Successfully reported this slideshow.
We use your LinkedIn profile and activity data to personalize ads and to show you more relevant ads. You can change your ad preferences anytime.

Approaches of Data Analysis: Networks generated through Social Media

1,230 views

Published on

Short Talk on SMART Data Sprint 2017 at NOVA University of Lisbon I iNOVA Media Lab.
23 January 2017

Published in: Social Media
  • Hey guys! Who wants to chat with me? More photos with me here 👉 http://www.bit.ly/katekoxx
       Reply 
    Are you sure you want to  Yes  No
    Your message goes here

Approaches of Data Analysis: Networks generated through Social Media

  1. 1. APPROACHES OF DATA ANALYSIS: Networks generated through Social Media NOVA University, Lisbon PhD candidate at UT Aus;n I Portugal @jannajoceli ˚ thesocialplaCorms.wordpress.com SMART Data Sprint 23-27 January 2017 Janna Joceli C. de Omena
  2. 2. Omena, 2017. Approaches of Data Analysis SMART Data Sprint 23-27 January 2017 ˚ Universidade Nova de Lisboa ˚ iNOVA Media Lab Black Mirror (2016), Nosedive We cannot speak of data analysis without considering the logic, features, grammars or the “ways of being” of social media.
  3. 3. Social Media Platforms SMART Data Sprint 23-27 January 2017 ˚ Universidade Nova de Lisboa ˚ iNOVA Media Lab channels of connectivity and sociability that must be taken as techno-cultural constructs objects of study + methodological process Social phenomena + means of media critique (Rogers, 2015) “structure of feelings” (Papacharissi, 2015) alternative form of journalism (Poell & Borra, 2012; Cardoso & Fátima, 2013; Malini et. al., 2014;) Concepts Programmability Popularity Connectivity Datification (Dijck & Poell, 2013) Logic Posts URLs Tweets Comments Replies Hashtags Location Memes Links Channels Grammars’ of Action (Agre,1994) Lack of neutrality Omena, 2017. Approaches of Data Analysis
  4. 4. SMART Data Sprint 23-27 January 2017 ˚ Universidade Nova de Lisboa ˚ iNOVA Media Lab Omena, 2017. Approaches of Data Analysis Social Media Studies with Digital Methods Machine-readable interfaces (Berlind, 2015) «Give third-parties access to data and functionalities that belong to the platform»
  5. 5. Omena, 2017. Approaches of Data Analysis What digital objects are available for data extraction? What media content can be part of my analysis? How far back in time can data be retrieved? What are the standard output files? SMART Data Sprint 23-27 January 2017 ˚ Universidade Nova de Lisboa ˚ iNOVA Media Lab Social Media Studies with Digital Methods
  6. 6. Omena, 2017. Approaches of Data Analysis Pages Groups Page Network Shared Links List of Events Users Key words Hashtags Loca;on Data extraction* Media content and digital objects Videos Channels Hashtags Loca;ons Follow Network Hashtags Posts (textual content: Cap;on, comments, replies) (visual content: videos, photos, memes) Page Like Network Groups Network Events URLs Tweets (textual content: Tweet text, men;ons, replies) (visual content: videos, photos, gifs) Geotags URLs Video Info (basic info an stats, comments, comments authors, interac;ons between users in the comment sec;ons) Video List and Network Channel Info and Network *Tools: Netvizz, Twitonomy, DMI-TCAT, YouTube Data Tools, Visual Tagnet Explorer, Tumblr Tool Media and users Info (textual content: cap;on, tags, users bio) (visual content: photo and video) (basic stats) Co-Tag Network Posts (textual content: summary, cap;on, tags, users bio) (visual content: photo and video) (basic stats) Co-Tag Network Output files CSV., TAB. GDF., XML., interac;ve chart
  7. 7. SMART Data Sprint 23-27 January 2017 ˚ Universidade Nova de Lisboa ˚ iNOVA Media Lab Omena, 2017. Approaches of Data Analysis Social Media APIs (limited data access) Pages or Open Groups Data = months or years Events = list of upcoming events (not past events) Twitter Search API = hours or few days (e.g. it returns to Twitonomy a sample of up to 3,100 tweets) Hashtag or locale based extraction = months or years (e.g. results will depend on the popularity of a hashtag and the adoption of the tag itself by users)
  8. 8. SMART Data Sprint 23-27 January 2017 ˚ Universidade Nova de Lisboa ˚ iNOVA Media Lab Network Analysis on Social Media Omena, 2017. Approaches of Data Analysis
  9. 9. SMART Data Sprint 23-27 January 2017 ˚ Universidade Nova de Lisboa ˚ iNOVA Media Lab Omena, 2017. Approaches of Data Analysis My personal friendship connec;ons on Facebook in January 2014. Extrac;on So`ware: Netvizz. Visualiza;on so`ware: Gephi.
  10. 10. SMART Data Sprint 23-27 January 2017 ˚ Universidade Nova de Lisboa ˚ iNOVA Media Lab Network Analysis on Social Media •  Explore associations •  Identify unexpected connections •  Key or marginal actors •  Mapping: Alliances and oppositions (Bounegru et.al, 2016) Program and anti-program (Rogers, 2017, forthcoming) Supporters and non-supporters (Omena, 2017, forthcoming) •  Clusters and weak/hidden ties •  Authority •  Activity (nodes properties) and weight of connections (edges properties) Omena, 2017. Approaches of Data Analysis
  11. 11. SMART Data Sprint 23-27 January 2017 ˚ Universidade Nova de Lisboa ˚ iNOVA Media Lab Choose a node attribute: Omena, 2017. Approaches of Data Analysis •  Degree = total n. of connections out-degree = activity in-degree = popularity •  People talking about = Debate (Facebook parameter: https://developers.facebook.com/docs/graph-api/reference/v2.1/page) •  Modularity = Clusters (community detection algorithm) R. Lambiotte, J.-C. Delvenne, M. Barahona Laplacian (2009). Dynamics and Multiscale Modular Structure in Networks. •  Betweenness or Bridgeness Centrality = Influence/Discriminate between local centers and global bridges (key players) (Ulrik Brandes (2001).A Faster Algorithm for Betweenness Centrality, in Journal of Mathematical Sociology 25(2):163-177); (Pablo Jesen et. al (2015). Detecting global bridges in networks. Journal of Complex Networks. Doi:10.1093/comnet/cnv022) •  PageRank = Authority/Importance (pagerank algorithm) Sergey Brin and Lawrence Page (1998).The Anatomy of a Large- Scale Hypertextual Web Search Engine, in Proceedings of the seventh International Conference on the World Wide Web (WWW1998):107-117
  12. 12. SMART Data Sprint 23-27 January 2017 ˚ Universidade Nova de Lisboa ˚ iNOVA Media Lab Hashtag Exploration #lovewins Bas;aan Baccarne, Angeles Briones, Stefan Baack, Emily Maemura, Janna Joceli, Peiqing Zhou, Humberto Ferreira. Digital Methods Summer School 2015, Does love win? The mechanics of meme;cs, heps://wiki.digitalmethods.net/Dmi/SummerSchool2015DoesLoveWin. Mapping: program and anti-program (Rogers, 2017, forthcoming) supporters and non-supporters (Omena, 2017, forthcoming) Omena, 2017. Approaches of Data Analysis
  13. 13. SMART Data Sprint 23-27 January 2017 ˚ Universidade Nova de Lisboa ˚ iNOVA Media Lab Omena, 2017. Approaches of Data Analysis Hashtag Exploration Bas;aan Baccarne, Angeles Briones, Stefan Baack, Emily Maemura, Janna Joceli, Peiqing Zhou, Humberto Ferreira. Digital Methods Summer School 2015, Does love win? The mechanics of meme;cs, heps://wiki.digitalmethods.net/Dmi/SummerSchool2015DoesLoveWin.
  14. 14. SMART Data Sprint 23-27 January 2017 ˚ Universidade Nova de Lisboa ˚ iNOVA Media Lab Omena, 2017. Approaches of Data Analysis Hashtag Exploration Bas;aan Baccarne, Angeles Briones, Stefan Baack, Emily Maemura, Janna Joceli, Peiqing Zhou, Humberto Ferreira. Digital Methods Summer School 2015, Does love win? The mechanics of meme;cs, heps://wiki.digitalmethods.net/Dmi/SummerSchool2015DoesLoveWin.
  15. 15. SMART Data Sprint 23-27 January 2017 ˚ Universidade Nova de Lisboa ˚ iNOVA Media Lab Hashtag Exploration #lovewins Page Like Network Exploring: associations and connections Page activity Debate within the network Main organizers of pro-impeachment protests in Brazil, 2015 Mapping: program and anti-program (Rogers, 2017, forthcoming) supporters and non-supporters (Omena, 2017, forthcoming) Omena, 2017. Approaches of Data Analysis
  16. 16. SMART Data Sprint 23-27 January 2017 ˚ Universidade Nova de Lisboa ˚ iNOVA Media Lab Page Like Network on Facebook Movimento Brasil Livre and Vem Pra Rua Brasil page like network (depth 1), March 2015. Node size: degree. Colours: clusters. Data extrac;on by Netvizz and vizualiza;on by Gephi. Movimento Brasil Livre and Vem Pra Rua Brasil page like network (depth 2), March 2015. Node size: degree. Colours: clusters. Data extrac;on by Netvizz and vizualiza;on by Gephi. (Omena and Rosa, 2015) Omena, 2017. Approaches of Data Analysis Vem pra Rua Brasil
  17. 17. SMART Data Sprint 23-27 January 2017 ˚ Universidade Nova de Lisboa ˚ iNOVA Media Lab The Tricks of Single Attributes Omena, 2017. Approaches of Data Analysis Movimento Brasil Livre and Vem Pra Rua Brasil page like network (depth 2), March 2015. Node size: in-degree. Colours: clusters. Data extrac;on by Netvizz and vizualiza;on by Gephi. Node size: In-Degree Colours: Modularity Node size: Out-Degree Colours: Modularity Movimento Brasil Livre and Vem Pra Rua Brasil page like network (depth 2), March 2015. Node size: out-degree. Colours: clusters. Data extrac;on by Netvizz and vizualiza;on by Gephi. Page Activity Page Popularity
  18. 18. SMART Data Sprint 23-27 January 2017 ˚ Universidade Nova de Lisboa ˚ iNOVA Media Lab The Tricks of Single Attributes Omena, 2017. Approaches of Data Analysis 1.  Activity (out-degree) does not call for popularity (in-degree).
  19. 19. SMART Data Sprint 23-27 January 2017 ˚ Universidade Nova de Lisboa ˚ iNOVA Media Lab The Tricks of Single Attributes Omena, 2017. Approaches of Data Analysis Movimento Brasil Livre and Vem Pra Rua Brasil page like network (depth 2), March 2015. Node size: degree. Colours: clusters. Data extrac;on by Netvizz and vizualiza;on by Gephi. (Omena and Rosa, 2015) Who generated more debate? (people talking about)
  20. 20. SMART Data Sprint 23-27 January 2017 ˚ Universidade Nova de Lisboa ˚ iNOVA Media Lab The Tricks of Single Attributes Omena, 2017. Approaches of Data Analysis 1.  Activity (out-degree) does not call for popularity (in-degree). 2.  Populate Facebook(e.g. MBL created 68 Facebook pages in 2015) or to have a high number of pages around the same topic does not mean to generate or create debate.
  21. 21. SMART Data Sprint 23-27 January 2017 ˚ Universidade Nova de Lisboa ˚ iNOVA Media Lab Hashtag Exploration #lovewins Exploring: associations and connections Page activity Debate within the network Main organizers of pro-impeachment protests in Brazil, 2015 Mapping: program and anti-program (Rogers, 2017, forthcoming) supporters and non-supporters (Omena, 2017, forthcoming) Omena, 2017. Approaches of Data Analysis Jornalistic Storytelling Exploring: Associations around single actors (ego-network) (Bounegru, et. al, 2016) “Connected China” (Thomson Reuters, February 2013) http://china.fathom.info/ Page Like Network
  22. 22. SMART Data Sprint 23-27 January 2017 ˚ Universidade Nova de Lisboa ˚ iNOVA Media Lab Omena, 2017. Approaches of Data Analysis http://china.fathom.info/
  23. 23. SMART Data Sprint 23-27 January 2017 ˚ Universidade Nova de Lisboa ˚ iNOVA Media Lab Omena, 2017. Approaches of Data Analysis http://china.fathom.info/
  24. 24. SMART Data Sprint 23-27 January 2017 ˚ Universidade Nova de Lisboa ˚ iNOVA Media Lab An analytical perspective: Omena, 2017. Approaches of Data Analysis i) Dominant voice ii) Concern iii) Commitment iv) Positioning v) Alignment Critical Analytics and Engagement Metrics (Rogers, 2016)
  25. 25. SMART Data Sprint 23-27 January 2017 ˚ Universidade Nova de Lisboa ˚ iNOVA Media Lab Data Critique Omena, 2017. Approaches of Data Analysis i)  Situate social media data in time and space ii) Social media APIs are never neutral iii) Social media data does not act out of context iv) Data is never ‘raw’ (Adapted from Dalton and Thatcher, 2016) Data are not simple evidence of phenomena, they are phenomena in and of themselves (Wilson, 2014). It (data) has always been “baked” through both its construction and its resulting interpretation (Gitelman, 2013). (apud Dalton and Thatcher, 2016, p.4)
  26. 26. NOVA University, Lisbon PhD candidate at UT Aus;n I Portugal @jannajoceli ˚ thesocialplaCorms.wordpress.com Janna Joceli C. de Omena Thanks for your time and attention =)

×