Home
The Clean CaDET Tutor hosts the educational aspect of the Clean CaDET project. It consists of several modules, collaborating to support instructors and learners in a series of intelligent tutoring use cases.
The Tutor supports instructors in:
- Knowledge and skill modeling and digital educational content creation.
- Registering, grouping, and enrolling learners to access educational content.
- Analyzing the progress of a learner group.
- Analyzing the statistics related to a skill or piece of educational content.
The Tutor supports learners in:
- Optimizing the time spent developing a skill and interacting with a piece of educational content to avoid over-practice and under-practice.
The following diagram illustrates the main building blocks of the Clean CaDET Tutor.
From left to right, the diagram showcases the primary actors interacting with the Tutor, the user interfaces, the application services, the core models, and the notable data stores.
We built the Tutoring system around the core models listed in this section. Most of the our research is dedicated to understanding and enhancing these models.
The Domain Model provides a structure for knowledge and educational content based on the KLI framework, which couples educational materials to applicable skills that the learner should develop.
The following page describes our Domain Model, its building blocks and their relationships in more detail.
The Learner Model gathers everything we know about the learner. A significant part of the learner model is the domain overlay model that describes the interaction the learner has with the domain model, including interactions with the educational content (e.g., viewing instructional items, answering assessment items) and an estimate of their skill mastery.
The following page describes our Learner Model, its building blocks and their relationships in more detail.
The models listed here represent parts of the problem domain that support our core models, but do not present significant advances or introduce any novelty.
The Enrollment Model is a simple, supporting structure tasked with gathering learners into groups. The groups provide a useful mechanisms for instructors to view the learners progress against their peers and to understand which skills the groups has trouble with.
The Tutor's application services coordinate the core and supporting models and technological features (e.g., repositories) to provide complete use cases to the users.
Most of the complexity related to the learner's interaction with the Tutor is stored in the Learner Model. The Learning Services present a thin layer between the user interface and the core models and are of little interest.
The Analytics Services provide several visualizations for instructors to analyze the interaction between the learners and the educational content. These analytics include:
- Viewing the KCM events across all KCs and learners.
- Viewing statistical information derived from events tied to specific KCs. This includes answering questions like:
- How many learners started interacting with a KC?
- How many learners went through (completed) all the educational content tied to a KC?
- How many learners mastered (passed) a KC?
- What is the learners' time to completion, visualized through a box plot?
- What is the learners' time to mastery, visualized through a box plot?
- Viewing the progress of a group of learners for a specific unit, where the questions from the previous (KC-related) bullet are now asked from the individual learner's perspective.
The current authoring services require some technical expertise to use. Knowledge units, knowledge components, instructional, and assessment items can be created using the following spreadsheet (it is populated with a "tutorial" unit meant to get the learner acquainted with the Tutor's features).
Once the spreadsheet is populated, it should be saved as an xlsx file. You may then run the following test to generate an SQL script that can be inserted to the database.
The repository that hosts these wiki pages contains code for the application services and the core models. The user interfaces interacting with this back-end part of the system are part of different repositories. Here we briefly discuss them.
Learners access most of the educational content through a browser front-end application, hosted in a dedicated code repository. This application also offers analytics tools for instructors.
Learners complete refactoring challenges by using their IDE (specifically Visual Studio Code) and our plugin for this environment. The plugin's code can be found here.