When I first came to MPOW, our group was piloting search technology that paired a video in RealPlayer side-by-side with a text transcript and then highlighted the words as the video played. It worked, but there was a significant bottleneck at the transcription stage.

To prepare the transcripts, we had to ship our videos on DVD to the vendor (who shall go unnamed) for them to complete the transcription which then took weeks or even months. The long-awaited end result was an XML document with each word and paragraph timecode-tagged. They told us they were using sophisticated automatic voice recognition technology and that the results then had to be proofread by humans. My suspicion is that most of the transcribing was people-powered. In any case, it was expensive and time-consuming to the point where the video content might be dated by the time it was finally available to the user.  It just didn’t seem to be scalable and I ultimately decided it wasn’t worth it.

In the past year, I’ve seen a couple of video-search platforms that do similar things but probably much better.  They’re still expensive though.

I’m anxious to see how well Google’s voice-to-text extraction works. I don’t hold lofty expectations since much of our content will have lots of medical terminology.  If it does work, the next step would be to embed the YouTube viewer into DSpace’s XMLUI and combine the searches somehow.

Indeed, one of Elsevier’s prototypes seems more fitting for high school or undergrad classroom discussions, especially if they were to take into account Lagendijk’s points about context.

Credit: RT @atmire: RT @jimtill: #OpenAccess Elsevier going the wrong way: http://bit.ly/D16Am

At MPOW we’re hoping to migrate the library’s static HTML web site to the Drupal CMS. We’ve been using Drupal successfully for a year and a half for an offsite affiliate project and for a SEPA grant-funded project.  We have access to develop the current site using .NET when we want, but we don’t have the resources to do any significant development. Lack of PHP (and MySQL) support on campus is the only thing that has kept us from making the jump to Drupal.  While we have an ample ‘Nix VPS to host it on, I’m concerned that we don’t have the resources to maintain and update Apache or the PHP and MySQL frameworks.

Our current Drupal projects are hosted on inexpensive shared hosting which has been mostly sufficient for what we need including shell access, SVN, CVS, and Apache .htaccess capability. I don’t have to worry about updates to Apache, PHP, MySQL, or the OS.  If we were to host the library’s web site offsite as well, I’d want the best of both worlds: reliable support, a managed environment, dedicated RAM and processor cycles, but also full access to Apache’s httpd.conf.  Pair Networks is attractive for all these reasons. Also, our current host doesn’t support Java/Tomcat/Jetty which means we’d still have to host Solr separately on campus and manage access to it from other applications. Tomcat or Jetty support on the same host would be nice, but it’s not a deal-breaker.

Why do we want to migrate to Drupal? For many of the same reasons hundreds of other libraries and thousands of organizations do:

• a huge, and growing, user community,
• an overwhelming number of user-contributed themes and modules that extend Drupal’s core functionality,
• Drupal’s flexibility for modeling and displaying content,
• Drupal’s taxonomy and tagging support.

But these are all means to the end. In the end, we want our web site to deliver information well and make it easy for people to find what they want or learn how to get what they want.  David Lee King, Topeka’s Digital Branch Manager, recently offered 10 reminders that we strongly identify with.

Based on our experience with Drupal so far and conversations we’ve been having for a long time, here are some things we want to do.

• Make it easier and more enjoyable for our staff to post or update content to the web site.
• Improve how we provide news about library resources and services. Patrons can subscribe by email or RSS to various “channels”, e.g. type of content, subject, author.
• Advertise “featured” content and give our “marquee” an extreme makeover that showcases what’s happening.
• Create subject guides and more audience-oriented channels of content without unnecessarily duplicating content. Example, “you’re into pathology? Here are the best available resources for you. Oh, and here’s what’s new in PubMed and here’s a librarian to contact. In the hospital? Sure. Here’s the set of online clinical resources, image collections, and what we have on the shelves.”
• Integrate metadata from Voyager, the institutional repository (DSpace), and Serials Solutions using the Extensible Catalog Drupal toolkit to create a better discovery layer across all our resources, i.e. open up the silos.… And to put it into context with the rest of our content.
• Integrate content from other sites.
• Manage FAQ’s.
• Improve library instruction, i.e. “where-to-find, how-to-use”, without duplicating content.
• Encourage conversation and feedback by giving patrons the ability to comment on the web site or in whatever social network(s) rule the day.
• Distribute content between multiple “affiliated” sites, between the Library web site and project/partner sites.
• Support mobile devices.

We’re confident we can accomplish them with Drupal (and yes, Solr) in a relatively short amount of time.

ApacheSolr provides search and faceting.

Better still, HILCC is licensed under Creative Commons.

This reminded me of the presentation given by someone from Georgia Tech libraries a couple of years ago at Open Repositories 2007. Nebraska, like GT, is taking a holistic approach to providing repository and publishing services. These are the best approaches I’ve seen toward making repositories and “digital libraries” successful.

"Do It For Me" instead of DIY

Find archivable articles first, then get author’s permission. "Works 90% of the time"

Author’s version of a manuscript shouldn’t look like a rough-draft manuscript. It should look professional and closely resemble the publisher’s final copy.

Self-archiving authors fail to include important metadata (abstract, citation, co-authors, date, copyright)

Adobe InDesign for typesetting and formatting

Usage Reporting

Bepress generates monthly download reports and sends emails to each author. "Cannot overestimate the power of those messages." Very valuable for promoting repository value to faculty. Library hears immediately from faculty who don’t receive their email reports.

Promoting

Solicit or place links to: Wikipedia, Online books page, Worldcat, subject websites

Actively solicit and publish original material

Among the most popular content

e.g. open-access dissertations. Dissertations mandated to go to ProQuest, but also encouraged to go into repository. Open-access versions downloaded 60x more.

e.g. book-length works: the otherwise unpublishable (too long, too esoteric, too strange). Dictionary of Invertebrate Zoology cancelled by university press just before printing. Published online 9/2005. Hopi Nation published over 25 years by various presses. Published online 9/2008. 523 downloads in first 5 weeks.

Tenure Concerns

Surprisingly, more senior faculty are higher on uptake. Junior faculty are more concerned about were they published and are suspicious of open access publishing.

Start with the older faculty since they’re most concerned with what will happen to their stuff

Benefits

Most recruiting is word-of-mouth

Faculty come to the library first for their publishing needs

from Increasing Use and Content Through Creative Service-Repository Bundling, Paul Royster.

It seemed simple enough.  My first thought was to pull an RSS feed from DSpace into Yahoo! Pipes or a similar service. Then I would deploy a javascript widget from there.

Problem #1: Ideally you don’t want your RSS feeds spouting 80 items starting back in the 1990’s.

Problem #2: By default, the data in the feeds isn’t rich enough.  I need journal, volume, issue, etc.

My next go at it was to use DSpace’s OAI-PMH provider. This makes more sense than RSS. The data is rich, it’s all there, and I can specify sets or date ranges using Collection handles.

So what’s the drawback? The default metadata format for OAI in DSpace is oai_dc, i.e. unqualified Dublin Core, making it nigh impossible to distinguish published date from deposited date and publication source from handle URI.  This would exponentially complicate my simple little Pipe and make my brain hurt…alot.

The solution? Apply a Metadata Crosswalk Plugin for exposing QDC in OAI.  DSpace 1.5 even includes a preliminary crosswalk for this.  One more reason to expedite moving to 1.5.x.  We’re almost there anyway.*

*We’re still running DSpace 1.4 in production which does have a crosswalk plugin implementation known as XSLT Crosswalk but I’d rather devote my time to the upgrade.

Lucky for you and me there is very good help available. Here’s what I used:

1. Jump to Step 4 of Rails Deployment and Installation at IT.Signals. You’ll need to customize your configuration with whatever ports and IP or server address you have. I also needed to create the tmp/pids directory inside my Bibapp installation directory. Replace [path to]/testapp with [path to]/[bibapp] as appropriate. Replace environment: production with environment: development if that’s how you roll.
2. Save your new mongrel_cluster.yml to the Bibapp config directory. Start up mongrel_rails cluster to make sure its working on all the ports you expect. Stop when you get to Step 5. We need to setup the cluster so that it will be restarted when the server is rebooted. At this point, we’re going to deviate from these instructions, but you may want to return to them later if you want load balancing across your cluster. Refer to the Nginx section in Step #5.
3. Copy the mongrel_cluster Recipe kindly provided by Joyent and paste it into a new XML document named mongrel_cluster.xml. As instructed, customize the <instance/> properties for your Bibapp install. Save this file to /var/svc/manifest/network/mongrel/ on your server. This file is known as a service manifest to SMF. Unix/Linux use etc/init.d and rc scripts for defining and administering services. Solaris 10 replaced this with a new framework known as SMF. SMF is well documented on Sun’s blog. The SMF Manifest Cheatsheet is another good reference to the manifest properties.
4. Jump to Step 3.4 Create service manifest of Convert legacy_run service without reboot. Run

      1: svccfg import [path to mongrel_cluster service manifest]

5. If successful cd to [bibapp] and proceed to Step 4 for enabling your newly installed mongrel_cluster service:

      1: svcadm enable mongrel/cluster:bibapp

As mentioned earlier, once you have your cluster running it makes sense to do some load balancing for your production environment.

Good luck!

This edition of E2 almost makes me wish we’d moved to Portland 10 years ago when we were considering it.  What’s most inspiring is that the city commissioners, mayor, and other spearheaders interviewed don’t just talk about emissions and traffic, but about reviving the city while also drawing boundaries to preserve farmland and forests.

I get the same stir of inspiration from Tom and Christine Sine.

Is any of this possible in our midwestern sprawl-addicted stateline-divided city?

Having been a frequent reader of Deepak Singh’s business|bytes|genes|molecules blog over the last year, I was interested to see him hired by Amazon as business development manager of Amazon Web Services.  On December 04, 2008, Deepak posted to the AWS blog the announcement of Public Data Sets on AWS.  PDS on AWS is a data sharing experiment that takes advantage of Amazon’s in-the-cloud storage and computing services.

Just two weeks after Deepak’s post, Clint Boulton at eWeek confirmed that Google had axed its own Research Datasets project along with other projects of questionable value to Google’s bottom line.  While sharing data across the web, publicly or not, will surely become more common among researchers, milking copious amounts of ad revenue from that sharing is less likely.

The storage and computation of large datasets appears to be more in line with the AWS business model and perhaps Amazon has the lead on scalable architecture to support cloud computing.   Even if large numbers of researchers and research projects store and crunch their data on the web, that in itself won’t score big in the social web scene.  Programmers, analysts, and machines are more likely to be interfacing directly with the data than are the research investigators themselves.

It’s yet to be seen what Microsoft’s strategy for data storage might be in the recently released Azure platform, but they obviously have eyes on the educational and research markets.  Products like live@edu and SharePoint are increasing Microsoft’s reach into the academic computing world.

The Microsoft Research group quietly released a beta version it’s own repository software, running on .NET and SQL Server of course, but this isn’t just a reformulation of Dspace using Microsoft ingredients:

“The platform focuses on the management of research assets-such as people, papers, lectures, workflows, data, and tags-as well as the semantic relationships between them.”

Sounds like they’re paying attention.  And they’re beginning to appeal in ecumenical fashion to the larger research community by offering things like OfficeSWORD and taking part in discussions about open research repositories.

What is most interesting from the perspectives of the library and the university’s research office is how these services will redefine our notion of the “institutional repository”.  On one hand, many IT services such as web hosting and email have been commoditized to the point that institutions, especially smaller publicly-funded campuses, are unable to resist the cost savings and agility that come from hosted services like live@edu.  Why not commit fully to the .NET architecture and have your institutional repository software and data hosted on Azure as well?

On the other hand, why not take advantage of AWS’ flexibility and scalability for storing data or running our repository application?

Regardless of the platform(s) we choose, our notion of “institutional repository” is going to be stretched as we want to aggregate data and services from multiple platforms.  How will our Dspace service reflect our data stored in AWS?  The building blocks are already in place to support more complex relationships between our repositories, services, and data.  The time has finally come to put them to work.