Designing human-centered technologies to mobilize social media data into institutional contexts

Abstract

Social media platforms have become an established and alternative mechanism for communities to mobilize and exchange information in response to humanitarian or local crises. Due to the richness of experiences accumulated on social media platforms, this content can be valuable for civil and non-profit organizations working to address social and development challenges. At the core of my dissertation is to examine what entails not only analyzing social media data but also the implications of integrating the insights obtained from that analysis into the context of actors and institutions who might act upon those insights, such as civil and non-profit organizations.

Using social media data as evidence by institutions to inform their work entails three main challenges —accuracy, representation, and context— due to the nature of social media data. Additionally, using this type of content to inform the design of interventions and technologies that will support the studied communities entails reflecting on how we make sense of data. Within the CSCW and HCI community, there has been a growing focus on using a computational approach to establish metrics and develop tools to analyze and make inferences from social media data. However, by constraining the examination of this type of data through the exclusive use of computational techniques, there is a high risk of neglecting the social, cultural, and temporal context of the data.

In response, my fieldwork consisted in following a mixed-methods approach to understanding the underlying situations that force communities to use social media platforms as a means of organization and the implications for non-profit organizations to make social media data actionable to inform their work. Based on the findings of my fieldwork, I designed, deployed, and evaluated a toolkit addressed to practitioners working in civil and non-profit organizations interested in using data from Twitter to identify local capacities, monitor community crises, and develop interventions. The toolkit comprises computational tools that allow searching, collecting, and analyzing data from Twitter. Additionally, the toolkit includes a manual and worksheets that guide practitioners to critically approach social media data and recognize the possibilities and limitations of this type of data by considering the challenges previously mentioned —accuracy, representation, and context.

In summary, the outcome of my research provides empirical evidence and situated tools for approaching social media data not as an independent identity but rather in light of the interplay between the online and offline behavior of the communities that produce such data. This dissertation offers two contributions to the growing body of work in HCI and CSCW invested in reflecting on the transformative character of data. First, it illuminates the large ecosystem of norms and practices of multiple actors, infrastructures, and databases that we need to consider to mobilize data from online platforms into institutional contexts. Second, the design of the toolkit proposes an actionable example of how to promote a situated examination of data. While my research has focused only on examining data from social media platforms, the contributions of my work are meaningful in the broader context of data-centric technologies. As we continue to deploy this technology, it is imperative to interrogate the assumptions and biases encapsulated within those technologies, specifically the data that feed them, and how they impact our understanding of human networks and communities.