=Paper=
{{Paper
|id=Vol-1688/paper-01
|storemode=property
|title=MoocRec.com : Massive Open Online Courses Recommender System
|pdfUrl=https://ceur-ws.org/Vol-1688/paper-01.pdf
|volume=Vol-1688
|authors=Panagiotis Symeonidis,Dimitrios Malakoudis
|dblpUrl=https://dblp.org/rec/conf/recsys/SymeonidisM16
}}
==MoocRec.com : Massive Open Online Courses Recommender System==
https://ceur-ws.org/Vol-1688/paper-01.pdf
MoocRec.com : Massive Open Online Courses
Recommender System
Panagiotis Symeonidis Dimitrios Malakoudis
Department of Informatics Department of Informatics
Aristotle University Aristotle University
Thessaloniki, 54124, Greece Thessaloniki, 54124, Greece
psymeon@gmail.com dmalakoudis@gmail.com
ABSTRACT
Massive open online courses (MOOCs) have recently gained
a huge users’ attention on the Web. They are considered
as a highly promising form of teaching from leading uni-
versities such as Stanford and Berkeley. MoocRec.com is
a web site that recommends courses to users so that, they
can acquire those skills, that are expected from their ideal
job posting. MoocRec’s recommendation engine is based on
Matrix Factorization (MF) model combined with Collabo-
rative Filtering (CF) algorithm, which exploits information
from external resources (i.e., users’ skills, courses’ character-
istics, etc.) to predict course trends and to perform rating
predictions according to them.
Categories and Subject Descriptors Figure 1: Users and Courses in the 2-D space.
H.3.3 [Information Search and Retrieval]: Information
Filtering
As shown in Figure 1, courses/users that are placed in
close distance, are the most suitable/similar to each other.
1. INTRODUCTION As shown, women prefer literature courses, whereas men
Massive Open Online Courses (MOOCs) platforms offer choose the technical ones. Specifically, the course “English
thousands of different courses and each course’s registra- Grammar and Style” can be recommended to Maria and
tion/enrolment can be in the hundreds of thousand students. Irene, whereas “From Java to Android” course is more suit-
It would be very useful, if someone could be recommended able to John. Please notice that matrix decomposition has
a course to acquire those skills, that are expected from his also revealed a second separation, which takes place among
ideal job description. people’s preference, towards practical and theoretical types
MoocRec.com is a web site that provides to users rec- of courses. In MoocRec.com, we predict users’ ratings over
ommendations of MOOCs. Firstly, users provide some in- courses based on matrix factorization (MF) technique, which
formation about their studies and their dream job. Then, exploits information from several external resources/matri-
MoocRec.com recommends to them related courses, to ac- ces.
quire the required skills for getting their dream job. The The rest of this paper is organized as follows. Section 2
heart of the recommendation engine of MoocRec.com is ma- summarizes the related work, Section 3 summarizes the sys-
trix decomposition over a user-course rating matrix R to re- tem’s architecture. Section 4 describes our recommendation
duce its dimensions and remove noise from data. To do this, engine. Finally, Section 5 concludes this paper.
we preserve a small number of k latent features (i.e., dimen-
sions) with the objective to reveal the mainstream users’
preferences. For example, in Figure 1, we plot users and
2. RELATED WORK
courses, assuming that k has been tuned to 2. Furnas et al. [1] proposed Singular Value Decomposition
(SVD) in Information Retrieval research field. More specif-
ically, SVD captures latent associations between the terms
and the documents. SVD is a well-known factorization tech-
nique that factors a matrix into three matrices. An instance
of SVD, known as UV-decomposition, searches for two ma-
trices (U and V ), whose their multiplication gives an approx-
imation of the original matrix R. A significant improvement
on the prediction accuracy of classic MF algorithm may be
obtained through the incorporation of implicit feedback into
the MF model [2, 3, 4].
Extensions of classic MF algorithm have been successfully example, course “Data Analysis for Life Sciences 6: High-
applied for recommendations in MOOC domain to address performance Computing for Reproducible Genomics” is rec-
the problem of high students’ drop-out rates from online ommended because it is related to Life Sciences and Ge-
courses. To reduce the high students’ drop-out rates, they nomics. Our recommendation algorithm, places first the
provide recommendations of useful forum threads to stu- courses which provide the greatest number of skills to the
dents based on their blog history inside a MOOC discussion target user. Moreover, courses that start soon or are self-
forum. For instance, Yang et al. [5] have designed a latent paced have priority towards others.
feature model to describe student behaviors inside a MOOC
forum.
3. SYSTEM’S ARCHITECTURE
Figure 2 introduces the architecture of the MoocRec sys-
tem. It consists of the web site, the recommendation engine,
the database and the web crawler.
Recommender Web Site
System User monitoring system
Rating system
Advanced matrix
factorization User profile based
recommender recommender
Content based Skill based Alerts
recommender recommender Search engine
Database Web Crawler
Users Alerts Searches MOOC Figure 3: Recommendation appearance and reason-
providers and social
Courses Watch lists
networks.
ing.
Skills User actions
Jobs Studies
5. CONCLUSIONS
In this paper, we proposed MoocRec.com, which exploits
Figure 2: MoocRec System Architecture information from external resources (i.e., users’ skills, courses’
characteristics, etc.) to provide course recommendations. In
future, we want to test experimentally our system to check
MoocRec.com integrates a search engine for MOOCs that
its accuracy effectiveness in terms of accurate recommenda-
are automatically retrieved using web content mining tech-
tions.
niques from MOOC providers such as edX and Coursera.
It also incorporates a MOOC recommender system, which
provides a target user with personalized content according References
to the skills he wishes to acquire. Therefore, we use content- [1] G. Furnas, S. Deerwester, and S. et al. Dumais. Informa-
based filtering in order to connect course’s description with tion retrieval using a singular value decomposition model
the desired skills. We also embed the matrix factorization of latent semantic structure. In Proc. ACM SIGIR Conf.,
technique and recommend MOOCs to a target user based pages 465–480, 1988.
on previous choices of other users.
Moreover, using alerts a registered user may be notified [2] Yehuda Koren. Factorization meets the neighborhood: a
of MOOCs when they become available. He may also rate multifaceted collaborative filtering model. In Proceeding
courses either explicitly or implicitly (log history). That is, of the 14th ACM SIGKDD international conference on
our monitoring system records the user’s actions and helps Knowledge discovery and data mining, pages 426–434.
in constructing his profile. ACM, 2008.
The web crawler scans edX and Coursera web sites and [3] Yehuda Koren and Robert M. Bell. Advances in Collabo-
enters all the MOOCs found in a MySQL database, In ad- rative Filtering. In Francesco Ricci, Lior Rokach, Bracha
dition, an additional web crawler mines from LinkedIn the Shapira, and Paul B. Kantor, editors, Recommender Sys-
skills which are connected to the top-rated jobs. Therefore, tems Handbook, pages 145–186. Springer, 2011.
the database holds information that correlates jobs to skills
and skills to courses, which makes course and job recom- [4] Arkadiusz Paterek. Improving regularized singular value
mendation possible. decomposition for collaborative filtering. In Proceedings
of KDD cup and workshop, volume 2007, pages 5–8,
2007.
4. RECOMMENDATION ENGINE
Figure 3 illustrates an example of the course recommen- [5] Diyi Yang, Mario Piergallini, Iris Howley, and Carolyn
dations which are provided by our system. An important Rose. Forum thread recommendation for massive open
characteristic of our recommendations is that the user is in- online courses. In Proceedings of 7th International Con-
formed about the reason he was recommended a course. For ference on Educational Data Mining, 2014.