Visualizing Peer Learning

Daniel Gruber, Patricia Goldweic, Philip Burns, and Bill Parod


Yellowdig Peer Learning

Screen Shot 2015-12-09 at 11.48.25 AM

Integrating Social Media to Elevate Engagement

Yellowdig is a Learning App piloted at Northwestern during the 2015-2016 academic year. The pilot was part of a Digital / Online Grant awarded to Medill Assistant Professor Dan Gruber.  Integrated in Canvas, Yellowdig allows students to easily share relevant resources on class topics within the Canvas environment. It is dynamic and engaging – resembling many of the social media platforms that students use in other aspects of their lives, but does not impose on those personal social networks. Yellowdig has been used by over 100 courses during the pilot.

The success of the Pilot resulted in Northwestern renewing the Yellowdig license for another three years.

Mechanisms to Enhance Learning and Teaching


“How do we use technology to help students and faculty learn more, specifically by giving them data and guidance and inspiration?”

Based on Fitbit CEO James Park’s vision for his company to help people get healthier and more active

Discussion Network Analysis

YellowdigVizPrototype1_detail_labels YellowdigVizPrototype1_labelsb

Our first prototype used a network graph model to represent Yellowdig pins (dark green diamonds), comments on those pins (light green circles), and users  (blue circles) who author each pin and comment. One can zoom in or out of the graph to see the overall structure (right) or a detail of the network (left).

Hovering on any node displays information about that node. This was useful for seeing overall structure as well as smaller, localized discussion structures.

Natural Language Processing

YellowdigVizBrands     YellowdigVizBrandsByWeekNoLegend

What brand names interest students? (left)

In each of the classes I teach, there is an important part called “In the News” which is intended to address the learning goal of linking academic frameworks to current events.We used Natural Language Processing such as Named Entity Recognition to identify companies and brands of interest to students in the course.

Which brand names generate the most comments by week? (right)

We looked at company name frequency over time. This graph shows prevalent companies by week (columns) in pins (top) and comments (bottom row). It was interesting to see how “pinning” of Nike (pink) in the 2nd week led to followup comments about Nike.

What analytics do students want?

garage_pic_w_students garage_logo

While reflecting on how to leverage both a network model (for discussion structure) and natural language processing (to surface discussion topics), we took a step back and arranged a session with the class at The Garage to gain their insights and learn what they would find most useful. The brainstorming was terrific – the students’ ideas were unburdened by implementation and offered direct insight into what visualizations are most useful for class discussion and why.

Some of their suggestions:

    • Show faces and images in the graph rather than symbols
    • Highlight sub-networks of specific discussions
    • Search and filter by topics, brands, and tags
    • Display social network graphs
    • Offer “connectedness” and engagement scoring
    • Display word clouds of frequent topics, brands, tags
    • Include links to pinned articles
    • Animate discussion network growth

Yellowdig Visualization – Spring 2016

Building on our prototype and great ideas from the student brainstorming session, the latest version of the Yellowdig Visualization App highlights user selected subgraphs, showing the selected node, its author, related comments, and their authors. This allows focus on specific discussions and reveals the associated “social” graph of the discussion’s participants.

The new visualization also echos the student’s Yellowdig board experience by using images from pinned articles and user profiles to reinforce connection with the Yellowdig “front end”.

Natural Language Processing (NLP) is revealed in the faculty view. NLP results are attached to their pins and comments allowing faculty to explore the graph by referenced company names and keywords. Faculty can use these terms as well as Yellowdig tags and students’ names to explore the graph from browsing lists on the right of the display.

The following video demonstrates these features as well as time-lapsed views of network growth by day and week.


The App integrates directly in Canvas for use by faculty and students and is available to any Northwestern courses using Yellowdig.We look forward to more great ideas from our faculty and students as this work evolves with diverse learning communities at Northwestern.

For more information, please contact Dan Gruber or Bill Parod, or visit the articles below which also describe this work.

Blog post: Three things you need to know about Yellowdig

Digital Learning | Learning Analytics | Innovations at Northwestern

Daniel-Gruber-750x1125Daniel Gruber is an Assistant Professor at the Medill School of Journalism, Media, Integrated Marketing Communications.  He is also an Assistant Professor of Management & Organizations in the Kellogg School of Management (Courtesy).  Before joining the Northwestern faculty in 2009, Dan completed his Ph.D. at the University of Michigan’s Ross School of Business.

Dan’s research contributions are at the intersection of organization theory and the areas of academic excellence across the Medill School.  He advances organization theory in the context of journalism, media, and integrated marketing communications.  Within the organization theory literature, Dan offers a unique perspective by studying news organizations, the media industry, and integrated marketing communications to conduct research on important topics such as sensemaking, disintermediation, and managing the unexpected. Email:


patricia_goldweicPatricia Goldweic is a Senior Software Engineer in Academic & Research Technologies, Northwestern Information Technology. Patricia is an experienced software developer and is known for her work over many years in creating learning tools and systems for teaching and learning in a variety of contexts and platforms. Her software design and implementation for this Yellowdig Visualization App is the most recent but one of several  applications she has developed for Canvas at Northwestern. Email:


phil_burnsPhilip Burns is a Senior Software Engineer in Academic & Research Technologies, Northwestern Information Technology. Phil has extensive experience in text processing and analysis. This work often involves collaborations with faculty on funded projects in the “digital humanities”. It also forms the text-analytic infrastructure for many of Northwestern’s teaching and learning applications. These services were used and enhanced for this project,  providing Named Entity Recognition and Keyword Extraction for Yellowdig pins and comments. Email:


BillParod_001Bill Parod is a Software Architect in Academic & Research Technologies, Northwestern Information Technology. Bill has worked on wide range of software development projects over many years. His recent focus has been on systems integration for our Learning Management System (Canvas), architecture and applications design for learning analytics, and consulting on grant-funded software development projects in various research domains. Bill oversees the technical planning for this and other Learning Analytics projects at Northwestern. Email:














Leave a Reply

Your email address will not be published. Required fields are marked *