Department of History, Faculty of Arts
University of Waterloo
Ian Milligan is an Associate Professor of History at the University of Waterloo, where he teaches Canadian and digital history. He is currently the principal investigator of the Archives Unleashed project, which seeks to make web archives accessible to humanities and social sciences researchers. Ian has published several books: the forthcoming History in the Age of Abundance? How the Web is Transforming Historical Research (April 2019), the SAGE Handbook of Web History (co-edited with Niels Brügger, 2018), Exploring Big Historical Data: The Historian’s Macroscope (co-authored with Scott Weingart and Shawn Graham, 2015), and Rebel Youth: 1960s Labour Unrest, Young workers, and New Leftists in English Canada (2014). In 2016, Ian was named the Canadian Society for Digital Humanities’s recipient of the Outstanding Early Career Award.
Working with Cultural Heritage at Scale: Developing Tools and Platforms to Enable Historians to Explore History in the Age of Abundance
The rise of the Web as a primary source will have deep implications for historians. It will affect our research — how we write and think about the past — and it will change how humanists and social scientists make sense of culture at scale. Scholars are entering an era when there will be more information than ever, left behind by people who rarely entered the historical record before. Web archives, repositories of archived websites dating back to 1996, will fundamentally transform scholarship, requiring a move towards computational methodologies and the digital humanities.
The talk explores this dramatic shift — and what is to be done about it — by arguing that historians will have to understand how to work with textual (and other) data at scale. Historians will soon need to become familiar, at the very least, with NLP techniques. This is not just a marginal problem: the need to explore the big data of the Web (and other digitized repositories) strikes to the core of our discipline.
All Historians Have to Begin to Work with Data
Initial moves towards digital methods have been very promising, as historians begin to study the 1990s. Even so, they will discover sooner than they think that one cannot write most histories of the 1990s or later without reference to web archives. They must be ready, but they are hamstrung. The profession has largely turned away from statistics and from quantitative methodologies more generally; and the web archiving analysis ecosystem is largely based on tools that require a high level of technical expertise. Access to web archives at scale requires, more often than not, fluency with command-line interfaces, access to high-performance computing, and storage at the terabyte scale. Historians need to analyze web archives to write histories, yet that requires skills and infrastructure beyond what one can reasonably expect of them. What, then, can be done?
Tools and Platforms: The Archives Unleashed Project
The talk introduces this problem, and discusses the process of developing tools and platforms to enable historians to explore this “age of abundance”. It does so by highlighting the Archives Unleashed Project, an interdisciplinary initiative funded by the Andrew W. Mellon Foundation. The project’s goal is to “make petabytes of historical Internet content accessible to scholars and others interested in researching the recent past”, and brings together a historian, a computer scientist, and a librarian to lead a team to develop such infrastructure. The project will achieve it in three main ways.
- The Archives Unleashed Toolkit is an open-source platform for analyzing web archives with Apache Spark. It is a scalable toolkit, based upon a process cycle that we have developed; we call it the Filter-Analyze-Aggregate-Visualize cycle. To use the Toolkit, a scholar first filters down a large web (a particular range of dates, a domain, or only pages with certain keywords present); analyzes (finds links, or named entities, sentiment, topics); aggregates (summarizes the output); and visualizes (either through various data tools or tabular data). The Toolkit, based on a command-line interface, is unfortunately very difficult to use.
- The Archives Unleashed Cloud is a web-based front-end for working with the Toolkit. It takes data from the Internet Archive and processes it into formats familiar to researchers: network diagrams, filtered text files, and other statistical information about a collection. We also provide all of this data for download with a bundled Jupyter Notebook. This allows scholars to use a web-based interface to perform basic data science operations on the data: draw on popular computational linguistics or data science Python libraries to process data and find answers. Suddenly, working with web archives is not so terrifying, and the users have been connected to the mainstream of the Natural Language Processing world.
- We run a series of datathons (three to date, as part of the Mellon grant). They bring together domain experts, researchers, and others to work with web archive data at scale and so help lower barriers; connect people interested in the topic and build community; and help develop a body of practice around web archiving collection and analysis practices.
The talk explores ways in which we can help historians move into an age when working with cultural heritage at scale is no longer a “nice to have” but a necessary component of studying periods from the 1990s onwards.