Working in Memphis I am fortunate to be in the same city as the ADL Workforce Co-lab located at the FedEx Institute of Technology on the campus of The University of Memphis. A few years ago the ADL Workforce Co-Lab held its first eLEarning Summit. There were guests and speakers from all over the world, but primarily many of us in from area companies were in attendance.
At the end of the Summit, Dr. Trey Martindale suggested we start a monthly “brown bag lunch” to discuss eLearning in the workforce. From that first lunch it has grown into the Workforce Elearning Resource Network (WERN).
Last week I attended our monthly meeting and thoroughly enjoyed a presentation by Dr. Xiang Hu. He gave us a demonstration of a project they are currently working on with the DoD in Intelligent Tutoring Systems (ITS). If you’re not familiar with ITS they can be very expensive and are typically only deployed in the Education, Medical, and Military industries. This presentation demonstrated new technology that is lighter, portable, and can run on any Learning Management System (LMS).
The most interesting thing about this ITS – AutoTutorLite is the Latent Symantic Algorythm. I just learned of it yesterday and am not completely confident I fully understand. Essentially, LSA pulls the “meaning” of a phrase or string of words and generates a dynamic response based on learner’s input. An analogy would be if you type the word “Doctor” in Google, the search engine dynamically builds a search list for you, yet it won’t recognize you may also want to see “Physicians.” From what I understand this is called ‘1st order of co-occurrence.’ 2nd order of co-occurrence would be for example if type the word “House,” it would return words like “Home,” “Kitchen,” “Dining Room” which is where the power of LSA comes in as it ‘thinks’ of what you are really wanting to say. Going even deeper into LSA’s co-occurrences, if I wanted to type “I feel sick” the algorythm would pull the meaning out of my phrase and not only return the words, ‘Doctor’ or ‘Physician’ it can generate a question like “What are your symptoms?” Amazing!
How this applies to Elearning is the ability to script hints and deep level questioning to learner’s responses. A concern was raised regarding spell check and how some learners do not type very well. We were quickly shown an example of a very poorly written sentence with misspellings and the ITS pulled the “meaning” of that phrase and understood what was being entered!
AutoTutorLite was developed from its predecessor AutoTutor to be lighter, portable, and run on any LMS. It’s published in the .swf (Flash) format and depending on the script will be less 300k in size. I can think of several opportunities where this could be used in the workforce and specifically in our industry. It’s more geared toward soft skills than instructional procedural training such as: How to deal with irate customers, Selling skills, Help Desk, Commercial Sales, etc.
Dr. Hu discussed the possibility of conducting a Proof of Concept for this application in the workforce. It may be possible to acquire an intern from his department and in turn deploy this technology on a very small scale to measure the impact.
Again, I’m not that confident that I understand that technology…yet 🙂