CHI2014 Workshop - Leveraging Video Interaction Data and Content Analysis to Improve Video Learning
1 like1,581 views
This document discusses the analysis of video interactions in online learning environments, focusing on how learners engage with video lectures in MOOCs. It presents data-driven approaches to improve video learning experiences by identifying interaction patterns, optimizing video interfaces, and understanding learner behavior such as pausing and re-watching. The research highlights the importance of collective interaction data for enhancing educational video design and proposes new methods for creating adaptable video interfaces.
![Lab
Study:
edX
&
On-‐Campus
Students
“It’s
not
like
cold-‐watching.
It
feels
like
watching
with
other
students.”
“[interaction
data]
makes
it
seem
more
classroom-‐y,
as
in
you
can
compare
yourself
to
what
how
other
students
are
learning
and
what
they
need
to
repeat.”](https://image.slidesharecdn.com/chi2014-workshop-140501181357-phpapp02/85/CHI2014-Workshop-Leveraging-Video-Interaction-Data-and-Content-Analysis-to-Improve-Video-Learning-33-320.jpg)
CHI2014 Workshop - Leveraging Video Interaction Data and Content Analysis to Improve Video Learning
- 1. Juho Kim (MIT CSAIL) Shang-‐Wen (Daniel) Li (MIT CSAIL) Carrie J. Cai (MIT CSAIL) Krzysztof Z. Gajos (Harvard EECS) Robert C. Miller (MIT CSAIL) 2014.04.27 CHI 2014 | Workshop on Learning Innovation at Scale Leveraging Video Interaction and Content to Improve Video Learning
- 2. Video Lectures in MOOCs
- 3. Classrooms: rich, natural interaction data armgov on Flickr | CC by-‐nc-‐sa Maria Fleischmann / Worldbank on Flickr | CC by-‐nc-‐nd Love Krittaya | public domain unknown author | from pc4all.co.kr
- 4. liquidnight on Flickr | CC by-‐nc-‐sa
- 5. How do learners use videos? Data-‐Driven Approach: Analyze learners’ interaction with the video player
- 6. Why does data matter? • detailed understanding of video usage • design implications for – Instructors – Video editors – Platform designers • new video interfaces and formats Improved video learning experience
- 7. ~40M video interaction events from 4 edX courses Learners Videos Mean Video Length Processed Events 127,839 862 7:46 39.3M Courses: Computer science, Statistics, Chemistry
- 8. How do learners use videos? • Watch sequentially • Pause • Re-‐watch • Skip / Skim
- 9. Collective Interaction Traces video 'me Learner #1 Learner #2 Learner #3 Learner #4 . . . . . . Learner #7888 Learner #7887
- 10. Collective Interaction Traces into Interaction Patterns video 'me interac'on events second-‐by-‐second in-‐video activity
- 11. Data-‐Driven Analysis and Design for Educational Videos
- 12. 1. Analyze interaction patterns scalable and automatic methods to interpret interaction data 2. Improve video learning video interfaces that adapt to collective learner interaction patterns Research Directions: Data-‐Driven Analysis and Design for Educational Videos
- 13. 1. Analyze interaction patterns scalable and automatic methods to interpret interaction data 2. Improve video learning video interfaces that adapt to collective learner interaction patterns Research Directions: Data-‐Driven Analysis and Design for Educational Videos
- 14. Interaction Peaks Temporal peaks in the number of interaction events, where a significant number of learners show similar interaction patterns video 'me interac'on events Understanding In-‐Video Dropouts and Interaction Peaks in Online Lecture Videos. Juho Kim, Philip J. Guo, Daniel T. Seaton, Piotr Mitros, Krzysztof Z. Gajos, Robert C. Miller. Learning at Scale 2014.
- 15. What causes an interaction peak?
- 16. Video interaction log data Video content analysis – Visual content – Text from transcript – Speech & acoustic stream
- 17. Observation: Visual transitions in the video often coincide with a peak.
- 18. Type 1. Returning to content interaction
- 19. Type 2. Beginning of new material interaction
- 20. Text Analysis • Topic transitions & interaction peaks – Topic modeling • Linguistic patterns & interaction peaks – N-‐gram analysis
- 21. node, optimal, goal cost, equal, path expand mean topic transition likelihood over time
- 22. N-‐gram Analysis “<start-‐of-‐sentence> So” • initiating an explanation “So let me spend a second on that,” “So that means,” • arriving at the take-‐home message “So we can get lots of information just from these five number summaries.”
- 23. N-‐gram Analysis “this is” • visual explanation “this is the double bonds here on this oxygen” • naming of a particular concept “and this is called a dislocation”, “so this is sometimes called the first quartile”
- 24. Acoustic Analysis Speaking rate Pitch
- 25. Automatic, multi-‐channel, scalable peak detection and classification
- 26. Video Analytics: “debugging” interface for instructors & editors
- 27. 1. Analyze interaction patterns scalable and automatic methods to interpret interaction data 2. Improve video learning video interfaces that adapt to collective learner interaction patterns Research Directions: Data-‐Driven Analysis and Design for Educational Videos
- 28. Data-‐Driven Interaction Techniques to Support Video Navigation
- 29. For learners: Data-‐Driven Video UI
- 30. Rollercoaster Timeline • Embedded visualization of collective interactions • Visual & physical emphasis on interaction peaks
- 31. Automatic Summarization • Keyword summary with word cloud • Visual summary with captured highlights
- 32. Automatic Side-‐by-‐Side View Pinned slide Video stream
- 33. Lab Study: edX & On-‐Campus Students “It’s not like cold-‐watching. It feels like watching with other students.” “[interaction data] makes it seem more classroom-‐y, as in you can compare yourself to what how other students are learning and what they need to repeat.”
- 34. Rethinking Educational Videos
- 35. Are behind-‐the-‐encoding-‐wall videos the best format? • Hard to edit once published • Only a single stream is published • Lack of useful metadata (concepts, difficulty…) • Hard to comment on, point to specific parts
- 36. Toward More Direct & Social Interaction for Video Learning • Alternative explanations from learners • Synchronous watching with other learners • Linking relevant resources with different levels of scaffolding • Experimenting with in-‐video examples & data
- 37. Leveraging Video Interaction and Content to Improve Video Learning Juho Kim MIT CSAIL juhokim@mit.edu juhokim.com