Tuesday, February 24, 2009

February 24, 2009

It's past time for another update, and things have been moving along slowly but surely. The most major accomplishment I can report is that I've finished a draft Web Site Collection Plan. I last mentioned this at the end of January. It's based on my assessment of other web site collection policies as well as the Collection Plan template from the Web-at-Risk project. This draft is available here.

Some parts of this policy are incomplete, simply because they will depend on an overarching digital preservation program. Other parts depend on current metadata practices here in Wilson Library that I haven't been able to pin down yet. Of course they're all flexible recommendations. I think it would be a good place to start, and possibly even a document that can be expanded to other collecting areas.

There's also been some headway on the Web Curator Tool (WCT) installation, thanks to help from staff at ibiblio.org. I gather that it is installed, but I could not yet get it to run. I'm sure this is due to my own tenuous knowledge of how to run things in a Unix environment.

Finally, I've been working on my poster presentation for DigCCurr 2009, which is looming on the horizon. The title is "Getting the Tar Off of Our Heels: Moving Forward with Archiving University of North Carolina at Chapel Hill Websites." Since this poster hinges on the successful trial of WCT, I'm hoping that I can get that tool running in some capacity soon. I've decided to develop a rubric for measuring different aspects of the web archiving tools I've trialed. The three facets will be (1) ease of installation/setup (2) ease of use (3) features. I'm trying to find a vetted set of characteristics for evaluating software, but am thus far not having luck. Barring that, the characteristics I have come up with are highly subjective, but hopefully still meaningful.

Tuesday, February 10, 2009

February 10, 2009

I've finished my little experiment, as mentioned in the last post. I have to say, it was pretty enlightening, and I'm glad I took the time to do it. (As a side note, when I was almost finished, I suffered a data loss - you would think that a student of digital curation would know better - and had to start over. So, let that be a lesson to me for backing up my own data.) So here's how it went.

Using a random number generator, I selected 20% of the completed records schedules. This gave me 16 different schedules, listed here:

K-12 International Outreach Office
Research & Evaluation (OIA)
Program in the Humanities and Human Values (CoAS)
City & Regional Planning
Dean's Office (School of Medicine)
Biochemistry & Biophysics (School of Medicine)
Medical Science Teaching Labs (School of Medicine)
Business & Finance (School of Dentistry)
Human Resources (School of Dentistry)
Student Services (Dept of Maternal & Child Health)
Center for Urban & Regional Studies
Manuscripts Department (Wilson Library)
Conservation (Wilson Library)
Grounds Services (Facilities)
Building Services (Facilities)
Housekeeping (Facilities)

I read each schedule, keeping track of which series were supposed to be transferred to University Archives (I'll call these "UARMS series" here). I then tried to locate each department's associated website, and was able to do so in all cases. Finally, I scoured these sites looking for any records that fall into the UARMS series.

Here's an example, Housekeeping (Facilities) has 6 series whose records are normally transferred to UARMS in paper format. Two of those six, photographs and policies/procedures, can be found on the Housekeeping website. This means that documents that fall into 33% of its UARMS series records types can be found to some degree in digital form on their website.

As anticipated, the website pages ranged widely in depth and type of content. Some were only one page of text, others were numerous pages deep with multiple file types present. Here are the general results of my survey:
  • The percentage of UARMS series document types that could be found on the departments' websites ranged from 0% to 67%.
  • The average number of those types found on the website across all sites was only 27%.
  • The department with the most record-worthy online content was, surprisingly, Grounds Services.
  • Many of these departments had scheduled records series in common. The most common were committee and council records (12 out of 16), correspondence (15 out of 16), photographs (12 out of 16), and polices and procedures (13 out of 16). Of these, only photographs were routinely found online (92% of the departments).
  • Photographs, policies/procedures, and conference/workshop (conducted) records were the most common types found online.
  • Only 10 of these 16 departments had "website records" on their record schedules, even though all had some sort of website.
After undertaking this survey, I decided to look at the finalized General Administration schedules and to do the same thing. Because there are so few, I did all of them:

Academic Affairs Administration (incl Int'l Programs)
CFNC-Pathways
NC in the World (NC Center for Int'l Understanding)
Licensure Office (Academic Affairs)
Research and Sponsored Programs (Academic Affairs)

To compare, here are some more findings:
  • The percentage of UARMS series document types that could be found on the departments' websites ranged from 10% to 57%.
  • The average number of those types found on the website across all sites was 39%.
  • Much more prevalent for these departments were reports (including annual) and publications.
Considering the caveats mentioned in my last post about my methodology, I still feel that this was an informative exercise.
  1. I am surprised that more information isn't online, especially considering that UNC is a state university. I had expected a lot more online information in the form of events, strategic planning documents, statistics, and meeting records. I was gratified by the fact that the General Administration sites, on average, did have a lot more of this type of information.
  2. In general, I was also surprised at how little information some of these departments have on their websites, and how out of date some of that information is.
  3. I am curious about the number of intranets that exist on campus. I found one or two sites that had log in pages, and one that required a log in to view committee minutes, which was a UARMS series. I think/hope that much of the information I mentioned in #1 may in fact be on the intranet instead of the public web.
  4. Harvesting these websites will pull in a lot fewer documents that qualify as records than I had expected.
Despite my statement in #4, I still think harvesting these sites would be a great asset for the following two reasons. First, those sites that do put up UARMS series often only have the most recent version of the record online. I imagine this is replaced each time a newer version comes out. Harvesting websites could capture all of those documents. Second, the "website records" series would be greatly enhanced with the captured pages. I personally feel that any department that has a website (and all of them did, in some form or another) should have that series on their schedule.

Again, my method has some faults. My sample is small compared to the gamut of UNC departments. Even though I chose randomly, many of the departments were from the same school/college. Also, those who have completed schedules probably are more records-aware in the first place, and may, as a result, have more content online. It also probably isn't fair to look for correspondence or some types of personnel records online, even though they are UARMS series. These aren't the types of things people would typically put on a website. Still, if nothing else, scanning through all of these websites in search of records has given me a better idea of what exactly this institution is putting out there and what, eventually, we would pull in.

Thursday, January 29, 2009

January 29, 2009

Here's an update on what I've been working with in the last few weeks. It's been difficult trying to figure out where to go this semester, especially while I'm waiting for some technology-related projects to move forward.

1. iRODS: I've been thrown head first into iRODS in an incredibly short but very interesting class. I've gotten the client installed and my group is now working on identifying policies we can use to create rules.

2. Web Archiving Policies: In continuation from last semester, I'm trying to draft some web archiving policies using the Collection Plan Template from the Web-at-Risk project as a guide. This has proven to be more difficult and more simple than I expected. Here's what I mean. Because the UARMS impetus for collecting websites is partially grounded in North Carolina Public Records Law, much of this information can be taken directly from the department's current scope used for paper records, and can be based on the department's established mission. On the other hand, areas dealing with digital object preservation, migration and workflow are more difficult to determine without any overarching digital object policies to which I can refer.

I know from past work experience how people abhor the word "policy." But I'm coming to see that policies can end up serving a unifying purpose and also save work for those down the road. The difficulty, I guess, is in having all interested parties agree on one policy. It's a lot of effort up front and probably seems to delay "real" work. Peddling around on your own can end up giving others something to work with, and I hope that will be the case here.

3. A Little Experiment: I've decided to try an experiment in order to determine exactly how many scheduled records would be pulled in from websites. Using a random sequence generator, I've selected 20% of the UNC departments with completed records schedules. I'm going through and listing the series that would normally be transferred to UARMS at some point in time. Then, as best I can, I'm assessing the departments' websites for those series. Although collecting records would not be the only reason we would harvest the sites, I'm interested in seeing exactly how often scheduled records are put online. Are departments really posting documents UARMS would normally collect in print?

I recognize that this is limited in certain ways. For instance, those who have completed individual records schedules may be more likely to publish their important documents online in the first place. Second, those who do not have completed records schedules are probably departments we would or should be more aggressive in harvesting. I may also fail to locate the records during my survey of the website. Still, I want to make sure I have a closer idea to exactly what a web harvest would accomplish from a public records compliance angle, and I think this little experiment will help.

4. Ongoing Efforts: I'm reading any new publications I come across on web archiving. I'm alsoruminating on my poster presentation coming up in April at DigCCurr 2009. With the delays I've had in getting some web archiving tools set up, I may need to take that poster in a different direction. There's still a bit of time, there. Whatever it ends up being, I hope someone will find it useful!

Tuesday, January 20, 2009

January 20, 2009

Through some listservs, I caught wind of a year-old initiative of which I was previously unaware: LiWA (Living Web Archive). In its own words:

"LiWA's aim is to develop new tools and approaches that can deal with these challenges and improve archive fidelity, coherence, and interpretability. The strategy is to frame them so that they are more resistant to future changes, by leveraging universal tools like the browser and by fostering community-based responses to new challenges."

In my words, this initiative seems to be tackling some of the pervasive issues with current web archiving technology, issues like capturing deep and hidden web content, and dealing with what they call "temporal coherence." This last concept, from what I can understand, encompasses the following type of problem: if it takes you 2 days to crawl a large site, and parts of that site change during that time period, have you captured the site faithfully?

Topical areas they're exploring include the aforementioned temporal coherence, as well as archive fidelity, spam cleansing, semantic evolution, the social web, and rich media. There are a number of presentations on the website explaining these concepts, and I'll be digging into them this week.

Wednesday, January 14, 2009

January 14, 2009

The semester begins, and it's time to revisit the goals. Here they are, from the beginning of last semester, and my thoughts on what's happened.

Goals:

  1. What is the current idea of a responsible web archiving strategy, including preservation, metadata, access?
  2. Work to refine scope - how much do we want to keep/is feasible to keep
  3. Identify what has been tried, is being tried
  4. Compare these with UNC's situation (resources/collections)
  5. Review current policy and procedure
A lot of progress has been made, and I feel much more aware of what is achievable during my fellowship.

Goal #1 - I have read up on quite a bit of "practice" but even more projection. Institutions are either treating websites just like other objects they would normally ingest, are keeping their distance from collecting websites, or are outsourcing to the Internet Archive. My main discovery is that UNC seems to be with the pack on most of the issues.
Goals #2 and #3 - How much and how often? These are the big questions for harvesting. For our purposes, I feel that selective and routine but infrequent crawls would work the best. Domain-wide harvests are typically done by national libraries, and this seems beyond our scope for now. We also want to make sure we're dovetailing with other institutions in our area, not duplicating work. It would be my hope that selective, routine harvests would fill in a lot of the public records sought by University Archives, fulfilling the university's legal obligations, and also round out current series in the Southern Historical collection.
Goal #4 - As I mentioned above, UNC appears to be milling in the pack when it comes to pursuing a web archiving plan. In order to move ahead, some simple infrastructure and policy commitments would need to be made.
Goal #5 - I feel that this is where I will be starting this semester. Reading on and developing a solid policy and procedure base for archiving websites is, I feel, the best preparation I can give should the UNC libraries decide to begin regular harvesting. This will also support my work with iRODS this semester, which needs a rule base built from policies.

Tuesday, November 25, 2008

November 25, 2008

The American Library Association Association for Library Collections and Technical Services recently held a 3 day email forum on web archiving. I was clued in to this forum by one of the archivists here, and I'm glad I got to lurk a bit. Although the conversation was a little disjointed, and dominated by several respondents, it was interesting and enlightening. I thought I'd put down some of my thoughts here.


I got the impression that many who spoke up on the list were focusing on government documents. This reflects what I've found in the literature - most published accounts of web archiving in the US also seem to target government sites. This makes sense, considering (1) those sites have, what is considered to be, vetted information (2) have information libraries often collect without heavy selection procedures (3) are perceived as more stable than others.


Of those who responded, the most often mentioned service or tool was the Internet Archive's Archive-it. Others include HTTrack, Context Miner, Fedora, OCLC's Web Archives Workbench, and del.icio.us. This forum was also where I learned about Zoetrope (see my November 20 post).

Manual vs. automated harvesting also seemed to be equally represented. Most still feel that curating the materials is the valuable service that librarians can provide over and above a broad approach like that of the Internet Archive.

Although many stated they weren't actually harvesting web content yet, some of those who were mentioned that they have looked for copyright statements on pages and also sought permission to archive. I think that, because so many expressed their newness to this topic, and because of the newness of web archiving in general, this topic hasn't yet gained a critical mass of experience or attention.


Another really interesting thing was the number of respondents who said they're creating MARC records for freely available website resources, without actually harvesting those resources. This really astonishes me. Most cited storage and technological limitations as barriers to harvesting. I understand that the librarians are interested in getting their users as much information as possible, and hope to do so through catalog searches, but with the ephemeral nature of web content, not to mention unstable URLs, I would imagine these links rot quickly and it would be incredibly demanding to maintain. I also wonder if a list on an archivists' listserv would garner different reactions and methods. More interesting was the claim that they only enter "stable" sites. Sort of an oxymoron. From reading between the lines, I gather that these are sites they feel have a lower possibility for changing URL, but I'm not sure. If there's one thing I've learned in my time at SILS (there's more than one thing, but I'll keep it to one for now), it's that digital objects are anything but stable and that it's best to never think of them as such.


On a related note, many also talked about collecting e-books and e-prints versus the entire website. This must be a difficult decision when you're looking at a document that you formerly collected in print, which is now available online. Many still seem to look for the online equivalent of books or pamphlets for pulling into their repositories. Drawing this distinction allows repositories to draw boundaries in the boundary-less web, but it also may mean they're missing valuable information that has been converted to a "non-book" format.

Again, these are just my thoughts and interpretations of the conversations from the forum and shouldn't be taken as a complete synopsis!

Thursday, November 20, 2008

November 20, 2008

Talk about a neat and shiny tool. Julien Masanes forwarded this link to a web archiving listserv I've subscribed to. It is definitely worth the time to watch the video:

http://uwnews.washington.edu/ni/article.asp?articleID=45255

Zoetrope looks like a fascinating way to actually access harvested web material. Developed by University of Washington and Adobe, it allows you to compare a page over time, multiple pages over time, or even sections of multiple pages (using "lenses") over time. It will even graph changes in a numerical entity (think product prices or temperatures). Although it harvests sites VERY often, I would hope it would still be useful for a site harvested only monthly or bimonthly. They mention eventually using the Internet Archive as a source for harvested sites, which may mean it can read and display .arc files. Even better - they plan to make it available for free.