Text mining can deepen classroom engagement


UMSI Associate Professor Kevyn Collins-Thompson has been awarded a $50,000 grant by Digital Education & Innovation (DEI) at the University of Michigan to fund his project, "Michigan Education Text Analytics (META): Supporting Engaged Teaching and Learning with an Educational Text Mining Platform.” This project aims to develop an education-focused text mining platform that will improve engagement in teaching and learning among students, teachers and administrators.

In text mining, texts are analyzed using statistical or other methods in order to turn the text into data that can be analyzed in new and useful ways. Common examples of text mining applications include the filtering out of junk mail by e-mail servers and the analysis of typed survey responses to find common trends.

Although text mining also has educational applications, currently, education-focused text mining requires users to integrate a number of different tools in complicated ways. With the META project, Collins-Thompson aims to create perhaps the first user-friendly interface for educational text mining, one that could be easily integrated into applications, including other U-M education tools such as Canvas or Coursera. This tool will utilize student-generated text, which is continually being produced in the form of assignments, online discussion messages, course evaluations, and other formats. This abundant text will then be used to develop a reusable educational text mining pipeline, the META server.

Figure 1: User interface design prototype for course review summary, showing course aspects (central sectors) with examples of associated adjectives, plotted by relative frequency (bar height) and sentiment (green=positive, red=negative). (‘CourseBoard’ design by UMSI students Adam Levick & Florence Lee.)

Once META has been developed, Collins-Thompson will use the platform to develop two applications aimed at increasing the efficiency and engagement of both faculty and students at U-M and beyond. One application would provide a summarization service for end-of-course student reviews in order to increase their usefulness. Currently, the avalanche of comments instructors receive at semester’s end is often overwhelming, leading instructors to focus on numerical ratings instead of written feedback. This application would use META to analyze comments in student feedback and draw out central themes, increasing the likelihood that instructors will actually read and use the feedback.

The second application has the potential to increase the efficiency of instructors by making it easier for them to monitor the understanding of large groups of students. For example, an instructor might ask hundreds of students in a freshman course to write a response to a prompt in order to check their understanding of a concept. While reading each student’s response and identifying students’ knowledge gaps might take hours by hand, text summarization tools developed by META will allow instructors to access almost real-time summaries of students’ typed responses. This will enable instructors to address any lack of understanding directly during the class or lab.

This project will also involve UMSI Research Assistant Professor Christopher Brooks, who will serve as co-principal investigator.

Digital Education and Innovation (DEI) is an initiative focused on shaping the future of learning and redefining public residential education at the University of Michigan by opening up new opportunities and enabling learners at U-M and around the world to engage in personalized, lifelong learning.

Posted on November 13, 2015