Picturing the Social: Talk for Transforming Digital Methods Winter School
Download as PPT, PDF2 likes4,798 views
The document discusses a presentation by Farida Vis on analyzing social media images at the Transforming Digital Methods Winter School, highlighting the massive daily sharing of images across platforms. It outlines the research project 'Picturing the Social' which aims to explore social media image data through interdisciplinary methods and develop tools for academic research. Key research questions focus on understanding the technical infrastructures and user experiences of social media image sharing, while emphasizing the importance of context in image meaning and visibility.
Picturing the Social: Talk for Transforming Digital Methods Winter School
- 1. Picturing the Social: Analysing Social Media Images A talk for the Transforming Digital Methods Winter School at the University of Exeter, 11 December 2014 Farida Vis, Information School University of Sheffield @flygirltwo
- 2. 750 MILLION - 1.8 BILLION IMAGES SHARED DAILY
- 3. Facebook daily image uploads: 350 million (November 2013) Instagram daily image uploads: 60 million (March 2014) Twitter: 500 million tweets daily (March 2014) 30% pics Snapchat daily snaps: 400 million (November 2013) Important: Lots also happening on messaging apps!
- 4. Now we carry a camera with us all the time Everyday snaps Witnessing events
- 5. High smartphone penetration (65% US/UK) 80% by January 2015, 90% towards end of 2017 Rise of platforms and apps focused on visual content Pinterest Tumblr Instagram Vine Snapchat
- 6. โa photo of Barack and Michelle Obama, posted immediately after Obamaโs re-election as President of the United States in November 2012, rapidly became the single most retweeted message in the history of Twitter.โ (โIntroductionโ, Twitter and Society, 2013)
- 7. Picturing the Social: transforming our understanding of images in social media and Big Data research (NB โ still images only!) ESRC Transformative Research grant (18-months, started in September 2014) Interested in everyday/mundane/exceptional -> Taking research beyond Twitter <- Twitter all social media | may not โ be โrightโ platform
- 8. The project topic screams out for interdisciplinarity! Farida Vis (PI) โ Media and Communication Simon Faulkner โ Art History/Visual Culture James Aulich - Art History/Visual Culture Olga Gorgiunova โ Software Studies/Sociology Mike Thelwall โ Information Science/software Francesco DโOrazio โ Industry/Media/software Anne Burns (RA) โ Visual Culture / Media Studies
- 9. Diverse and cross sector (with industry)!
- 10. Picturing the Social conference #VisSocMedConf (all audio on SoundCloud)
- 11. โข Screen Shot 2014-07-09 at 06.05.51 Creating PhD/postdoc opportunities in the Lab VisSocMedLab Twitter Pinterest SoundCloud Storify
- 12. Research Questions driving Picturing the Social project: 1. How can these new forms of social expression be understood through studying their technical infrastructures? 2. How can a bottom-up understanding of social media users inform how these structures are experienced? 3. How can we devise comprehensive methodological and theoretical strategies for the capture and interpretation of social media image data at different scales? 4. How can the study of images become a central focus within social media and Big Data research and have an impact on key stakeholders and end-users? FSA: various | WEF: UGC, verification, ethics (beyond true/false)
- 13. Images largely ignored in social media research Not easy to โmineโ Hard to figure out meaning Context! Context! Context! Huge interest in industry (rapid new developments!)
- 14. And recent similar breakthrough developments at Stanford and Google. Though # of pizzas disputed.
- 15. Rapid rise in visual social media analytics platforms
- 16. Topic based image searching Location based image searching
- 17. Using cutting edge industry tools + firehose social media data to build image tool for academia Based on team + wider user requirements DโOrazio: social media data and beyond; tracking content across the social web. Thelwall: Using public APIs to test and experiment โQualitative data on a quantitative scaleโ (DโOrazio, 2013) Aim: free image tool for research community
- 18. Tools developed by Francesco DReโOarl aizninoovation: weighs the visibility
- 19. Tools using public APIs, including Twitter, YouTube, Tumblr, Flickr. Since start of project: image downloads for Twitter and Tumblr available! Tools developed by Mike Thelwall
- 20. Building new theory: traditional broadcasting model Production of message Message = text Reception of the message
- 21. Picturing the Social approach Structures Users Content
- 24. Facet Methodology (Mason 2011) Cut gemstone and their facets โThe aim of our facet methodology approach is to create a strategically illuminating set of facets in relation to specific research concerns and questions: not a random set, or an eclectic set, or a representative set, or a total set. The rigour or the approach comes ultimately from researcher skill, inventiveness, insight and imagination โ in deciding how best to carve the facets so that they catch the light in the best possible way.โ (77: 2011)
- 25. IInformed by Facet Methodology โข Episodic (event driven) โข Temporal โข Ontological โข Cultural โข Place/location
- 26. โข Episodic: Thatcher death tweets โข Temporal: Slices of time: 24 hrs (public) | 1 sec (firehose) โข Ontological: What is the thing? โข Cultural: The atlas project โข Place: Sheffield (geolocation, location, location) โข Make up of these teams: key is to unsettle
- 27. Thatcher Death Tweets Lead: Simon Faulkner Simon Faulkner talk at Picturing the Social Conference: โSome Initial Thoughts on the โThatcher Death Tweetsโโ โPilot studyโ to inform the rest of the facet projects. Initial: 100K from GNIP (to get the conversation started) Now: re-build and re-run queries in Pulsar (cross platform!)
- 28. Cross-cutting methods โข 36 in-depth interviews with: industry, academia, social researchers โข (users) Nationally representative survey - If we want to know how/why users create content and meaning: we should ask them not simply project! โข (users) Series of focus groups โข (users) 12-month digital ethnography โข Building a tool (based on user requirements!!!) โข Throughout all: structures, users, content
- 29. Thinking a bit more about structures โข How platforms make images visible? โข How tensions between companies make images more/less visible โ dynamic field (capital = key) โข How do APIs give access to the dataโwe wantโ? โข How companies act as data intermediaries โข How do these both strip and add metadata? โข How interfaces and dashboards add further layers and structure our thinking and research? โข How we know quite little about the tools we use.
- 30. How do social media companies make images visible?
- 31. Seen on a friendโs FB wall because she had liked it. Close friend -> surfaced on my newsfeed as well.
- 32. Algorithmic visibility: seeing the image through Edgerank How does Edgerank โseeโ? Finding and (re)tracing the image Image as โimageโ | Image as โlinkโ
- 33. Think critically how we study โthe feedโ โthe streamโ and at the same time realise theyโre not the only game in town! Visibility not a โgivenโ โ linked to a series of small actions, the connections between users and value attributed by the algorithm. How can we think about new scopic regimes in this context? How can we make their workings visible? Belting: images as โnomadsโ (and also shape shifters?)
- 34. Photo wars: Twitter vs Facebook/Instagram Sharing an image on Twitter through the ages Tips from 2009: 1. Via a Twitter Image Sharing Service 2. Via SMS or Email 3. Via Brightkite or FriendFeed 4. Via Skitch 5. Via an Encoded Tweet August 2011: Twitter rolls out direct image upload December 2012: Twitter no longer displays Instagram pics. Twitter cards switched off. Users only view images โnativelyโ September 2014: Future of Twitpic uncertain
- 37. Users hack solutions using IFTTT recipes
- 38. Using APIs and data intermediaries: GNIP Beyond Twitter! What do they offer? Firehose data = โcomplete accessโ Including historical access to Twitter
- 39. โManaged public API accessโ GNIP
- 40. Data enrichment + historical data GNIP
- 41. Data sources: DataSift (used for Pulsar) Twitter: 200+ million active users Tumblr: 300+ million active users
- 42. Tools for querying historical Twitter data TextSifter: data downloaded in DiscoverText
- 43. Limited access APIs: for specific users Reading the API fine-print: Instagramโs Terms of Use 3. You cannot use the Instagram API to crawl or store usersโ media without their express consent.
- 44. How do our academic (Twitter) tools work? โข Due to open access API ecosystem, we have witnessed an explosion in Twitter tools. But how do we evaluate them? NB - Do we evaluate them? โข What kinds of research questions do they allow us to answer? What kind of research question do we ask? โข What kind of โmethodโ is using an API? โข How do these tools allow us to search for data? โข What methods do these tools foreground? โข What research possibilities do they close off? โข NB - No โForrester Waveโ for academic social media tools. Maybe we should change that!
- 45. Picturing the Social โข Moving beyond the descriptive and counting โข Moving beyond studies about single platforms โข How are they connected within wider ecosystem? Include all other media! โข How can we study the circulation of these images? โข How can we develop approaches that take into account structures, users and content? โข Where does the social come in? What can we learn about these new social worlds?
- 46. visualsocialmedialab.org @VisSocMedLab f.vis@sheffield.ac.uk @flygirltwo