Jordi Sancho - Learning Analytics - LASI Bilbao 2015
- 1. Learning analytics to ignite massive collaborative projects Jordi Sancho Interactive Media Lab - LMI Social work department - Faculty of Education University of BarcelonaLearning Analytics Summer Institute, June 23th Deusto, Bilbao
- 2. Problem: ? Massive collaboration projects in real life but not in education (we can’t analyse and assess them) ? Student’s assignments are wasted efforts (motivation, learning and outcomes). Solution: To develop new analytic methodologies that scale the assessment and ignite massive collaborative learning projects. SUMMARY
- 4. BIG FAILURE
- 5. Redice 04 Project 10 subjects 10 teachers 900 students Develop and link the most important concepts (in your opinion) for each subject (wiki)
- 6. BIG FAILURE
- 8. We learnt…
- 9. What do we need to guide and assess an elephant stampede?
- 10. Macro level 1. Classify collaborators and contents according to their role on the process. 2. Classify created documents according to content. 3. Classify collaborators according to their type and role of contribution. 4. Classify documents according to their quality (teacher’s criteria). Individual level Assign and visualise all the above items for each student and for each document. To manage a massive collaboration project on content creation, we need to:
- 11. Accepted the challenge: PhD thesis in 2012
- 12. 1. Classify collaborators and contents according to their role and importance on the process (SNA)
- 13. Editor 1 Editor 2 Page 1 Page 2 Editor 3 Page 3
- 14. Page 1 Page 2 Page 3 Editor 1 1 0 1 Editor 2 1 1 0 Editor 3 0 0 1 Editor 1 Editor 2 Editor 3 Editor 1 0 1 1 Editor 2 1 0 0 Editor 3 1 0 0 Page 1 Page 2 Page 3 Page 1 1 0 1 Page 2 1 0 0 Page 3 1 0 0
- 18. 2. Classify created documents according to content (Factorial analysis)
- 21. 3. Classify collaborators according to their type and role of contribution. (k-means clustering)
- 23. Central group -Cluster 5- CORREC_CORRECTA INFO_RELLEV EXEMPLE_RELLEV EXEMPLE_IRRELL CORREC_INC ESBORRA_CORRECTAM ESBORRA_INCORR REESTRUCT_CORRECT REESTRUCTURA_INCORR COMBINA_ARG CITA_FONTS INFO_IRRELL CREA_ENLLAC FORMAT_WIKI 10137503,5.0 10220556,5.0 94176261,5.0 Correcció ortogrà?ca correcta Introdueix informació rellevant Esborra correctament Esborra incorrectament Reestructura correctament Reestructura incorrectament Cita fonts Introdueix informació irrellevant Crea enlla? (hipervincle) Format wiki Combina argumentacions Correcció ortogrà?ca incorrecta Exemple irrellevant Exemple rellevant Variable values Variables Editors Activity +-
- 26. 3. Classify documents according to their quality (teacher’s criteria). (Naive Bayes predictive algorithm)
- 27. Dichotomisation criteria Type 1 texts Type 2 texts Language analysis Language analysis Learning Algorithm Detects and identifies Patterns Probabilistic classification Text to classify Input From cluster analysis
- 28. May the algoritm/programme identify if contents have been developed following teacher’s directives? Text model Type 1 Text model Type 2
- 29. May the algoritm/programme identify if contents have been developed following teacher’s directives?
- 30. 75% Concepts developed following directives. 62% Concepts not developed following directives.
- 31. Wiki Metodologia Anàlisi factorial Algoritme predictiu naive Bayes Informació Continguts Wiki Tipologies d’editors Clústers per k-means Classificació continguts segons aportacions Blocs temàtics Importància editors Importància continguts Anàlisi de xarxes socials Editors Aportacions
- 33. Grau entre 1 i 9 Grau entre 21 i 54 Grau entre 10 i 20 Més de 32 edicions (percentil 95) 11-31 edicions Menys 11 edicions Tamany del node segons la mesura d’intermediació Individual information
- 34. Grau entre 1 i 9 Grau entre 21 i 54 Grau entre 10 i 20 Més de 32 edicions (percentil 95) 11-31 edicions Menys 11 edicions Tamany del node segons la mesura d’intermediació Grau entre 1 i 9 Grau entre 21 i 54 Grau entre 10 i 20 Més de 32 edicions (percentil 95) 11-31 edicions Menys 11 edicions Tamany del node segons la mesura d’intermediació 10273594 Degree centrality: 9 Intermediation: 4th (2,59%) C2: Informers and more Teoria econòmica (5th editor) Social (10th editor) Individual information
- 35. 10137503 10187391 10220556 10222461 10235050 10259885 10273594 10528372 10854480 10858746 10859855 10861701 10876445 10878431 10878442 10882900 10883084 10889830 10897331 10914061 10916334 10916916 11002040 11003521 11006376 11011405 11017963 11019072 11165512 11763334 11763850 11771561 94176261 97176752 98072306 98205505 Clúster 4 Clúster 3 Clúster 5 Clúster 2 C 2: Molta informació i altres C 3: Editors de format C 4: Vinculadors C 5: Grup motor General information
- 36. Factor 1 Factor 2 Factor 3 Factor 4 Factor 5 Factor 6 Factor 7 Factor 8 Factor 9 Factor 10 Més de 40 edicions 30- 40 edicions 20-30 edicions Menys de 20 edicions Tamany del node segons el grau de centralitat General information
- 37. Macro level 1. Classify collaborators and contents according to their role on the process. 2. Classify created documents according to content. 3. Classify collaborators according to their type and role of contribution. 4. Classify documents according to their quality (teacher’s criteria). Individual level Assign and visualize all the above items for each student and for each document. To manage a massive collaboration project on content creation, we need to:
- 38. CONCLUSIONS
- 39. Starting with the usual teaching process and “poaching” data that makes sense. 1. WHERE IS YOUR STANDING? Designing learning processes with the analytics in mind. vs. We can think big. We can rethink learning. We should ignite massive collaborating projects (doing the same but just marginally better is not enough)
- 40. 18 05 13 87 2. DRAWING NEW MAPS HAS PROBLEMS
- 43. “ Unknown Parts”
- 44. Sometimes the addition of an improved methodology produces a loss of useful knowledge. In the short time. Are we still there?
- 45. Sancho, J. (2012) Muerte y resurrección de la universidad (again) en manos de la colaboración masiva: avanzar los MOOC. Dins de Bergmann, Juliana i Grané, Mariona (Editores) La universidad en la nube. Barcelona: Transmedia XXI. Sancho, J. (2011) La evaluación de proyectos colaborativos a gran escala basados en wikis mediante el análisis de redes sociales. Dins de Cano, E. (Editora)(2011) Aprobar o aprender. Estrategias de evaluación en la sociedad red. Barcelona: Transmedia XXI.
- 46. Thank you! Jordi Sancho jsancho@ub.edu Twitter: @JordiSanchoS