Instead of churning out computer science papers for students to turn in as their own work, a machine learning tool is helping faculty members and students discover learning resources that they may not have found any other way. Over the last year and a half, a team at Pennsylvania State University has been working on this technology in a quest to banish expensive textbooks and make the process of finding open educational resources less time-consuming.
"A big part of that was experimentation around to what degree can we teach the machine to understand the keywords we're providing, and then also match that with educational resources that would be relevant," said Kyle Bowen, director of education technology services at Pennsylvania State University
The software tool BBookX searches through open education resource repositories to find materials that faculty members may want to mix into their course content. Faculty members can look for specific articles to share with their students outside the confines of a book or mash up a bunch of articles into a new textbook. The technology looks a bit like Google in its search options and feels a bit like Netflix, where you can give a thumbs up or down to resources — but it doesn't follow either of those models closely.
A faculty member using this tool would first create the shell of a textbook and a table of contents, which includes chapter titles with descriptions of topics that would be covered. In these descriptions, the author can type in key words, phrases and learning objectives. Based on these key words, the software runs a search in open educational resource repositories and returns a list of potentially relevant articles.
The faculty member can then choose whether to keep each of those articles. By doing another search with the same keywords, more relevant results are given because the software's algorithms have learned something about what the user wants.
After the articles have been accepted, the software compiles them by topic in a way that's easily mixable. The faculty member can then add in case studies, review questions and other personal material to round out the book. The final book can be posted in the learning management system for students to access.
By using this software, faculty members can discover resources that they may not have normally found in search engines because of their own bias or content expertise. This summer, Bart Pursel, faculty programs coordinator at Pennsylvania State University — who teaches a course on information, people and technology — created a 15-chapter textbook for the course with BBookX.
His search for materials turned up articles about Belgian inventor Paul Otlet, who created an analog search engine in the 1900s with millions of facts about the world catalogued on index cards. For a fee, people could mail in their questions, and Otlet's team would look for the answer among the cards and send it back. While he didn't know about this man's invention before, Pursel made his work a narrative thread throughout the course as students and professors considered what challenges Otlet faced then and how they relate to today's databases, search engines and indexes.
"It will still surface interesting things that instructors maybe haven't thought of or haven't seen before that can then make an interesting contribution to their course," Pursel said.
The education technology services team will continue piloting the tool with different faculty members as team members look for ways to refine the technology and discover possible applications. And they'll be putting it in the hands of students who can compile their own learning resources.
