Resurrecting My Revolutionsing Social Link Neighborhood in Bringing Context to the Disappearing Web

Resurrecting My Revolution
Using Social Link Neighborhood in Bringing Context
to the Disappearing Web
Hany SalahEldeen & Michael Nelson Resurrecting My Revolution TPDL 2013
Hany M. SalahEldeen & Michael L. Nelson
Old Dominion University
Department of Computer Science
Web Science and Digital Libraries Lab.
TPDL 2013

• From Twitter, Websites, Books:
• The Egyptian revolution.
• From Twitter Only:
• Stanford’s SNAP dataset:
• Iranian elections.
• H1N1 virus outbreak.
• Michael Jackson’s death.
• Obama’s Nobel Peace Prize.
• Twitter API:
• The Syrian uprising.
Six Socially Significant Events
Hany SalahEldeen & Michael Nelson 02 Resurrecting My Revolution TPDL 2013

Social Events Having a Bimodal Time Distribution

Resources Missing & Archived
Collection Percentage Missing Percentage Archived
23.49%H1N1 Outbreak 41.65%
36.24%Michael Jackson 39.45%
26.98%Iran 43.08%
24.59%Obama 47.87%
10.48%Egypt 20.18%
7.04%Syria 5.35%
31.62% 30.78%
24.47% 36.26%
25.64% 43.87%
26.15% 46.15%

Missing and Archived Percentages Across Time

Previous Conclusions
• Measured 21,625 resources from 6 data sets in
archives & live web.
• After a year from publishing about 11% of
content shared on social media will be gone.
• After this we are losing roughly 0.02% daily.

New Research Questions
• Validity: is our estimation model still valid?
• Existence Stability: do resources on the live web
remain missing?
• Archival Stability: do resources in public archives
persist?
• Social Context: how can we extract social context of
missing resources and potential replacements?

Revisiting Existence
• From previous study:
• Rerunning after a year:
MJ Iran H1N1 Obama Egypt Syria
Measured 37.10% 37.50% 28.17% 30.56% 26.29% 31.62% 32.47% 24.64% 7.55% 12.68%
Predicted 31.72% 31.42% 31.96% 30.98% 30.16% 29.68% 29.60% 28.36% 19.80% 11.54%
Error 5.38% 6.08% 3.79% 0.42% 3.87% 1.94% 2.87% 3.72% 12.25% 1.14%
Average Prediction Error = 4.15%

Revisiting Archival
• From previous study:
• Rerunning after a year:
MJ Iran H1N1 Obama Egypt Syria
Measured 48.61% 40.32% 60.80% 55.04% 47.97% 52.14% 48.38% 40.58% 23.73% 0.56%
Predicted 61.78% 61.18% 62.26% 60.30% 58.66% 57.70% 57.54% 55.06% 37.94% 21.42%
Error 13.17% 20.86% 1.46% 5.26% 10.69% 5.56% 9.16% 14.48% 14.21% 20.86%
Average Prediction Error = 11.57%
in all cases, our archival predictions were too optimistic

Measured Vs. Predicted

Interesting Phenomenon: Reappearance
On The Live Web And Disappearance From
The Archives
Event MJ Iran Obama H1N1 Egypt Syria Average
% Re-appearing on the web 11.29% 11.48% 6.63% 3.68% 4.21% 1.97% 6.54%
% Disappearing from archives 9.98% 11.17% 15.65% 5.46% 2.81% 2.25% 7.89%
% Going from 1 memento to 0 2.72% 2.89% 4.24% 1.96% 0.23% 0.28% 2.05%

Reappearing And Disappearance
Predictions

Tweet Existence
Problem:
We don’t have the URIs of most of the tweets
Solution:
compute loss of other tweets linking to the URI:
for each resource in the datasets:
-Extract all the tweets that link to the resource using Topsy
API (up to 500 tweets)
-Check existence of each tweet.
- (Topsy (mostly) does not delete indexed tweets)

Using Topsy API
Get all the
tweets having
the same URI
Check how many
still exist on the
live web

Tweet Existence
Event MJ Iran Obama H1N1 Egypt Syria Average
Average % of missing posts 14.43% 14.59% 10.03% 7.38% 15.08% 0.53% 10.34%

Tweets Disappearing Across Time

Context Discovery And Shared
Resource Replacement
Problem:
140 characters limits the description of the linked
resource. If it went missing, can we get the next best
thing?
Solution:
• Shared links typically have several
tweets, responses, and retweets
• We can mine these traces for context and viable
replacements

Context Discovery
Linking to:
http://beta.18daysinegypt.com/

Use Topsy to Discover Tweets
with the Same Link

Tweet Text Replacement
• From all extracted tweets, extract the best replacement tweet
having the longest common N-gram
Replace with
more descriptive
one

Resource Replacement
Assume that the resource linked in the tweet
disappeared.
We mine the list of tweets for:
• Hashtags
• User mentions
• Co-tweeted URIs

Co-Tweeted Resources
A missing resource could be described or replaced by another
resource that have been shared within the same tweet.
replaces

Co-Tweeted Resources
A missing resource could be described or replaced by another
resource that have been shared within the same tweet.
replacesOr

Build a Tweet Document
A tweet document represents the concatenation of all
extracted tweets:
do you have a story to tell about your 18 days of
revolution? share it or contact sara 18days brand new
interactive storytelling project on egyptian revolution a
very creative platform to tell your story daysinegypt
marches heading to tahrir square now from all over cairo
it's all over again use the website to document your
revolutionary stories and share them with the world! check
out awesome documentary project crowdsourcing a
people's narrative of the egyptian revolution …
”
“

Tweet Signatures
Tweet Document:
Tweet Signature = top 5 most frequent terms from Tweet Document
documentary project daysinegypt check sourced
”
“do you have a story to tell about your 18 days of
revolution? share it or contact sara 18days brand new
interactive storytelling project on egyptian revolution a
very creative platform to tell your story daysinegypt
marches heading to tahrir square now from all over cairo
it's all over again use the website to document your
revolutionary stories and share them with the world! check
out awesome documentary project crowdsourcing a
people's narrative of the egyptian revolution …

Query Google w/ Tweet Signature

Search Engine Results
The original
resource

Search Engine Results
The original
resource
Others are
good replacement
candidates

Recommendation Evaluation
We extract a dataset of resources that are currently
available:
• Pretend these resources no longer exist (for a baseline)
• Each of the resources are textual based
• Each resource have at least 30 retrievable tweets.
 Extracted 731 unique resources

Recommendation Evaluation
We use boiler plate removal library* to remove the
template from the:
•linked resources
•top 10 retrieved results from Google
We use cosine similarity to compare the documents
* https://github.com/misja/python-boilerpipe

Similarity Measures In Resource
Replacements

Results
41% of the test cases we can find a
replacement page with at least 70% similarity
to the original missing resource
The search results provide a mean reciprocal
rank of 0.43

Conclusions
• We validated our model in predicting the resource existence
on the current web with ~4% error after one year.
• The archival prediction on the other hand produced a large
error ~11.5%
• We explored a phenomenon of reappearing on the web after
disappearing (6.54%), and disappearing from the archives
too (7.89%).
• The removal of search engine caches in the most recent
Memento revision could be a possible explanation of the
disappearance from the archives.

Cont. Conclusions
• Measured the estimated percentage missing from
the tweets ~10.5%.
• Utilized Topsy API to extract context information
about tweets with missing resources.
• Investigated the viability of finding a replacement
resource to the missing one.
• In 41% of the cases we were able to extract a
replacement resource that is 70% or more similar to
the missing resource.

Resurrecting My Revolutionsing Social Link Neighborhood in Bringing Context to the Disappearing Web

More Related Content

Viewers also liked (20)

More from Michael Nelson (9)

Recently uploaded (20)

Resurrecting My Revolutionsing Social Link Neighborhood in Bringing Context to the Disappearing Web