Computational Thinking

• Title: Computational Thinking: Automated Formative Assessment of Parametric Modelling Assignments
• Project Period: 1 January 2018 to 31 December 2018
• Funding Source: NUS LIFT
• Investigators: Patrick Janssen (PI)
• Budget: SGD $111,744

Research Overview

This research proposal addresses the teaching of Spatial Computational Thinking using am existing visual programming platform called Möbius Modeller (Janssen et al. 2016). This platform is a web-based visual programming environment that allows 3D models to be generated and manipulated directly in the web browser.

This research proposes to develop an extension of the Möbius Modeller for automated formative assessment. With this system, students upload their assignment submissions to a server, and the system then automatically assesses the validity of the submissions, giving immediate feedback.

Research Aims and Objectives

The aim of the research is to develop and evaluate the Automated Assessment System for performing the automated assessment of student submissions of Möbius assignments.

For each assignment, both students and tutors create Möbius Flowcharts. Students need to create a Model Generation Flowchart that generates a target 3D model (with specific data properties and attributes as described in the assignment). Tutors need to create a Model Validation Flowchart that checks the 3D models generated by the student Flowcharts. The Automated Assessment System will process each submission and check its validity by executing both Flowcharts. First the student Flowchart is executed, thereby generating a model. The validity of this model is then checked by executing the tutor Flowchart, which will generate some validation results.

Automated assessment process: 1) execute the student's Model Generation Flowchart; 2) execute the tutor's Model Validation Flowchart; 3) return the results to the student.

For the tutors, they need to create a Model Validation Flowchart that will check the validity of the student models. This Flowchart may use various analysis techniques, including:

• Schema Analysis: The model is validated against a JSON schema. This schema may be specific to the assignment, identifying elements, attributes, and properties to be present in the model.
• Spatial Analysis: The model is spatially compared to a target model. A tolerance may be given, so that any deviations below that tolerance can be deemed acceptable.
• Analysis of Embedded Data: The model is analysed and data embedded in the model is extracted and checked.
• Analysis of Derived Data: The model is analysed and additional data is derived from the model and checked. For example, certain areas or volumes may be calculated. Or for more complex scenarios, specific simulations may need to be executed, generating additional data.

The Model Validation Flowchart may combine these types of analysis in various ways. In general, the first step is always to validate the models against a JSON schema. This may then be followed by other more detailed types of analysis.