Social media data archives in an API-driven world

Abstract

In this article, we explore the long-term preservation implications of application programming interfaces (APIs) which govern access to data extracted from social media platforms. We begin by introducing the preservation problems that arise when APIs are the primary way to extract data from platforms, and how tensions fit with existing models of archives and digital repository development. We then define a range of possible types of API users motivated to access social media data from platforms and consider how these users relate to principles of digital preservation. We discuss how platforms' policies and terms of service govern the set of possibilities for access using these APIs and how the current access regime permits persistent problems for archivists who seek to provide access to collections of social media data. We conclude by surveying emerging models for access to social media data archives found in the USA, including community driven not-for-profit community archives, university research repositories, and early industry-academic partnerships with platforms. Given the important role these platforms occupy in capturing and reflecting our digital culture, we argue that archivists and memory workers should apply a platform perspective when confronting the rich problem space that social platforms and their APIs present for the possibilities of social media data archives, asserting their role as "developer stewards" in preserving culturally significant data from social media platforms.

Counter-archiving Facebook

“…API-based research offers methodological comfort in providing researchers with structured data, a clear demarcation of the types of data that can be extracted, their volume and the restrictions on their use, which are specified in the platform’s terms of use (Venturini and Rogers, 2019). Although previous research proposed using APIs for archiving social media (Acker and Kreisberg, 2019; Littman et al., 2018; Lomborg, 2012), the majority of social media researchers have used API data for immediate analysis, rather than for long-term preservation, appraisal of sources or lending access to others.…”

Using Machine Learning to Enhance Archival Processing of Social Media Archives

“…As early as 1992, a statistical content analysis demonstrated that a statistical approach could assist in archival description of the content of early social media implementations, and especially in surfacing latent patterns in that content [24]. In their examination of using platform APIs to identify social media content for preservation, Acker and Kriesberg cautioned that that this approach presents a new evidentiary and administrative challenge for archivists because it can result in decontextualizing social media data streams [1]. Most notably, the 2018 failure of the Library of Congress to be able to provide access to the entirety of its massive Twitter Archive has resulted in a number of more selective efforts to archive and provide access to Twitter content [6].…”

“…The aggregated data is what comprises the COVID-19 Hate Speech Twitter Archive (CHSTA), which we release publicly on Github in order to encourage its use by other researchers. 1 Table 1 shows the number of tweets in CHSTA that were collected through each wave. Although the coverage of each wave differs by a few days, we find that wave 4 has significantly more tweets compared to that of other waves.…”

Perspective: Acknowledging Data Work in the Social Media Research Lifecycle

“…However, sharing "officially" and transparently not only requires effort but also certainty about the legal and ethical limitations that apply for a specific dataset. There is hence little (but growing) evidence of official sharing (for an overview see Thomson, 2016;Acker and Kreisberg, 2019) but also an unmeasured "gray market" in which data is shared informally amongst researchers within the same group or field (Weller and Kinder-Kurlanda, 2015).…”