IU Technology Architecture Lodge
Random and not so random thoughts from Raymond Yee, primarily on the scholarly and educational use of the Web, libraries, educational technology, and information management
Interop soup: MOA2, METS, IMS, RSS, OPML
Posted by Raymond Yee, 3/6/02 at 10:25:00 AM. FOR MORE UP-TO-DATE STUFF
http://iu.berkeley.edu/rdhyee/discuss/msgReader$849?mode=day&print-friendly=true
Rationale
I'm interested in submitting an article about the importance and practical state of interoperability among three types of systems as represented by their respective standards:
- METS (metadata encoding and transmission standard) -- successor to MOA2 (Making of America II) for digital libraries (http://www.loc.gov/standards/mets/ and http://sunsite.berkeley.edu/MOA2/)
- IMS Content Packaging and metadata standards for use in educational technology (including learning management systems, course authoring software) (http://www.imsproject.org/content/packaging/ and http://www.imsproject.org/metadata/index.html)
- XML standards popular on the web for news, content, and weblogging syndication (RSS 0.9x, RSS 1.0 -- and perhaps OPML) (http://groups.yahoo.com/group/rss-dev/files/specification.html http://backend.userland.com/rss092 and http://www.opml.org/ )
The three communities mentioned (digital libraries, educational technology, and "next-generation-web techies/early adopter users") have common interests but are working on different standards. I want to write about interoperability among these different standards, primarily from the viewpoint of practical implementation at this point. The primary scenario driving the article will that of someone who wants to make use of rich content from digital libraries, marked up METS format with educational technology software (learning management systems) and in web authoring (the user wants to easily grab hold of digital objects in the library and integrate them with his Web writing environment)
The article will be about the practical issues of interoperability, how to map the various standards to one another, and what software is available today to make use of these objects.
One goal is to transform one of the MOA2 sample objects in a very simple HTML page. Another is to try out XSL-FO to see whether I can get a PDF file out of the MOA2 XML. As I mentioned previously, we're very interested in interoperability among libraries, museums, and learning management systems. A proof-of-concept I want to develop is to interconvert MOA2, EAD, IMS XML formats. Surely, someone must have done these interop experiments....but I haven't found any mention of such work in my cursory web search.
What is METS?
I also spent some time studying the METS standard (which builds on MOA2). I didn't have a clear understanding of the relationship between the two. From what I understand, METS is a generalization and an updating of MOA2. Consequently METS is expressed in terms of an XML schema rather than the DTD of MOA2. However, it would be nice to see some HTML interface to a METS object -- are they out there?
WebCT IMS Content Import Utility
For over a year now, we have had the idea that we should be able to generate content in an IMS compliant fashion that can be imported into learning management systems. I'm currently working on creating a very simple test content package to import into WebCT through WebCT's Content Migration Utility. Questions I have include: how difficult is it to generate such material? Can we use Microsoft's LRN 3.0 Toolkit be used to generate materials for IMS? How hard will it be to write some prototype software that will take materials from our digital archives and send the materials right into WebCT? Does Blackboard have IMS compliant tools?
So what have I learned so far? About the WebCT tool, I was happy overall but clealy see the tool as in the early stages of development. There are some very important limitations:
- Top-most manifest must be a course manifest. "This limits the import functionality to creating courses, either empty courses or courses populated with tools and content. IMS-compliant content cannot be imported into existing courses." (The scenario I often consider is wanting to import content from digital repositories into existing courses without wiping out everything else.
- The use of this content migration tool is limited to WebCT administrators -- and not the instructors. The functionality of the tool is not encapsulated as a web service, limiting any dynamic behavior. As an instructor, I'd want to be able to browse to some useful resources (images, web pages) and then easily fire them off to my learning management system.
- I was pleasantly surprised that there is export functionality. (Some of us thought that WebCT might want to get people to move their materials into WebCT but would want to make it difficult to migrate content out.)
I look forward to trying out the migration tool. I'll start with trying to important an empty course -- and then move to trying to important a list of URLs. I'll also have to learn more about the WebCT interface so that I can better understand the strengths and limits of WebCT -- and how to figure out how data are conceptualized in WebCT. (For more info, take a look at this presentation on the WebCT toolkit.)
I also took a look at the Microsoft LRN Toolkit 3.0. I was envisioning a tool that would help me package up Microsoft Office documents for learning management systems. I was looking for ways to convert Word and Excel documents -- but instead found tools to convert Powerpoint and FrontPage documents. Hmmm.. The Toolkit includes an editor for creating IMS manifests (for content packaging and metadata). The most interesting part of me is the LRN Viewer: "The LRN Viewer generates a Dynamic HTML (DHTML) view of LRN content, adding a syllabus and navigation features, and provides a framework that supports the SCORM run-time environment. The LRN Viewer uses the Microsoft XML Parser with Extensible Stylesheet Language (XSL) to generate the DHTML view." (You might to see a powerpoint presentation by Microsoft on the LRN toolkit)
For more info on IMS content packaging and tools, take a look at an overview.
MOA2 to HTML
I have been continuing the MOA2 work that I wrote about on Friday. Specifically, I spent this afternoon learning more about XSL through writing a XSLT stylesheet to do a very simple rendition of MOA2 (Making of America II) objects. The Making of America II is a Digital Library Federation project to create a proposed digital library object standard by encoding defined descriptive, administrative and structural metadata, along with the primary content, inside a digital library object. There are some sample objects encoded in the MOA2 standard.
I've been interested in first learning ("grokking") the standard so that I learn how to convert MOA2 objects into other formats -- and vice versa. It's also been a good opportunity to dig into XML technology, specifically XSLT.
Consider the Patrick Breen diary. A tutorial was written based on a subset of the diary. I took the XML for the sample diary and applied the XSL stylesheet I just wrote (forgive the infelicities--I'm still a XSLT novice) to get the following output HTML. Please don't focus on the presentation in the HTML -- it wasn't meant to be visually attractive. Rather, I wrote the stylesheet to extract part of the content to present in one page and for me to understand the relationships among the various parts of the MOA2 standard. (Note that the stylesheet should work for the other MOA2 sample objects.)
Converting MOA2 to RSS 0.92
I've been continuing my prototyping of how digital objects from libraries can be shared and sent into other digital boxes. RSS is a a common XML format for the syndication of news and blogging items. Since Userland products (such as Manila and Radio) use RSS 0.92 (as opposed to RSS 1.0), I decided that for demonstration purposes, I would figure out how to convert MOA2 objects into RSS 0.92 so that the blogging and syndication tools in Radio/Manila can be used to manipulate the library digital object.
Here below is a rendition of a RSS 0.92 channel derived from an excerpt from the Patrick Breen diary at the Bancroft Library. The RSS file is computed from the original MOA2 file via an XSL transformation. One of the big problems of doing a MOA2 to RSS 0.92 mapping is that the MOA2 format is much richer format, containing lots of information with no corresponding "slot" in RSS 0.92. For the purposes of this demonstration, I ended up mapping a lot of the administrative and descriptive metadata in the MOA2 format -- not a totally satisfactory solution. Because RSS 1.0 has room for rich extensibility, it is perhaps a better format for the syndication of library objects. (I will have to look into using RSS 1.0). (We have always thought that if one wants to make resources from a digital library available in an easy to manipulate format, one must make it easy to also syndicate the metadata with the objects -- so that all the provenance information flows easily with the objects themselves. RSS 0.92 is not really geared for such a task -- but there are tools that use RSS 0.92 today.) (BTW, I've limited the display to only five items....)
| ||||||||||||||||||||||||||
What can you do with RSS 0.92?
Once you have an RSS 0.92 channel, you can use tools like Radio to easily aggregate these mateials and publish them to another source. Here's a screenshot of the Breen RSS channel in my Radio aggregator. Here's a screenshot of my adding a comment to one of the items. You can then see the final result on my Radio blog.
MOA2 to OPML
Since MOA2 is a hierarchical structure, another hierarchical structure might provide a better mapping than RSS. OPML is used by Userland for such things as outlines in Radio and directories and slides in Manila.
I've converted the MOA2 representations of the sample of the Breen diary and the Yoshiko Uchida scrapback into OPML. You can look at these objects now within a Manila site. For example, you can browse to the front cover of the scrapbook with this Manila directory.
For more information, look at the MOA2 to OPML XSLT file that I used to do the transformation, the resulting OPML for the Yoshiko Uchida diary, and the OPML generated via the W3C XSLT service.
First pass at MOA2 to IMS-CP
Later this morning: I'm almost there with a first pass at a mapping. I have converted the Uchida diary MOA2 object into an IMS package. Here's a screenshot that shows the Uchida diary as a learning object displayed in the Microsoft LRN 3.0 Viewer. Here's the same part of the diary displayed in the MOA2 viewer developed by the Library.
Here's the imsmanifest.xml that I've generated. If you want to look at interact with the scrapbook with the Microsoft LRN 3.0 viewer, download the viewer and open the imsmanifest.xml file through the LRN Toolpad->Open LRN Viewer button. If you read the imsmanifest.xml, you will notice that I have yet to do anything useful with the metadata in MOA2 (See MOA2 document for the Uchida diary). That's a euphemism -- I haven't translated any of the metadata yet! This is what I'm working on this afternoon....
BTW, here's the XSL file that I wrote to do the conversion.
To do: metadata mapping between MOA2 and IMS
The IMS CP packages I've generated so far have skipped over the metadata in the MOA2 file. I am trying to figure out how to map this metadata into IMS.
In all the other conversions from MOA2 (to RSS 0.92, OPML, sample HTML), I haven't thought too deeply about how to map all the metadata from MOA2 primarily because it was clear that such mpapings would be "lossy", and hence the challenge of preserving the information was outside of the scope of this round of prototyping. These formats don't have provisions for holding detailed metadata. Nor can they be extended to handle arbitrary metadata. (RSS 1.0 will be an interesting case of something different.)
IMS Content Packaging (IMS-CP) is different -- because there is a place for metadata, not only the recommended MD structure of IMS Learning Resource (Meta-Data) but any other metadata (represented by another schema that you want). In theory, one should be able to do a completely faithful, "lossless" mapping. But my first tact was not to do any metadata because the task of faithful mapping is a more complex one.
I can think of at least two strategies to move ahead. The most straightforward is to convert MOA2 metadata to GDM. Since GDM is based on MOA2, this metadata conversion should be fairly straightforward and still preserve the original metadata, keep all the metadata key from a library perspective. Another approach is to map GDM to IMS-MD. This is a bigger challenge because there is a translation between metadata of interest to the library community and that of interest to developers of learning objects and software systems. Will the best way be to map metadata tags or preserve the GDM metadata, map only the obvious corresponding elements?
But it's one I'll take on tomorrow (I forgot that this afternoon I need to work on other things!)
After I work through the MOA2 to IMS conversion tasks, I'll turn to looking at METS, the successor to MOA2.
Last update: Monday, July 21, 2003 at 6:03:16 PM.
This site is using the Vanilla Manila 1999 theme.
The opinions or statements expressed herein should not be taken as a position
of or endorsement by the University of California, Berkeley. Nor should the
opinions or statements expressed herein be taken as a position of or
endorsement of the University of California, Berkeley. Links on these pages to
commercial sites do not represent endorsement by the University of California
or its affiliates.