The paper presents an ongoing development of an analytical tool for the clickstreams of video lectures. This work was motivated by the increasing number of video tutorials in graphical software applications. Although the clickstreams have been used to examine students’ detailed behaviors, few analytical tools have been designed from the instructors’ viewpoint. The tool we have developed includes four major interfaces: (1) accumulated behavior, (2) individual time history, (3) video control, and (4) a note-taking area. We also designed a global time controller, which users may move in all of the interfaces except the note-taking area, thus synchronizing the time stamps among the other interfaces. An actual CAD video lecture clickstream of 53 students and 103,679 clicks was used to validate the tool. We found that the analytical tool can help identify students' behaviors.
1. A Clickstream Analytical
Tool for Video Lectures
Yao-Yu Yang, Chia-Hsin Liu, Yi-Hsuan Lin, Shih-Chung Kang
Computer-Aided Engineering Div. of Civil Engineering
Dept.,
National Taiwan University
Yao-Yu (Ben) Yang, Ph.D student, NTU
16th October, 2016
web: yyben.tw
2. Students' Watching Patterns are Important
for instructors.
More than 60% MOOC instructors thought
students' watching patterns would help them:
(1)find struggling parts, most engaging content,
and trouble students
(2)improve topic presentations
(3)understand some assignment issues
2
Stephens-Martinez, K., et al. (2014). Monitoring moocs: which information sources do instructors value?
Proceedings of the first ACM conference on Learning@ scale conference, ACM.
3. Students' Watching Patterns are Important
for instructors.
More than 60% MOOC instructors thought
students' watching patterns would help them:
(1)find struggling parts, most engaging content,
and trouble students
(2)improve topic presentations
(3)understand some assignment issues
3
However, the instructors need analytical tools to identify
students’ learning process from the activity log.
5. • Seeking actions indicate that the students
were definitely watching at the video.
We focused on seek log
5
Seeking action
6. 6
We used the seek log which comes from one
of video lectures of Engineering Graphics.
Engineering Graphics Course - lecture topic: Dimension Style
In the video, 53 students applied seeking actions.
7. 7
The seek log is hard to read.
We applied 2 visual approaches to plot the seek log
in order to obtain the individual time history.
The seek log of the video
8. Visual approach I
Backward seeking
Time axis
Forward seeking
8
AB < A B<
A: Control Start Point
B: Control End Point
9. Visual approach I
Backward seeking
Time axis
Forward seeking
9
AB
A B
A: Control Start Point
B: Control End Point
10. Visual approach I
10
Act.3:
Forward seeking.
Act.4:
Backward seeking.
Act.1:Continuously seeking forward.
Act.2: Jumped to the beginning
An example of individual time history
11. Visual approach I
11
Animation makes the individual time history clear.
The animation of the individual time history for student ID 51.
12. Visual approach II
12
Total number of views
Total number of skips
The accumulated behavior chart shows
the total number of views and skips
for every second of the video.
14. Type I:Overview first
Type II:Smoothly watch
Type III: Skip and never watch a segment
Type IV: Concentrate on a segment
14
Four watching patterns recognized
15. What is the video content for those extreme
values on the accumulated behaviour chart?
15
Accumulated behavior chart Video content
16. What is the video content for those extreme
values on the accumulated behaviour chart?
16
Accumulated behavior chart Video content
17. 17
The lowest part was describing the topic of the video lecture at the beginning
The peak was talking about spacing of baselines in CAD,
which was a detailed operation.
What is the video content for those extreme
values on the accumulated behaviour chart?
18. What is the video content for those extreme
values on the accumulated behaviour chart?
18
Accumulated behavior chart Video content
19. 19
The highest was informing the numbers of a parameter setting
which had already shown on the video.
The lowest was introducing annotation lines in detail.
What is the video content for those extreme
values on the accumulated behaviour chart?
20. 20
•The two visual approaches we developed help
instructors understand students’ learning
process.
•We summarized video watching patterns into
four types, and found out the popular and
boring part of the video.
Conclusion
21. 21
•CAD instructors may evaluate their teaching
strategy and the effectiveness of the video
lectures.
•We plan to analyze all the video lectures of the
Engineering Graphics course for a bigger
picture.
Implication and future work