Bountiful Harvest: Curation, Collection and Use of Web Archives

The theme for the ARA Annual Conference 2017 is: ‘Challenge the Past, Set the Agenda’. I was fortunate enough to attend a pre-conference workshop in Manchester, ran by Lori Donovan and Maria Praetzellis from The Internet Archive, about the bountiful harvest that is web content, and the technology, tools and features that enable web archivists to overcome the challenges it presents.

Part I – Collections, Community and Challenges

Lori gave us an insight into the use cases of Archive-it partner organisations to show us the breadth of reasons why other institutions archive the web. The creation of a web collection can be for one of (or indeed, all) the following reasons:

  • To maintain institutional history
  • To document social commentary and the perspectives of users
  • To capture spontaneous events
  • To augment physical holdings
  • Responsibility: Some documents are ONLY digital. For example, if a repository upholds a role to maintain all published records, a website can be moved into the realm of publication material.

When asked about duplication amongst web archives, and whether it was a problem if two different organisations archive the same web content, Lori put forward the argument that duplication is not worrisome. The more captures of a website is good for long term preservation in general – in some cases organisations can work together on collaborative collecting if the collection scope is appropriate.

Ultimately, the priority of crawling and capturing a site is to recreate the same experience a user would have if they were to visit the live site on the day it was archived. Combining this with an appropriate archive frequency  means that change over time can also be preserved. This is hugely important: the ephemeral nature of internet content is widely attested to. Thankfully, the misconception that ‘online content will be around forever’ is being confronted. Lori put forward some examples to illustrate the point for why the archiving of websites is crucial.

In general, a typical website lasts 90-100 days before one of the following happens:

  1. The content changes
  2. The site URL moves
  3. The content disappears completely

A study was carried out on the Occupy Movement sites archived in 2012. Of 582 archived sites, only 41% were still live on the web as of April 2014. (Lori Donovan)

Furthermore, we were told about a 2014 study which concluded that 70% of scholarly articles online with text citations suffered from reference rot over time. This speaks volumes about preserving copies in order for both authentication and academic integrity.

The challenge continues…

Lori also pointed us to the NDSA 2016/2017 survey which outlines the principle concerns within web archiving currently: Social media, (70%); Video, (69%) and Interactive media and Databases, (both 62%).  Any dynamic content can be difficult to capture and curate, therefore sharing advice  and guidelines amongst leaders in the web archiving community is a key factor in determining successful practice for both current web archivists, and those of future generations.

Part II – Current and Future Agenda

Maria then talked us through some key tools and features which enable greater crawling technology, higher quality captures and the preservation of web archives for access and use:

  • Brozzler. Definitely my new favourite portmanteau (browser + crawler = brozzler!), brozzler is the newly developed crawler by The Internet Archive which is replacing the combination of heritrix and umbra crawlers. Brozzler captures http traffic as it is loaded, works with YouTube in order to improve media capture and the data will be immediately written and saved as a WARC file. Also, brozzler uses a real browser to fetch pages, which enables it to capture embedded urls and extract links.
  • WARC. A Web ARChive file format is the ISO standard for web archives. It is a concatenated file written by a crawler, with long term storage and preservation specifically in mind. However, Maria pointed out to us that WARC files are not constructed to easily enable research (more on this below.).
  • Elasticsearch. The full-text search system does not just search the html content displayed on the web pages, it searches PDF, Word and other text-based documents.
  • solr. A metadata-only search tool. Metadata can be added on Archive-it at collection, seed and document level.

Supporting researchers now and in the future

The tangible experience and use of web archives where a site can be navigated as if it was live can shed so much light on the political and social climate of its time of capture. Yet, Maria explained that the raw captured data, rather than just the replay, is obviously a rich area for potential research and, if handled correctly, is an inappropriable research tool.

As well as the use of Brozzler as a new crawling technology, Archive-it research services offer a set of derivative data-set files which are less complex than WARC and allow for data analysis and research. One of these derivative data sets is a Longitudinal Graph Analysis (LGA) dataset file which will allow the researcher to analyse the trend in links between urls over time within an entire web collection.

Maria acknowledged that there are lessons  to be learnt when supporting researchers using web archives, including technical proficiency training and reference resources. The typology of the researchers who use web archives is ever growing: social and political scientists, digital humanities disciplines, computer science and documentary and evidence based research including legal discovery.

What Lori and Maria both made clear throughout the workshop was that the development and growth of web archiving is integral to challenging the past and preserving access on a long term scale. I really appreciated an insight into how the life cycle of web archiving is a continual process, from creating a collection, through to research services, whilst simultaneously managing the workflow of curation.

When in Manchester…

Virtual Archive, Central Library, Manchester

I  couldn’t leave  Manchester without exploring the John Rylands Library and Manchester’s Central Library. In the latter, this interactive digital representation of a physical archive combined choosing a box from how a physical archive may be arranged, and then projected the digitised content onto the screen once selected. A few streets away in Deansgate I had just enough time in John Rylands to learn that the fear of beards is called Pogonophobia. Go and visit yourself to learn more!

Special collections reading room, John Rylands Library, Manchester

Leave a Reply