We are currently analyzing the notional machine and conceptual assertions of artificial intelligence from popular MOOCs. The goal is to find out the similarity and disparity of the conceptual assertions and notional machine representation across the MOOCs while they introduce and define AI.

I have developed low-cost physical interactive models to make STEM concepts alive with Makey Makey and Scratch such as:

1. Making primitive variables alive

2. Making playdoh artifacts alive

3. Making Monalisa alive with AI Integration

    Research Questions

 1. How do interactive physical models influence student engagement and conceptual understanding in STEM subjects? 

2. How do we integrate interactive physical models into the K-12 curriculum?

3. What added value does AI bring to these interactive models in terms of personalization, assessment, and engagement? 

4. How can such AI-integrated models be effectively scaled across diverse classroom settings?

My novel approach to elicit and classify mental models of Java arrays can be applied to create scalable assessment of other fundamental programming concepts (e.g., variables, objects) and identify students’ misconception in different points of programming curriculum. By building on this prior work, my future prospective research goals are the following: 

1. Expand MMT to Other Programming Concepts and Develop Instructor Dashboard for Real-Time Concept Diagnosis: Develop Mental Model Tests for additional core CS1 concepts such as: variables, objects and create interactive dashboard for Instructors to real-time concept diagnosis. 

2. Integrate the MMT with Intelligent Tutoring Systems (ITS): Integrate the MMT into an ITS that provides targeted feedback and custom learning pathways based on detected mental models. 

3. Investigate the Effectiveness of Explanative Diagrams in CS1 Students’ Mental Models: Utilizing the MMT to assess mental models, I aim to investigate the effectiveness of Mayer's explanative diagrams based on cognitive theory of multimedia learning in CS1 students’ mental models. 

4. Longitudinal Analysis of Mental Model Evolution and Identify Liminal Space using MMT among CS Students: Utilizing the MMT as a tool to assess mental models, I plan to conduct a longitudinal study that tracks students’ mental models across multiple semesters and identify liminal space of programming concepts. 

My dissertation aimed to investigate novice programmers' mental models of arrays. Below are the key contributions of my dissertation work:

• Definition of mental models applicable for programming concepts: I perceive the mental models of a programming concept as they contain a set of assertions for each part (of the structure) and state changes. Here, an assertion is a single belief or notion in a human’s mind.  [Dissertation]

• Development of Mental Model Test: Based on the two crucial characteristics of mental models- correctness and consistency, I developed a quantitative approach to elicit mental models which is named in my dissertation - the Mental Model Test (MMT). [Poster: SIGCSE '2021]

• Identification of Misconceptions: With the MMT, I identified several misconceptions before and after classroom instruction of arrays. [Dissertation] [FIE '2023] 

• Measurement of Mental Model Correctness after Classroom Instruction: Even after classroom instruction I observed lower correctness in the mental models of arrays state changes (e.g., declaration) than the parts (e.g., name). [SIGCSE '2024]

Key Findings [Slides]:

• Students’ mental model correctness and consistency for array’s state changes (e.g., declaration) are lower than the parts (e.g., name) components. [Dissertation]

• Prior knowledge affects initial mental models.

In this cross-sectional study of 102 US hospitals, 68% of hospitals in the sample offered proxy accounts to caregivers of adult patients, 45% of the hospital personnel surveyed endorsed sharing of login credentials, and 19% of hospitals that provided proxy accounts enabled patients to limit the types of information seen by their caregivers. 

[JAMA Internal Medicine 2020] 

I have analyzed the diagrams of 15 commonly used CS1 Java textbooks to measure the quality of diagrams using Richard E. Mayer's framework of effective diagrams. 

Findings: I found that none of the textbook diagrams could retain the criteria to become effective diagrams possibly failing to effectively impact novice programmer's mental models.

[ITiCSE 2020](28% acception rate), [Poster: SIGCSE 2020], [Slides]

As a requirement of my Cognition course instructed by Dr. Alexia Galati, I synthesized literature of mental model across different disciplines

Slides ->

NSF funded - Connected Learner project at the University of North Carolina at Charlotte is a re-orientation of undergraduate computing and informatics education to focus on student learning that connects to peers, the profession, and the community. The project vision is to transform the student entering an undergraduate computing and informatics program from a person with an interest in computing – to a person with an identity as a computing professional. 

I analyzed student and teaching assistants' survey data as part of the evaluation process of the project. 

[FIE 2019] [Poster: SIGCSE 2020]