In my previous posts on this blog I have surveyed the digital preservation state of the District of Columbia Public libraryâ€™s Washingtoniana collection. This survey was preformed viaÂ an interview with Digital Curation Librarian Lauren Algee Â using the NDSA levels of digital preservation as a reference point.
In our survey we discovered that the DCPL Washingtoniana collection has very effective digital preservationÂ which through a combination of knowledgeable practices and the Preservica service (an OAIS compliant digital preservation service) naearly reaches the 4th Level in every category of the NDSA levels of Digital Preservation. With this in mind my next step plan for the archive looks at a number of areas the archive has been interested in expanding and presenting some thoughts on where they could begin taking steps towards preservation of those materials.
Of particular interest in this regard is the collecting of website materials. Being dynamic objects of a relatively new media, collecting these items can be fairly complex as it is hard to precisely pin down to what extend is a website sufficiently collected. Websites may appear differently on different browsers, they may contain many links to other websites, they change rapidly, and they often contain multimedia elements. As such outlined below will be a policy which discusses these issues and specifically offers a digital preservation plan for websites.
Website Digital Preservation Policy for the Washingtoniana collection
The Washingtoniana collection was founded in 1905 when library director Dr. George F. Bowerman began collection materials on the local community. The collection stands as one of the foremost archives on the Washington, D.C area, community, history, and culture. Naturally it makes sense then with the increasing movement of DC social life and culture to online or born digital platforms that the Washingtoniana collection would consider collecting websites.
The same criteria for determining selection of materials for Washingtoniana materials should apply here. Websites should be considered if they pertain to Washington, DC or its surrounding areas, events that take place in or discus that area, pertain to prominent Washington D.C. related persons, DC related institutions, or websites otherwise pertaining to Washington D.C. community, arts, culture, or history.
Like any physical preservation decision, triage is an essential process. Websites that are likely to be at risk should be high priority. In a sense all web content is at risk. Websites that are for a specific purpose, or pertain to a specific event may have a limited operational window. Websites for defunct businesses, political election sites, and even an existent website on a specific day may be vulnerable and thus a candidate for digitization. In addition the materials in question should not be materials which are being collected elsewhere, and should be considered in relation to the rest of the collection.
Although automation tools may be used for identification, discretion for selection is on librarian hands. In addition, suggestions from patrons relevant to the collection should be considered, and a system for managing and encouraging such suggestions may be put in place.
A metadata standard such as MODS (Metadata Object Description Standard ) should be used to describe the website. MODS is a flexible schema expressed in XML, is fairly compatiable with library records, and allows more complex metadata than Dublin Core and thus may work well. Metadata should include but not be limited to website name, content producers, URL, access dates, fixity as well as technical information which may generated automatically from webcrawlers such as timestamps, URI, MIME type, size in bytes, and other relevant metadata. Also, extraction information, file format, and migration information should be maintained.
A variety of collection tools exist for web archiving. The tool selected should be capable of the below tasks as outlined by the Library of Congress web archiving page
- Retrieve all code, images, documents, media, and other files essential to reproducing the website as completely as possible.
- Capture and preserve technical metadata from both web servers (e.g., HTTP headers) and the crawler (e.g., context of capture, date and time stamp, and crawl conditions). Date/time information is especially important for distinguishing among successive captures of the same resources.
- Store the content in exactly the same form as it was delivered. HTML and other code are always left intact; dynamic modifications are made on-the-fly during web archive replay.
- Maintain platform and file system independence. Technical metadata is not recorded via file system-specific mechanisms.
A variety of tools are capable of this task, a web crawler such as the Heritrix open source archival webcrawler or a subscription solution Archive-IT should be used. Both are by the Internet Archive, however the first is more of an open source solution while the second is a subscription based service which offers storage on Internet Archive servers.
Upon initial collection fixity should be taken using a Checksum system. This can be automated either with a staff written script or a program like Bagit, which automatically generates fixity information. This information should be maintained with the rest of the metadata for the digital object.
Websites should be kept in the most stable web archival format available. At the moment of this posts writing that format should be the WARC (Web ARChive) file format. This format allows the combination of multiple digital resources into a single file, which is useful as many web resources are complex and contain many items. Other file formats may be accepted if archived webpages are received from donors.
Upon initial ingestion items may be kept on internal drives, and copied to at least one other location. Before the item is moved into any further storage system the file should be scanned for viruses, malware, or any other undesirable or damaging content using safety standards as agreed upon with the division of IT services. At this point fixity information should be taken as described above, and entered into metadata record.
Metadata should be described as soon as possible, as to which point the object with attached metadata should be uploaded into The Washingtonianaâ€™s instance of Preservica.
Although Preservica automates much of the preservation process, a copy of the web archive should be kept on external hard drives. On a yearly interval a selection of the items within the harddrive should be checked against the items in Preservica to insure the Preservica fixity checks and obsolesce monitoring are working as desired.