1. Towards learning resources rankings in
MOOCs: A Pairwise based Reputation
Mechanism
R. Centeno, M. Rodríguez-Artacho, Félix García, Elio
Sancristóbal, Gabriel Díaz, Manuel Castro
miguel@lsi.uned.es
UNED University, Spain
2. Content
• Reputation
• Integrating remote laboratories in MOOCs
• Evaluating MOOC content
• A Pairwise reputation mechanism for MOOCs
• Conclusions
IEEE EDUCON 2015, Tallinn
3. Reputation
• many types: professional links, friendships, purchases, ...
• complex: dynamism, complexity of the social structure, many
nodes (users, entities, ..)
• interaction results are unpredictable (which seller to select,
which hotel to book, …)
• how can we predict future behaviours?
3
Complex Social Networks Reputation
IEEE EDUCON 2015, Tallinn
4. Reputation mechanisms in social
networks
IEEE EDUCON 2015, Tallinn
objective: extract reputation of entities (users, objects, …)
how: gathering and aggregating opinions
Examples:
5. Reputation mechanisms in social
networks
IEEE EDUCON 2015, Tallinn
objective: extract reputation of entities (users, objects, …)
how: gathering and aggregating opinions
Examples:
6. Reputation mechanisms in social
networks
Qualitative vs.Quantitative “Experts” vs.
Users
Unbalanced source text oppinion vs.
Numerical rating
Lack of accuracy numerical ratings
IEEE EDUCON 2015, Tallinn
7. MOOC Example DIEEC UNED
Integrating remote laboratories in MOOCs
◌ Module 1: Simulator.
◌ Module 2: VISIR.
◌ Module 3: Working with resistors. Ohmic values. Voltage divider.
◌ Module 4: RLC circuits. RL, RLC & RC circuits.
◌ Module 5: Working with diodes. Differences between 1N4007 & BAT42.
Halfwave rectifier. Voltage drop on diode.
◌ Module 6: Low-pass filter. Mean value, voltage ripple, load regulation
and line regulation.
◌ Module 7: Zener diode. Zener diode as voltage regulator. Zener diode as
clipper. Construction of the current-voltage characteristic curve.
◌ Module 8: Operational amplifier. Non-inverting amplifier. Inverting
differentiator. Inverting amplifier
IEEE EDUCON 2015, Tallinn
9. MOOC Example DIEEC UNED
Organization
Access to experiments is provided by the
MOOC’s portal through an integrated
scheduling/booking system
The initial settings allow 16 simultaneous
users per 60 minutes slot and for each
user a maximum of two simultaneous
slots booked and a limitation of 14 slots
per course
With these settings, VISIR allows up to 384
students to experiment with any of the
designed practices of the MOOC
11. Cuantitative vs. Comparative reviews
Easier for users to state opinions when the
query compare objects in a pairwise fashion
Ben-Hur 9
Casablanca 7.4
Gone with the wind 8.2
IEEE EDUCON 2015, Tallinn
12. Cuantitative vs. Comparative reviews
• Potential problems of numerical oppinion
– Passive wait for users to grade
– Influenciable and manipulable viral ratings
• Reputation based on comparative reviews
Pairwise reputation mechanism (PWRM)
IEEE EDUCON 2015, Tallinn
13. MOOC formalization
M = {U,R,LR,LU} is a MOOC where:
U = {u1,..,un} users (students AND teachers)
R = {r1,..,rj} learning resources
LR = { <ui,rj> / ui € U; rj € R } user ui has uploaded
resource rj
LU = { <uk,rm> / uk € U; rm € R } user uk has used
resource rm
IEEE EDUCON 2015, Tallinn
14. MOOC formalization
• Live envorinment in term of resources
• Objective build ranking of resources within
the MOOC
• Asumption: Users have set of resources’
preferences on a subset
• Let O set of oppinions where oi=<ri,rj>
representing a pairwise query
R' Í R
IEEE EDUCON 2015, Tallinn
16. Conclusions
• Reputation mechanism to be applied within MOOCs
• Massive implies accuracy, but resources are often not
provided by many users
• Clustering criteria Selection process
According to typology (Multimedia, video, audio, …)
According to metadata (Pedagogical objectives,
granularity, etc. )
IEEE EDUCON 2015, Tallinn
17. Towards learning resources rankings in
MOOCs: A Pairwise based Reputation
Mechanism
R. Centeno, M. Rodríguez-Artacho, Félix García, Elio
Sancristóbal, Gabriel Díaz, Manuel Castro
miguel@lsi.uned.es
UNED University, Spain
Thanks!
Notes de l'éditeur
Oppinions and social networks
How accurate users provide ratings predict behaviours and confidence of results.