abroad, Oscar is not here today. As I said earlier, I am the second author, and the third author is my college the professor Oswaldo Solarte.
- #4: Before getting down the presentation, I want to spend a half of a minute to tell you about our University and the authors. Colombia is a country in South America. The Valle del Cauca region is located in the south-west of Colombia. The University of Valle have 11 campuses along this region in different cities. Two of them in the main city, Cali, and 9 in other cities. This is a big state university in Colombia using blended education.
- #5: So, This is the presentation overview for today: First, I am going to present the motivation of this research. Second, I would like to set the research question Followed, I am going to exposed the pillars of this research. Then, the recommender system process will be explained, going step by step along with a short demo of the system and the functional testing results For the end, some conclusions and the on-going research.
- #7: Many RecSys take advantage of users’ logs, feedback and system personalization to produce recommendations for others. For instance: people who bought A-item also bought B-item, people who search X-item also search for Y-item etc.
- #8: So do TEL RecSys to Search for good items Predict satisfaction Generating a top list of items (books, papers, learning objects, learning material, etc.) and so on and so for…
- #9: However, good recommendations in learning could be far away from “like-minded students” data. “Like-minded” students not necessarely are students with similar learning performance. They could be close friends, but not with similar performance. “Like-minded” students not necessarely can give good advice in learning to their folks. maybe Because not necessarely are studing the same, or maybe because they are good at different subjects than their friends.
- #10: So, we are looking for students with similar learning curves. That is, former students similar in learning curves to a current student. We want to take into consideration successful experiences of this kind of similar former students to bring learning recommendations. For us, a successful experience is when a learners achieve a learning goal (competence level) by doing something that they can share with their community (it is important to said that he/she maybe did something that was not planned by the teacher). Thus, we need to build a system that support Social Knowledge Management of successful experiences in the learning path and on the other hand, a systems that manage a social assessment of students. This intellectual capital represents a valuable currency in the new Social Knowledge Learning
- #22: Video of 2 minutes to show theTEL RecSys funtionalities
- #23: Recommendation items and qualifications were generated synthetically because of the cold start problem. We run 3 funtional testing case scenarios Funtional testing checks for expected results after executing a piece of software. Thus, it assess the correctness of the recommender’s functionality Obtained results correspond to expected results, and the functional tests applied to recommender system were satisfactory
- #27: This is the Center for Innovation in Education that was build last year in Universidad del Valle in alliance with the Colombian Government. This is a center to teach, research, and development of educational content.
- #28: In this center we are conducting two massive blended courses (TIT@ and CREA-TIC) to teach Colombian teachers of primary and middle school to use ICTs in their learning designs. We are in the first steps to use the software suite and its recommender system in these courses, as to conduct research experiments in this context.
- #29: Thank Universidad del Valle for funding the research project NUBE‑UV (CI 2756). Thank professor Martha Millán for her valuable consulting on the Knowledge Discovery in Databases Area Thank EISC, University of Valle for inscription, plane tickets, and accommodation to assist to EC-TEL 2014
- #30: Thanks for your attention I am here to receive your questions and comments. Oscar Chavarriaga, Beatriz Florian-Gaviria, and Oswaldo Solarte EISC, University of Valle, Cali - Colombia {oscar.chavarriaga, beatriz.florian, oswaldo.solarte}@correounivalle.edu.co