CSAIL Research Abstracts - 2005 link to http://publications.csail.mit.edu/abstracts/abstracts05/index.html link to http://www.csail.mit.edu
bullet Introduction bullet Architecture, Systems
& Networks
bullet Language, Learning,
Vision & Graphics
bullet Physical, Biological
& Social Systems
bullet Theory bullet

horizontal line

The Classroom Learning Partner: Promoting Meaningful Instructor-Student Interactions in Large Classes

 Kimberle Koile & Howard Shrobe


To develop innovative technology to improve student experiences and learning in large classes by increasing interaction between instructors and students.


Personal interaction between instructor and student in large classes is almost impossible.  How do we increase that interaction, so that classes become more a two-way conversation between instructor and students?  One way is to give students the ability to engage in hands-on activities that yield immediate feedback through interaction with instructors and peers [1]. This technique, termed formative assessment [2], has proven successful in large and small classes.  In large classrooms that employ a wireless polling system called Personal Response System, or PRS, for example, students use a transmitter to submit answers to multiple-choice, true and false, or matching questions.  The results are tabulated and displayed in the form of a histogram on the instructor’s computer. A system such as PRS provides a way for students to communicate their misunderstandings to an instructor.  Instructors, however, are limited to asking questions having pre-existing sets of possible answers, i.e., close-ended questions, which assess recognition rather than recall.

In small classes, instructors can engage the students in a wider variety of in-class exercises than in large classes, since an instructor only has to evaluate a small number of answers.  Students can work problems at a blackboard, on paper, or using Tablet-PC-based systems (e.g., [3]).   Can this technique be used in a large classroom, e.g., with 200 students, where the logistics of managing 200 student answers could easily overwhelm an instructor?


The Classroom Learning Partner (CLP) will support in-class exercises in a large class, while also enabling instructors to use the wide variety of exercises possible in small classes.  The key idea:  Aggregate student solutions into a small number of equivalence classes and present the summary information to the instructor, e.g., in the form of a histogram. 

CLP will use as its framework an existing Tablet-PC-based presentation system,  Classroom Presenter [4], which supports student wireless submission of digital ink answers to in-class exercises. Using the Classroom Presenter system, an instructor lectures and annotates slides with digital ink. The slides and ink are displayed simultaneously on a large screen and on students’ Tablet PCs.1 When an instructor displays a slide containing an exercise, the students work the exercise using digital ink on their tablets, then anonymously submit their answers to the instructor via a wireless network. Using Classroom Presenter in this way works well in classes of size eight or smaller, as instructors are easily overwhelmed by more than eight solutions [5]. We will extend Classroom Presenter so that in-class exercises can be used as a teaching method in large classes: We are developing software that aggregates student answers into a small number of equivalence classes by comparing those answers to instructor-specified right answers and wrong answers, and/or by clustering student answers.  The software will present the results to the instructor (and students), e.g., in the form of a histogram and representative answers.


With CLP, we expect that instructor and student will interact more often and in a more meaningful way than has been possible in large classes to date.  Through the use of in-class exercises, they will be able to interact via  formative assessment, which increases student learning by enabling instructors to (1) get feedback about student learning during the teaching of a topic, when they can modify their explanations to fit student misunderstandings; (2) assess student learning with less effort and time than with traditional assessment methods such as exams; (3) explicitly show students that they care about their students’ learning; (4) increase student confidence and rapport by allowing students to be active participants in their own learning; and (5) help students feel comfortable by letting them submit answers anonymously. 

Current Status

We are in the beginning stages of the Classroom Learning Partner project.  We have collected examples of in-class exercises and student answers for MIT's introductory undergraduate computer science class (6.001) run without technology, and used that data to create exercise and answer ontologies.  We anticipate aggregating answers in the form of text (e.g., filling in decodings for bit strings), text and markings (e.g., labeling boxes in a flow diagram), and sketches (e.g., drawing a search tree).  We have organized the system architecture into three components and begun work on each:  (1)  An instructor authoring tool, with which an instructor will prepare lecture slides that include exercises annotated with answer type, correct answers, and incorrect answers if known; (2) a digital ink to structured object translator, which will take students' submitted answers in the form of digital ink and translate each into an object of the specified answer type (thereby providing semantics for the ink); and (3) an aggregator, which will  group student answers into equivalence classes, summarize the results, and present a histogram and representative answers to the instructor, who can then display the results to the class.  In a later version, we imagine adding a component that sends individual replies to students based on comparing their answers to right and wrong answers, offering explanations of misunderstood concepts that are related to particular wrong answers (e.g., a la intelligent tutoring systems).

We plan to deploy a prototype in 6.001 recitations during academic year 2005-2006, and a second version in 6.001 lecture (with a typical enrollment of 350) the following year. We also will disseminate our system to the Classroom Presenter community and make it available to others via the web.

1 We will allow instructors to disable ink transmission if they prefer that students take their own notes.

Research Support

This research is supported by the MIT iCampus program.


[1]  R.R. Hake. Interactive-Engagement versus Traditional Methods: A Six-Thousand Student Survey of Mechanics Test Data forIntroductory Physics Courses, American Journal of Physics, 66(1), pp. 64-74.

[2]  T.A. Angelo and K.P. Cross. Classroom Assessment Techniques: A Handbook for College Teachers, Jossey-Bass, San Francisco, 1993.

[3]  B. Simon, R. Anderson, C. Hoyer, and J. Su.   Preliminary Experiments with a Tablet PC Based System to Support Active Learning in Computer Science Courses. In 9th Annual Conference on Innovation and Technology in Computer Science Education (ITICE), 2004.

[4]  R. Anderson, R. Anderson, B. Simon, S. Wolfman, T. VanDeGrift, K. Yasuhara,  Experiences with a Tablet PC Based Lecture Presentation System in Computer Science Courses, SIGCSE, 2004.

[5]  R. Anderson. personal communication, 2005.

horizontal line

MIT logo Computer Science and Artificial Intelligence Laboratory (CSAIL)
The Stata Center, Building 32 - 32 Vassar Street - Cambridge, MA 02139 - USA
tel:+1-617-253-0073 - publications@csail.mit.edu
(Note: On July 1, 2003, the AI Lab and LCS merged to form CSAIL.)