By: Audrey Malagon, Batten Associate Professor of Mathematics, Virginia Wesleyan University, @malagonmath

Due to COVID-19, colleges & universities moved quickly last spring to remote teaching and learning. Many of us used online learning tools for the first time or in new and unfamiliar ways. Even during less turbulent times, we are increasingly relying on web-based tools to deliver content, perform assessments, and collect data on student learning. The Distributed Open Education Network (Doenet) project is a collaborative effort that combines innovations of previous online mathematics teaching and assessment endeavours. Doenet combines instructor designed content, free student access, and educational research tools. Here Duane Nykamp (University of Minnesota), the principal investigator on the project, tells us more about the platform they are designing.

Tell us about the philosophy and goals of Doenet.
Our philosophy is that education should equip students to effectively engage with the web as a whole, rather than restricting their exploration to content provided by the instructor and isolated within a learning management system. By encouraging authors and instructors to place their content on the open web and then following a learner's activity as they explore multiple sites, Doenet can collect anonymized data (with permission) that can be analyzed to evaluate the impact of content. The ultimate goal is to discover effective content and learning strategies, which will enable Doenet to guide learners to content that will best facilitate their learning.

Doenet’s tools will allow content creators to integrate interactive content with text, video, and assessments. We also lay the groundwork for converting materials into accessible formats compatible with assistive technology.

Where did the idea for Doenet come from, and how did you form a team of collaborators?
The collaboration that spawned Doenet started when Mike Weimerskirch introduced the teams behind Ximera (Jim Fowler and Bart Snapp) and Math Insight (Duane Nykamp) realizing they shared both educational goals and philosophy.  The core Doenet team has grown to span three institutions.  The University of Minnesota group is Duane Nykamp, Mike Weimerskirch, Jonathan Rogness, Kevin Charles, and Vinitha Konduru; Stan Pride played an integral role in the early stages of development.   Mathematics professors Jim Fowler and Bart Snapp make up the Ohio State University contingent.  Matt Thomas, Ithaca College, provides the undergraduate mathematics education research expertise.

How will higher education be better off as a result of Doenet?
One impact will be reducing the financial burden on students. We openly license Doenet’s source code as well as the content and learning experiments that we develop. When relying on publishers for content and learning tools, students can find themselves in a course where the only option for completing coursework is through purchasing access to publishers’ tools.  By creating and encouraging the development of open, high-quality online content, we can give instructors options that will save students a considerable amount of money.

Universities benefit by keeping the knowledge for assessment and the data on student interactions inside the academy instead of under the control of commercial entities. Rather than asking students to pay commercial third parties to design their assessments and grade their work, we enable faculty to author engaging questions and activities, many of which can be autograded. By ensuring that the knowledge for how to perform assessment remains in the academy, we help universities retain their roles as a source of trusted credentials.

What have you learned so far in this project? What’s the biggest change or adjustment you’ve had to make?
One of the first projects we have been developing for the Doenet platform is DoenetML, the Doenet markup language. DoenetML, based on PreTeXt, is a language for authoring richly interactive content that can be displayed and automatically graded in a web browser, even if one's device is offline. One challenge has been to design a system that, on one hand, can infer an author's likely intent from simple descriptions, but, on the other hand, has a rich semantic vocabulary for designing complex interactive experiences. We have had to rewrite core pieces of the DoenetML software multiple times as we gained experience with how an author might use the system, and DoenetML evolved significantly from our original expectations.

Balancing privacy and security concerns with the need to track students across websites is a continuing challenge, especially given how bad actors can exploit personal data. As web browser vendors change their policies regarding permissible tracking mechanisms, our technology for tracking students across websites has had to evolve as well, in consultation with security experts and community members. Our solution centers around putting individuals in control of sharing their own data, and we hope to build confidence through complete transparency of how the data is anonymized, aggregated and shared to benefit the entire community.

Learn more about NSF DUE 1915363

Full Project Name: Collaborative Research: Empowering faculty to run online learning experiments
Abstract: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1915363
Project Contact: Dr. Duane Nykamp, nykamp@umn.edu

*Responses in this blog were edited for length and clarity.