Summer School

4th Annual PSLC LearnLab Summer School

Application Process Closed

• When: Monday, July 7, 2008 - Friday, July 11, 2008
• Where: Carnegie Mellon University, Pittsburgh, PA, USA
• Cost: Participation is free!
• Contact: Michael Bett -
• Important Dates:
  • The deadline for applications is Midnight April 2, 2008.
  • Admission decisions will be made by April 9, 2008.
  • Call For Applications

  Download Last Years Presentations

  The PSLC Summer School will be an intensive 1-week course on technology-enhanced learning experiments and building intelligent tutoring systems. The summer school will provide a conceptual background and considerable hands-on experience in designing, setting up, and running technology-enhanced learning experiments, as well as analyzing the data from those experiments in a technology supported manner.
  
  The summer school will last five days. Each day will include lectures, discussion sessions, and laboratory sessions where the participants will work on developing a small prototype experiment in an area of math, science, or language learning. The participants will use state-of-the-art tools including but not limited to the Cognitive Tutor Authoring Tools and other tools for course development, TuTalk tools for authoring natural language dialog, TagHelper tools for semi-automated coding of verbal data, and storing and mining of student interaction data.
  
  On the last day, student teams will present their accomplishments to the rest of the participants, followed by a "graduation" party. Participants will be expected to do some preparation before the summer schools starts.
  
  The summer school will be organized into three parallel tracks: tutor development (TD), /in vivo /experimentation (IV), and data mining (DM).  The tracks will overlap somewhat but will differ significantly with respect to the hands-on activities, which make up about half the summer school. Although you as participant will be assigned to one of the tracks, based on your preferences stated in the application, it will be possible to "shop around" - that is, participate in activities of tracks other than the one to which you have been "officially" assigned. Our primary concern is that the summer school will be a good learning experience for you.
  
  The summer schools involves intensive mentoring by PSLC researchers, which starts by e-mail before the summer school (in order to select a subject domain and task for the project, where appropriate) and continues during the summer school with a good amount of one-on-one time during the hands-on sessions. The mentors will be assigned based on interest as stated in the application. (All participants will have the opportunity to interact with all course instructors, but will interact more frequently with their designated mentor.)
  The following researchers will function as mentors:
  
  Ken Koedinger
  Kurt VanLehn
  Carolyn Rose
  Pamela Jordan
  Noboru Matsuda
  Vincent Aleven
  Bob Hausmann
  
  *The Three Tracks*
  
  First, let us describe the tracks in more detail.
  
  

  *IV track:* if you are in the "in vivo" track, your goal will be to design and create a prototype of an /in vivo/ experiment for one of the LearnLabs.  That is, you will identify a hypothesis about learning that you find interesting (perhaps one you were interested in already), and will work on designing one or more experimental treatments to test that hypothesis. (The control condition should be the corresponding activity in a standard LearnLab course.) You will create a prototype of the treatment and will work also on creating (prototypes of) assessments for measuring robust learning.  The experimental treatment must be a variation of  material taught in the standard LearnLab courses, so you will study the scope and sequence documents of the relevant course to find a place where the experiment would "fit". You may also spend some time familiarizing yourself with the course materials, which may include cognitive tutors or other on-line instruction. The prototype would preferably be made with one of the PSLC's enabling technologies (e.g., it could be a computer tutor made with CTAT, or a dialogue system made with TuTalk) but if it is more appropriate to use standard tools such as Powerpoint, that would be fine also.

  *DM track:* if you are in the data mining track, your goal will be to analyze an educational data set using data mining tools and methods. The data set could be one of the data sets currently in PSLC's Data Shop or you could bring your own. (Best to work this out beforehand with your mentor - see below.) A typical data set would be a detailed record of the interactions that students had with a computer tutor over a (possibly extended) period of time, but other data sets are welcome/interesting too, including verbal protocol data. You will identify a driving question that guides your analysis (quite possibly, one that you were interested in already), such as to understand how students' strategy use evolves over time, or what aspects of students' meta-cognitive abilities are related to learning. For verbal data, the driving question might be to find out how topic coverage in student tutor-dialogues relates to learning.  You will need to familiarize with the data, for example, by playing the student working through the activities that the data pertain to, operationalize your hypothesis into more detailed analysis questions, run the analyses with various tools (e.g., Taghelper, the Data Shop, or Excel) and algorithms (e.g., standard machine learning algorithms), interpret the results, perhaps recode the data, etc.

  *TD track:* in the tutor development track, your goal will be to implement a prototype computer-based tutor, using authoring tools developed by PSLC reseachers, such as CTAT (the Cognitive Tutor Authoring Tools) which supports the creation of intelligent tutoring systems, or TuTalk, which is used to develop tutorial dialogue systems that interact with students in natural language. Both CTAT and TuTalk have been designed for non-programmers. You will be able to use these tools even if you have no programming experience. Depending on your interest, your tutor might be related to a planned or possible experiment (perhaps an /in vivo/ experiment), or it might be related to a tutor development project that you are involved in or are planning to start up, or a course that you are teaching. CTAT-built tutors typically focus on multi-step problem solving as is often found in math, physics, or chemistry, but they are also being applied with increasing success and frequency to language learning, where the exercises presented to students often have smaller granularity. Tutorial dialogue systems built with TuTalk focus on directed lines of reasoning, designed to help students to understand important concepts in the domain or to address common misconceptions. They can, but don't have to be, connected to a tutor that helps with problem solving. During the week, you will start out by doing some cognitive task analysis to understand the nature of the problems for which your tutor will provide tutoring. Then, depending on your interest, you will use one or more of the tools described above to implement a computer-based tutor. By the end of the week, you will have a prototype running. In fact, if you decide to focus on intelligent tutoring systems development, you will already have implemented some intelligent tutor behavior by the end of day 2 (an Example-Tracing Tutor).