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.

Doctoral student discussion forum on MOOCs

351 views

Published on

Keynote talk given in Doctoral student discussion forum on MOOCs, held in Beijing, China.

Published in: Science
  • Be the first to comment

  • Be the first to like this

Doctoral student discussion forum on MOOCs

  1. 1. Understanding and Engaging MOOC Learners: A Data-driven Approach Guanliang Chen Web Information Systems, TU Delft https://angusglchen.github.io/
  2. 2. Background
  3. 3. Background Learning data
  4. 4. Background KnowledgeLearning data
  5. 5. Background Knowledge Application to learning Learning data
  6. 6. Background Knowledge Application to learning Learning data
  7. 7. Background Knowledge Application to learning Learning data Social Web data
  8. 8. Background Knowledge Application to learning Learning data Social Web data +
  9. 9. Background Knowledge Application to learning Learning data Social Web data + How can the Social Web data be utilised to better understand and engage our MOOC learners?
  10. 10. Learner Profiling Beyond the MOOC Platform ACM WebScience 2016 & ACM Learning at Scale 2016 (Best Paper Nominee) Guanliang Chen, Dan Davis, Jun Lin, Claudia Hauff, and Geert-Jan Houben. Beyond the MOOC platform: Gaining Insights about Learners from the Social Web, ACM WebScience, pp. 15-24, 2016. Guanliang Chen, Dan Davis, Claudia Hauff and Geert-Jan Houben, Learning Transfer: does it take place in MOOCs?, ACM Learning At Scale, pp. 409-418, 2016.
  11. 11. Whythis research? Learner Engagement, retention, … During MOOC
  12. 12. Whythis research? Learner Before MOOC NOTHING Engagement, retention, … During MOOC
  13. 13. Whythis research? Learner Before MOOC NOTHING Engagement, retention, … During MOOC NOTHING After MOOC
  14. 14. Howto solve the problem? We propose: a deeper understanding about learners can be gained by exploring their traces in the Social Web.
  15. 15. Whatresearch questions?
  16. 16. Whatresearch questions? 1 On what Social Web platforms can a significant fraction of MOOC learners be identified? 

  17. 17. Whatresearch questions? 1 On what Social Web platforms can a significant fraction of MOOC learners be identified? 
 Are learners who demonstrate specific sets of traits on the Social Web drawn to certain types of MOOCs? 
2
  18. 18. Whatresearch questions? 1 On what Social Web platforms can a significant fraction of MOOC learners be identified? 
 Are learners who demonstrate specific sets of traits on the Social Web drawn to certain types of MOOCs? 
2 To what extent do Social Web platforms enable us to observe (specific) user attributes that are highly relevant to the online learning experience? 
 3
  19. 19. Learner Identification across Social Web platforms
  20. 20. Learner Identification across Social Web platforms Email Login name Full name
  21. 21. Learner Identification across Social Web platforms 1) Explicit discovery via emails Email Login name Full name
  22. 22. Learner Identification across Social Web platforms 1) Explicit discovery via emails Email Login name Full name Profile pictures Profile links
  23. 23. Learner Identification across Social Web platforms 1) Explicit discovery via emails 2) Search Email Login name Full name Profile pictures Profile links
  24. 24. Learner Identification across Social Web platforms 1) Explicit discovery via emails 2) Search Email Login name Full name Profile pictures Profile links Compare: 1. Profile link 2. Profile pictures 3. Login & Full names
  25. 25. Learner Identification across Social Web platforms 1) Explicit discovery via emails 2) Search Email Login name Full name Profile pictures Profile links Compare: 1. Profile link 2. Profile pictures 3. Login & Full names MATCH !
  26. 26. Social Web platforms involved in our work
  27. 27. Matching Results for 18 DelftX MOOCs
  28. 28. Matching Results for 18 DelftX MOOCs Lowest Highest Overall Gravatar 4,37% 23,49% 7,81% Twitter 4,99% 17,58% 7,78% Linkedin 3,90% 11,05% 5,89% StackExchange 1,23% 21,91% 4,58% GitHub 3,43% 41,93% 10,92%
  29. 29. Matching Results for 18 DelftX MOOCs Lowest Highest Overall Gravatar 4,37% 23,49% 7,81% Twitter 4,99% 17,58% 7,78% Linkedin 3,90% 11,05% 5,89% StackExchange 1,23% 21,91% 4,58% GitHub 3,43% 41,93% 10,92% On average, 5% of learners can be identified on globally popular Social Web platforms. 

  30. 30. Learners on Linkedin - Using job titles & skills to characterise learners
  31. 31. Learners on Linkedin - Using job titles & skills to characterise learners Data Analysis MOOC - Software Engineer - Business Analyst - …
  32. 32. Learners on Linkedin - Using job titles & skills to characterise learners Data Analysis MOOC - Software Engineer - Business Analyst - … Design Approach MOOC - Co founder - UX designer - …
  33. 33. Learners on Linkedin - Using job titles & skills to characterise learners - Visualised by applying t-SNE techniques.
  34. 34. Learners on Twitter - To predict learners’ demographics (e.g., age & gender)
  35. 35. Learners on Twitter - To gather social relation information Framing MOOC Functional Programming
  36. 36. Learners on Twitter WITH friends WITHOUT friends # Learners 637 1292 Completion rate 28,57% 23,99% Avg. time watching videos 116.61 min 119.5 min Avg. # questions attempted 117.7 102.9 Avg. #posts per learner 0.96 0.76
  37. 37. Learners on Twitter WITH friends WITHOUT friends # Learners 637 1292 Completion rate 28,57% 23,99% Avg. time watching videos 116.61 min 119.5 min Avg. # questions attempted 117.7 102.9 Avg. #posts per learner 0.96 0.76 MORE ENGAGING
  38. 38. Learners on GitHub - To what extent do learners transfer their acquired knowledge into practice? - Learning transfer is the application of knowledge or skills gained in a learning environment to another context. - A more important measure of learning in MOOCs than retention, engagement, or completion rate.
  39. 39. Learners on GitHub 3 months 2.5 years + 0.5 years + +FP101x logs Surveys Coding data
  40. 40. Learners on GitHub 3 months 2.5 years + 0.5 years + +FP101x logs Surveys Coding data Are changes made in a functional language?
  41. 41. Learners on GitHub- Are “Github learners” different? GitHub learners Non-GitHub learners # Learners 12,415 25,070 Completion rate 7.71% 4.03% Avg. time watching videos 49.1 min 27.7 min Avg. # questions attempted 31.3 17.5 Avg. accuracy of learners’ answers 23.4% 12.9%
  42. 42. Learners on GitHub- Are “Github learners” different? GitHub learners Non-GitHub learners # Learners 12,415 25,070 Completion rate 7.71% 4.03% Avg. time watching videos 49.1 min 27.7 min Avg. # questions attempted 31.3 17.5 Avg. accuracy of learners’ answers 23.4% 12.9% MORE ENGAGING
  43. 43. Learners on GitHub- Are “Expert learners” different? Expert GitHub learners Novice GitHub learners # Learners 1,721 10,694 Completion rate 15.0% 6.5% Avg. time watching videos 78.6 min 44.4 min Avg. # questions attempted 57.9 27.0 Avg. accuracy of learners’ answers 38.0% 21.1%
  44. 44. Learners on GitHub- Are “Expert learners” different? Expert GitHub learners Novice GitHub learners # Learners 1,721 10,694 Completion rate 15.0% 6.5% Avg. time watching videos 78.6 min 44.4 min Avg. # questions attempted 57.9 27.0 Avg. accuracy of learners’ answers 38.0% 21.1% MORE ENGAGING
  45. 45. To what extent do engaged learners exhibit learning transfer? >30%10-30%<5% 5-10%
  46. 46. Which type of learner is more likely to display learning transfer? Intrinsically motivated Extrinsically motivated
  47. 47. Which type of learner is more likely to display learning transfer? Intrinsically motivated
  48. 48. Experienced Inexperienced Which type of learner is more likely to display learning transfer?
  49. 49. Experienced Which type of learner is more likely to display learning transfer?
  50. 50. Take-home Messages On average, 5% of learners from 18 DelftX MOOCs can be identified on 5 globally popular Social Web platforms. 
1
  51. 51. Take-home Messages On average, 5% of learners from 18 DelftX MOOCs can be identified on 5 globally popular Social Web platforms. 
1 Learners with specific traits prefer different types of MOOCs.2
  52. 52. Take-home Messages On average, 5% of learners from 18 DelftX MOOCs can be identified on 5 globally popular Social Web platforms. 
1 Learners with specific traits prefer different types of MOOCs.2 Learners’ post-course behaviour can be investigated by using their external Social Web traces.3
  53. 53. From Learners to Earners: Enabling MOOC Learners to Apply their Skills and Earn Money in an Online Market Place IEEE Transactions on Learning Technologies Guanliang Chen, Dan Davis, Markus Krause, Efthimia Aivaloglou, Claudia Hauff and Geert-Jan Houben. Can Learners be Earners? Investigating a Design to Enable MOOC Learners to Apply their Skills and Earn Money in an Online Market Place, IEEE Transactions on Learning Technologies.
  54. 54. Whatis the problem? EX101x: Data Analysis to the MAX()
  55. 55. Whatis the problem? EX101x: Data Analysis to the MAX() Most successful learners are already highly educated. Learners from developing countries 
 are underrepresented.
  56. 56. Whatis the problem? EX101x: Data Analysis to the MAX() Most successful learners are already highly educated. Learners from developing countries 
 are underrepresented. But, MOOCs aim to educate the world!

  57. 57. Howto solve the problem? We propose: learners can be paid to take MOOCs.
  58. 58. Howto solve the problem?
  59. 59. Howto solve the problem? Learners MOOCs take
  60. 60. Howto solve the problem? Learners MOOCs take Freelance Platforms connect
  61. 61. Howto solve the problem? Learners MOOCs take Freelance Platforms connect solve
  62. 62. Howto solve the problem? Learners MOOCs take Freelance Platforms connect solve pay
  63. 63. Setup 1 2 3 Weekly spreadsheet “bonus exercises” drawn from UpWork (manually checked) in EX101x. Accuracy check. Quality check (code smells).
  64. 64. Howare learners doing?
  65. 65. Howare learners doing? Learners can solve real-world tasks in good quality.
  66. 66. Howare learners doing? Learners can solve real-world tasks in good quality. Real-world tasks improve learners’ engagement.
  67. 67. Recommender System We have built a working recommender and deployed in a MOOC for experiment.
  68. 68. Overall Knowledge Application to learning Learning data Social Web data +
  69. 69. Overall Knowledge Application to learning Learning data Social Web data +
  70. 70. Overall Knowledge Application to learning Learning data Social Web data + To better understand our learners & To better engage our learners.
  71. 71. Overall Knowledge Application to learning Learning data Social Web data + To better understand our learners & To better engage our learners. Many learners/courses, plenty of data, lots of potential unexplored.
  72. 72. Thank you for your participation! http://bit.ly/lambda-lab

×