De- and Reassembling Data Infrastructures
Download as PPTX, PDF0 likes210 views
The document discusses the role of digital methods in e-research, focusing on the collection and analysis of data from various online platforms like Twitter and Instagram. It highlights the challenges and requirements for data infrastructures, including data production and cross-platform syndication, as well as the methodologies and tools necessary for effective data analysis and visualization. Overall, it emphasizes the importance of aligning data sources, tools, and visualization techniques with the research objectives and the diverse needs of different publics.
De- and Reassembling Data Infrastructures
- 1. Carolin Gerlitz (based on joint work with Liliana Bounegru & Jonathan Gray) University of Siegen – Digital Methods Initiative Amsterdam Infrastructuring eResearch Workshop, Dec 7 2016 DE- & REASSEMBLING DATA INFRASTRUCTURES
- 2. DIGITAL METHODS Widely used term. Repurposing of (1) digital data (2) technical features (3) analytical capacities. How can links, likes, shares, comments etc. be used for research? (Rogers 2013)
- 3. DIGITAL METHODS Structured extraction & analysis of data. Two main objectives: (1) study sociality online (2) understand medium specificity and socio-technical configurations Tool development, repurposing & training.
- 4. EXAMPLE: TCAT Twitter Capture and Analysis Toolkit (TCAT). Developed by Erik Borra & Bernhard Rieder. Data collection & analysis of Twitter data based on Streaming API.
- 5. DE-ASSEMBLING Digital methods rely on the participation of a variety of actors and entities. Data & tool chaining. What are the data infrastructures that underpin dm work? What challenges do they pose?
- 6. DATA PRODUCTION Data production as distributed accomplishment of users, platform activities, capture mechanisms, third party apps & cross-platform syndication.
- 7. (1) COMMENSURATION Cross-platform syndication, different interpretation of platform features, bots & automation. How to commensurate data from heterogeneous sources (Espeland & Stevens 1998)? s
- 8. (2) MULTIVALENT YET BIASED Data set out to cater to different analytical interests of stakeholders. At the same time: support some forms of analysis more than others (interestedness).
- 9. DATA EXTRACTION Scraping, crawling, API retrieval. Reliant on platform data structures and API politics, tools, plugins and scripts. Platforms determine the conditions of access to their data. Instagram Hashtag Explorer
- 10. DATA ANALYSIS Reliant on further tools for querying data, calculating metrics, stats or combination of data formats. DMI TCAT
- 11. (3) METHODOLOGICAL UNCANNY Open or commercial tools resonate with known methods – but not quite (Marres & Gerlitz 2015).
- 12. DATA VISUALISATION Visualisation standards and data outputs. Which data formats are amenable for which visualisation technique? What interestedness does visualisation introduce? D3, tableau, Gephi
- 13. (4) TOOL CHAINNG Assembling of different data sources & tools for different tasks into a methodological apparatus. Cascades of inscriptions (Ruppert et al 2013).
- 14. (5) DISTRIBUTED TOOL MAKING Many general purpose tools (incl. extensive documentation). Heterogeneous developers and emergent standards. Which tools can be chained? How can open source tools be maintained and scaled up?
- 15. (6) DATA PUBLICS Data assemble heterogeneous publics with different objectives, interests, skills & needs (Ruppert 2015, Birchall 2015). Researchers, companies, organisations, activists, journalism.
- 16. ALLINGING DATA INFRATSRUCTURES Methodological work as de- & reassembly. Specific to needs of publics. Alignment & mal-alignment of data sources, tools, visualisations and research objectives: need for repositories and shared dev.
- 17. (RE)IMAGINING DATA INSTRUCTURES From data literacy to data infrastructure literacy (Gray et al. 2017). Accounting for inscription, alignment and malalignment. Enable to re-think, re-assemble and re-align infrastructures. Methodological infrastuctural imagination (Bowker 2014).