Occupy Video as Data: Visualizing Temporal Narratives

By / / dataset, process, tools and scripts / 3 comments

How do we make sense of specific events occurring during the Occupy movement through narratives emerging from social media over time?

When an NYPD office pepper sprays peaceful protestors, the event is immediately captured on camera phones with subsequent citizen responses on YouTube and Twitter and news broadcasts capturing conflicting narratives.

We are seeking to create a framework for analyzing video and social media feeds that help develop a rich timeline of events and inter-related media reporting and crowd-sourced responses.

Our team is exploring this approach using a single event like the Pepper Spraying of protestors by NYPD officer Tony Bologna or the eviction of protestors from Zucotti Park on November 15th, 2012 by examining news articles, YouTube videos (and related metadata), and Twitter feeds. The following key elements are being developed in parallel:

1. Manual search and curation of all relevant news articles and videos of the event composed on MediaThread

2. Extraction of metadata from all YouTube videos using google’s Gdata API and Phython programming.

3. Filtering and compilation of Twitter feeds from the day of the event using data from R-Shief showing whether the news originated on Twitter or migrated there from other sources (and which ones).

4. Visualization of all the inter-related events using timeline tools such as SIMILE

While mock-up examples are shown here, we have some working prototypes in development that we will share shortly.

Example Timeline Image from SIMILE Tool at MIT

About Nitin

Nitin Sawhney is a Professor of Practice in the Department of Computer Science at Aalto University in Helsinki, Finland. As a human-centered design researcher he examines the critical role of technology, civic media, and urban interventions in crisis and contested spaces. Nitin engages user experience design, participatory action research, sensory ethnography, and multi-modal (speech/audio) approaches for complex contexts of human-machine interaction. Nitin has recently been conducting research at the intersection of AI and HCI for real-time news as well as humanitarian and crisis situations. http://www.NitinSawhney.org

3 comments

  1. mymedianow
    Reply

    Twtter data strategy and implementation thus far: (24 March 2012)
    (update by Ulrich)

    What we (me and Christo) did was download the file of all tweets containing the hashtag #occupywallstreet from R-shif. We then filtered this huge file of tweets using mySQL as follows:
    -show only tweets containing the word zuccotti, and of these
    1) show the date and hour each tweet was made
    2) show the number of tweets per hour (and retweets per hour = tweets – 1)

    This can be seen in the doc in the dropbox called zuccottieviction_tweets_octnov.csv

    What we are showing and comparing using this data is:
    -how tweets increase and subside during the time of the zuccoti park eviction event
    -we can then compare this rise and fall with other media to see where the information originated
    -this data also allows us to see what type of tweet content created the most retweets and from which users did the most information activity (most retweets) originate. Again we can then compare this with the number of views, etc. on other media platforms so as to to see if twitter was an originator of information or a follower.

  2. Pingback: Hackathon 2 Roundup | Occupy Research

  3. Pingback: Occupy Video as Data presentation at NYU, 4/18 « #OccupyData NYC

Leave a comment