Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
Measuring Quality Content
Search
Adam Hyland
August 04, 2012
Research
2
76
Measuring Quality Content
Presentation to Wikimania 2012 on Article Feedback Tool statistics.
Adam Hyland
August 04, 2012
Tweet
Share
More Decks by Adam Hyland
See All by Adam Hyland
Here Comes (a significant fraction of) Everybody
protonk
0
75
Boston Data Swap: Data Vis Under Uncertainty
protonk
0
53
Why Nate Silver is Famous
protonk
1
120
Data Visualization under Uncertainty
protonk
0
760
Phillips Academy Wikipedia Introduction
protonk
0
87
Other Decks in Research
See All in Research
データサイエンティストの採用に関するアンケート
datascientistsociety
PRO
0
970
数理最適化に基づく制御
mickey_kubo
5
660
SatCLIP: Global, General-Purpose Location Embeddings with Satellite Imagery
satai
3
210
定性データ、どう活かす? 〜定性データのための分析基盤、はじめました〜 / How to utilize qualitative data? ~We have launched an analysis platform for qualitative data~
kaminashi
6
1k
CSP: Self-Supervised Contrastive Spatial Pre-Training for Geospatial-Visual Representations
satai
3
200
Ad-DS Paper Circle #1
ykaneko1992
0
5.5k
Type Theory as a Formal Basis of Natural Language Semantics
daikimatsuoka
1
220
2025年度 生成AIの使い方/接し方
hkefka385
1
690
Weekly AI Agents News!
masatoto
33
68k
クラウドのテレメトリーシステム研究動向2025年
yuukit
3
950
NLP2025SharedTask翻訳部門
moriokataku
0
290
引力・斥力を制御可能なランダム部分集合の確率分布
wasyro
0
150
Featured
See All Featured
CoffeeScript is Beautiful & I Never Want to Write Plain JavaScript Again
sstephenson
161
15k
JavaScript: Past, Present, and Future - NDC Porto 2020
reverentgeek
48
5.4k
BBQ
matthewcrist
89
9.7k
A better future with KSS
kneath
239
17k
Writing Fast Ruby
sferik
628
61k
Building a Modern Day E-commerce SEO Strategy
aleyda
41
7.3k
Statistics for Hackers
jakevdp
799
220k
Navigating Team Friction
lara
187
15k
A Tale of Four Properties
chriscoyier
160
23k
Fight the Zombie Pattern Library - RWD Summit 2016
marcelosomers
233
17k
Optimising Largest Contentful Paint
csswizardry
37
3.3k
Rebuilding a faster, lazier Slack
samanthasiow
81
9k
Transcript
Measuring Article Quality Peer Review and the Article Feedback Tool
Adam Hyland protonk @ en-wp
Look Familiar?
Maybe This Version?
None
Article Feedback Tool • Deployed in 2010 • Version 4
(the current version) ramped up in 2011 • Designed to offer an avenue for reader feedback • High volume of reader feedback
• 6 months of public data • 795,353 articles --
2,487,522 responses
Featured Articles (FA) • 3,599 articles (0.09% of all articles)
• 2,267 Featured Lists (FL) • Most rigorous peer review process on the English Wikipedia • Very sensitive to editor preferences • Some idiosyncrasies
Good Articles (GA) • 15,357 articles • Relatively rigorous peer
review (yes I know reasonable minds may disagree) • Less idiosyncratic than FA in some ways • Perhaps less dependent on editor preference
Data • Article name • Length (in bytes) • GA/FA
status (including former/not- promoted) • Some user data
None
Beyond Summaries • Reader ratings follow pageviews • Predominantly non-editors
• Popular articles: • Call of Duty • Justin Bieber • Jimmy Wales (avg. rating: 1.10585)
Power Laws Everywhere!
Classical(ish) Models • Logistic regression model supports a relationship between
rating and likelihood of FA/GA • Linear model does, but with a twist • Can’t escape Cambridge Endogeneity Police!
None
Data Mining • Predicting featured status from reader ratings and
minimal meta-data. • Bayesian classifier able to roughly predict featured status (with a high false positive rate)
But the system’s changing! • AFT v4 is a multi-category
quantitative measure • AFT v5 is, roughly, YES/NO • Is this a problem? • Frank Harrell and the perils of dichotomization.
Actual Reader Ratings
Another Look
For the skeptics
Information • We can imagine we might not lose information
in shifting to v5 • This is born out by the classifier, to some degree. • We don’t lose a lot of power when dichotomizing individual ratings
A Look Ahead • Really exciting! • Great compliment to
current research methods • Long exposures can help discover reader/editor divergence • Predictive analytics • Need more open data
Questions? • Of course you have questions! • All work
is or soon will be available on github under a free license • Full writeup on en-wp forthcoming