veyepar: Using PyStruct to Classify MOOC Discussion Forum Posts

Author(s):	Kyle Shaffer
Location	Room CR6
Date	jul Sat 16	Days Raw Files
Start	14:15	First Raw Start	12:50
Duration	00:55:00	Offset	1:24:11
End	15:10	Last Raw End	14:20
Chapters	00:00 0:06:01 0:06:07 0:28:50 0:29:59 0:35:15 0:35:18 0:35:21 0:35:23 0:59:59 1:25:02
Total cuts_time	89 min.

https://pygotham.org/2015/talks/282/using-pystruct-to-classify-mooc-discussion-forum-posts
raw-playlist raw-mp4-playlist encoded-files-playlist mp4 svg png

assets release.pdf Using_PyStruct_to_Classify_MOOC_Discussion_Forum_Posts.json
logs
Admin: episode episode list cut list raw files day marks day marks day image_files

State:
Locked:	clear this to unlock
Locked by:	user/process that locked.
Start:	initially scheduled time from master, adjusted to match reality
Duration:	length in hh:mm:ss
Name:	Video Title (shows in video search results)
Emails:	email(s) of the presenter(s)
Released:	has someone authorised pubication
Normalise:
Channelcopy:	m=mono, 01=copy left to right, 10=right to left, 00=ignore.
Thumbnail:	filename.png
Description:	This talk will detail how I approached the task of classifying discussion forum posts in a massive open online course (MOOC). In particular, the talk will focus on using the PyStruct library as a tool to implement a conditional random field for this problem. Many linear classifiers have been shown to be effective at text classification, however discussion forums present a distinct sequential structure that can be leveraged. This makes this problem an excellent candidate for the application of structured prediction models. After presenting the study background, a brief overview of structured prediction will be presented focusing on the PyStruct library. In addition to presenting how PyStruct was used for this task, I will also present an overview with code examples for constructing features that were fed to the models and will detail any Python tools (such as scikit-learn and NLTK) used for these tasks. The results of the study show that framing this as a structured prediction task leads to substantial improvements in precision and F1 scores compared with a linear classifier. markdown
Comment:	Cancelled (I think) production notes

2016-07-16/12_50_49.ts Apply: 12:56:50 - 12:56:56 ( 00:00:06 ) S: 12:50:49 - E: 13:20:48 D: 00:29:59 (Start: 361.0) (End: 367.0) show more...

vlc ~/Videos/veyepar/big_apple_py/pygotham_2016/dv/Room_CR6/2016-07-16/12_50_49.ts :start-time=0361.0 --audio-desync=0

	Raw File	Cut List
	12:50:49		seconds: 361.0 Wall: 12:56:50
Duration	00:29:59
	13:20:48		seconds: 367.0 Wall: 12:56:56
Comments:
mp4 mp4.m3u	dv.m3u	Split: Sequence:: delete

2016-07-16/12_50_49.ts Apply: 12:56:56 - 13:19:39 ( 00:22:43 ) S: 12:50:49 - E: 13:20:48 D: 00:29:59 (Start: 367.0) (End: 1730.0) show more...

vlc ~/Videos/veyepar/big_apple_py/pygotham_2016/dv/Room_CR6/2016-07-16/12_50_49.ts :start-time=0367.0 --audio-desync=0

	Raw File	Cut List
	12:50:49		seconds: 367.0 Wall: 12:56:56
Duration	00:29:59
	13:20:48		seconds: 1730.0 Wall: 13:19:39
Comments:
mp4 mp4.m3u	dv.m3u	Split: Sequence:: delete

2016-07-16/13_20_49.ts Apply: 13:26:05 - 13:26:08 ( 00:00:03 ) S: 13:20:49 - E: 13:50:49 D: 00:30:00 (Start: 316.0) (End: 319.0) show more...

vlc ~/Videos/veyepar/big_apple_py/pygotham_2016/dv/Room_CR6/2016-07-16/13_20_49.ts :start-time=0316.0 --audio-desync=0

	Raw File	Cut List
	13:20:49		seconds: 316.0 Wall: 13:26:05
Duration	00:30:00
	13:50:49		seconds: 319.0 Wall: 13:26:08
Comments:
mp4 mp4.m3u	dv.m3u	Split: Sequence:: delete

2016-07-16/13_20_49.ts Apply: 13:26:08 - 13:26:11 ( 00:00:03 ) S: 13:20:49 - E: 13:50:49 D: 00:30:00 (Start: 319.0) (End: 322.0) show more...

vlc ~/Videos/veyepar/big_apple_py/pygotham_2016/dv/Room_CR6/2016-07-16/13_20_49.ts :start-time=0319.0 --audio-desync=0

	Raw File	Cut List
	13:20:49		seconds: 319.0 Wall: 13:26:08
Duration	00:30:00
	13:50:49		seconds: 322.0 Wall: 13:26:11
Comments:
mp4 mp4.m3u	dv.m3u	Split: Sequence:: delete

2016-07-16/13_20_49.ts Apply: 13:26:11 - 13:26:13 ( 00:00:02 ) S: 13:20:49 - E: 13:50:49 D: 00:30:00 (Start: 322.0) (End: 324.0) show more...

vlc ~/Videos/veyepar/big_apple_py/pygotham_2016/dv/Room_CR6/2016-07-16/13_20_49.ts :start-time=0322.0 --audio-desync=0

	Raw File	Cut List
	13:20:49		seconds: 322.0 Wall: 13:26:11
Duration	00:30:00
	13:50:49		seconds: 324.0 Wall: 13:26:13
Comments:
mp4 mp4.m3u	dv.m3u	Split: Sequence:: delete

Rf filename:	root is .../show/dv/location/, example: 2013-03-13/13:13:30.dv
Sequence:
get this:	check and save to add this