Where do images fit in the era of ‘Big Data’?

Where do images fit in the era of ‘Big Data’? - Analysing social media images from a Big and small data perspective.

Farida Vis, Information School

University of Sheffield

@flygirltwo

READING THE RIOTS

ON TWITTER

Rob Procter (University of Manchester)Farida Vis (University of Leicester)

Alexander Voss (University of St Andrews)[Funded by JISC] #readingtheriots

What role did social media play?

2.6 million riot tweets (donated by Twitter) –

700,000 individual accounts

Initially:o Role of Rumourso Did incitement take place? [no –#riotcleanup]o What is the role of different actors on Twitter?

Role of Rumours

Guardian Interactive Team (Alastair Dant)

http://www.guardian.co.uk/uk/interactive/2011/dec/07/london-riots-twitter

Data Journalism Award (sponsored by Google)

Data visualizations:

what are they and what do they want?

400 million tweets/day (March 2013)

40 million Instagram images/day (January 2013)

-> 59% posted to Twitter

-> 98% posted to Facebook

130 million active Instagram users

16 billion images on the service

1 billion likes a day (June 2013)

Where do images fit in the era of ‘Big Data’?

Big Data – text + number driven

Images: undervalued, underexplored

Not by the users

Big Data small data

Big Data small data

Quantitative Qualitative

Big Data small data

Quantitative Qualitative

Big Data small data

Quantitative QualitativeWHY?

Context

Context

Context

UK riots – 2.6 million tweets Egypt protest – 1FB image

READING THE RIOTS

ON TWITTER

Rob Procter (University of Manchester)Farida Vis (University of Leicester)

Alexander Voss (University of St Andrews)[Funded by JISC] #readingtheriots

No analysis of the images circulated on Twitter

The burning bus

Image sharing

2.6 million tweets: 10K unique shortened links - 19K shares

(1) image sharing platform

(2)video sharing platform

(3)social media platform

(4)mainstream media – riot coverage

(5)mainstream media – other

(6)Alternative media

(7)Blogs (included in Technorati top 100)

(8)Blogs – other

(9)Websites – news focused;

(10) Websites – other

(11) Spam

(12) Broken link

(1) Police car (burning, attack, and aftermath)(2) Bus (burning, aftermath, and altered image)(3) Other Vehicle (burning, attack, and aftermath)(4) Building (burning, aftermath, before/after shots)(5) Looting (in the act, aftermath, trophy shots)(6) Screenshots (TV screens)(7) Street scenes (8) Police(9) Arrests (10) Image of text (screen grab other than TV screen, sign,

newspaper front page)(11) riot clean up(12) unclear (13) Other(14) excluded (not about riots, not single still image, broken link,

image removed etc.).

Total image shares according to image categories

‘Although the Twitter user chose the viewing position and shared the image through Yfrog the original image data was created by one of Google’s ‘numerous data collection vehicles’ using their R5 ‘panoramic camera system’’ (Anguelov et al., 2010, pp. 32-33).

Re-use of images originally created by Google Streetview data collection vehicle

Deleted contenthttp://twitpic.com/62m6nx

The burning bus – 57 unique URLs

Image wall – ten categories

(1) The moment the bus went up in flames

(2) Smoking bus (close up of the bus, air full of thick smoke)

(3) Bus on fire (close up of the bus, engulfed in flames)

(4) Sky News – TV visible in shot

(5) Sky News – screengrab (TV not visible, better quality)

(6) Burning bus, police and crowds (high quality news image)

(7) Bus consumed by fire (poor quality focused screengrab)

(8)‘Call of Duty’ (same as smoking bus, but altered with text)

(9)Aftermath (carcass of the burnt bus in close up)

(10) Other

Viewing within the domestic space

Showing the first screen on the ‘second screen’

Journalists joining in

Images uploaded on Twitpic AND Flickr using different devices – different audiences/users

NB riots not Instagram-ed

Was not there, claimed to be

Altered images

Tottenham as a war zone

Role of popular culture

Journalist as witness

This reminds us of the way in which John Berger notes

the significance of the act of photography in terms of

the statement: ‘I have decided that seeing this is worth

recording’ (Berger, 1972, p. 179, emphasis in original).

Further work needed on how images were discussed

How to present/display the data?

Lev Manovich

Direct Visualization

Lev Manovich

http://www.manovich.net/

Aby Warburg mnemosyne

UK riots – 2.6 million tweets Egypt protest – 1FB image

Algorithmic visibility:

seeing the image through Edgerank

Finding and (re)tracing the image

Photographer of the images reproduced in Boing Boing

Photographer of the original image

Included in the Storyful guidelinesfor social media image verification

Who is the photographer?

Image altered?

Sequence of images around the specific image

See: http://blog.storyful.com/2012/04/24/inside-storyful-storyfuls-verification-process/#.Ucyb5-CPDap

And more

Verifying social media images

Where do images fit in the era of ‘Big Data’?

Big Data small data

Quantitative Qualitative

Big Data small data

Quantitative Qualitative

Big Data small data

Quantitative QualitativeWHY?

• Vis, F., Faulkner, S., Parry, K., Manyukhina, Y., & Evans, L. (in press), ‘Twitpic-ing the riots: analysing images shared on Twitter during the 2011 UK riots’, in Weller, K., Bruns, A., Burgess, J., Mahrt, M., and Puschmann, C. (Eds.) Twitter and Society, New York: Peter Lang.

• Vis, F. (in preparation), ‘From Egypt to Wallstreet: tracing a Facebook protest image across social media and the Internet’