Historians who work with, or who are thinking about working with, web archives will be excited about the announcement that Archive-It Research Services made on March 17th. They’re widely expanding the sort of data that they provide to researchers. As they put it in their announcement:
The service will allow any Archive-It partner to give users, researchers, scholars, developers, and other patrons easily-analyzed datasets that contain key metadata elements, link graphs, named entities, and other data derived from the resources within their collections. By supporting access in aggregate to partner archives, ARS will facilitate new types of use, research, and analysis of the significant historical records from the web that Archive-It partners are working to collect, preserve, and make accessible.
They’re making three types of datasets available, the first of which are WAT files, which contain metadata about websites.
From WATs, you can get metadata descriptions for websites, the links that they point towards, the anchor text of those links, and crawl information. Ian Milligan, one of the co-authors of this blog, has been using WAT files to analyze Canadian political history websites: see some results here and here (including a guided video tour of some results).
LGA and WANE files are unfamiliar to the two authors of this blog, although they look to be very useful! LGA files accelerate the ability to do longitudinal link analysis from WAT files. The examples they give are actually from the Canadian Political Interest Groups collection! Finally, WANE files use Stanford NER to extract information relating to people, organizations, and locations. Using derived text from a web archive, Milligan plotted all the locations mentioned in GeoCities – you can see the results here.
To get these files, consult the service details page. In short, if you’re an Archive-It partner you can order it internally through your dashboard. For the rest of us researchers, you just need to send an e-mail in with some information, and start the process.
In short: an amazing move that’s really going to unlock these files. WARC files are really big – too big, for most systems – whereas the LGA, WAT, and WANE model is going to unlock accessible web archive research. Kudos to Archive-It.