IMS Learning Information Services: What a Solution Looks Like

In an earlier post, I outlined the motivating pain that brought the working group members to the table. In this post, I’m going to list out the highlights of the solution we came up with to address that pain. Again, this post is focused mainly on the important but unsexy problems of SIS/LMS integration that IMS LIS was intended to address. I’ll get to some of the sexier implications in a future post.

As I mentioned in the previous post, the draft of the LIS specification is now complete and is circulating among IMS members for a few months before it is released for public review. We (meaning Oracle) have also done a fair bit of integration testing with several other LIS implementers, so we have a fairly good idea of how this is all going to work in the real world. 

Here are some highlights of what we think we have accomplished with LIS:

• Flexible course to course site mapping: If you recall from the previous post, one gap in the earlier spec that created all kinds of problems was that the class section data coming out of the SIS did not reflect the realities of how teachers and students want to view those sections in the LMS. We created a kind of semantic hint that now carries this information the LMS without dictating what the LMS must do with it. This should by itself prevent many of the problems that colleges face today.
• Near-real-time provisioning: LIS provides an event-driven integration mechanism, meaning that data about a student’s registration in a class can be sent immediately to the LMS upon registration. The student should get instant access. This capability is designed to work in tandem with batch methods, so if the two systems somehow get out of sync, they can be easily resynchronized.
• Higher assurance of out-of-the-box integration: This is happening in several ways. First, Oracle already has production-ready integration tested with Sakai (in partnership with Unicon) and Schools on Facebook (which Michael Staton guest blogged about), and we have started testing integration with Desire2Learn and Moodle (in partnership with Moodlerooms). We are ready to test with Blackboard, ANGEL, and others as soon as they are ready to come to the test floor. The best guarantee of interoperability is actual tested interoperability. Second, LIS requires implementers to provide WSDL in order to form a reliable service contract between the two systems. And finally, the working group is developing a conformance profile that implementers must adhere to in order to guarantee interoperability. The LIS specification tries to be forward-looking and provide enough flexibility that people can meet a variety of needs with it. The trade-off is that it is broad enough to be interpreted and used differently, which tends to break interoperability. Conformance profiles provide further constraints on how implementers agree to use the specification for a particular purpose. They don’t have to follow the conformance profile, but adopting schools can look to see whether they have in order to attain some assurance of interoperability.
• Grade reporting: The grade reporting capability is greatly improved in LIS. Oracle is only implementing final grade reporting in our first version of SAIP, but the spec allows very fine-grained reporting of grades. I’ll have more to say about this in a future post.
• Routing-friendly: Because LIS provides a SOAP-based binding, LIS messages can be routed using fairly standard middleware. This means, for example, that you could send courses from the English department to Moodle and courses from the math department to Desire2Learn. As far as I know, nobody has implemented this kind of routing yet, but it’s definitely on the road map for Oracle. 
• Integration with more apps: There are many applications used by colleges that need information about courses and students but that will probably never be enhanced to integrate with LIS web services. Wikis spring to mind, for example. LIS provides an LDAP binding, which means that applications that integrate with LDAP in a reasonably robust way (i.e., they can store groups as well as individuals in LDAP) can be enhanced to integrate with LIS fairly easily. LDAP integration also provides a bridge between LIS and identity management systems. Oracle’s SAIP does not yet support the LDAP binding but will support it in the near future.