Mason Journals (George Mason Univ.)
Not a member yet
3256 research outputs found
Sort by
Murphy, Gretchen. Hemispheric Imaginings: the Monroe Doctrine and Narratives of U.S. Empire
No full text
A Novel Approach to Detecting AI-Generated SQL Queries and Explanations By Utilizing Parallel Dual-Encoder Architecture
No full textAI code generation disrupts programming education by facilitating academic dishonesty but also presents risks to software development, introducing unintended code into complex systems. Specifically for SQL, these risks escalate dramatically because unlike application crashes from coding errors, faulty AI-generated queries corrupt, delete, or expose sensitive data, possibly costing companies billions in losses and regulatory penalties. Traditional detectors for AI generated code are multifaceted and are not specific to SQL Code. This research presents a novel detection system for AI-generated SQL queries and their explanations using a parallel dual-encoder architecture which processes SQL code through specialized transformers (DeepSeek-Coder-V2, StarCoder 2) while simultaneously analyzing explanatory text using language models (DeBERTaV3, LLaMA 3). The detector was trained and evaluated on LeetCode SQL problems using human solutions from LeetCode posts, complemented by AI-generated solutions from GPT-o3, Claude 4 Opus, Gemini 2.5 Pro, and DeepSeek R1 generated by the research team. Our Multi-Model Feature Fusion with Binary Classifier Training achieved 98% accuracy and 98% F1-score, demonstrating consistent performance in all tested AI models. This framework provides educators and organizations with a reliable tool for maintaining academic integrity and database security while establishing a foundational methodology for AI-generated code detection research. Future work involves modifying the dataset to address syntax and structural differences across various SQL dialects, such as PostgreSQL, MySQL, and SQLite
High-performance Photosensor Based on a Nanocomposite of porous graphene and metal nanoparticles
No full textPhotodetectors are critical components widely used in optical sensing and industry, with various applications ranging from autonomous vehicles to consumer electronics. Compared to conventional materials like silicon, laser-induced graphene (LIG) serves as an ideal material for photosensors due to its high broadband light absorption and mechanical durability. However, the effective photoresponse, the current generated per unit of light power, of porous graphene can be significantly enhanced through the addition of metal nanoparticles. We utilize the plasmonic effects of palladium nanoparticles (PdNPs), which concentrate light into intense local fields that increase electron-hole pair generation in the LIG, to boost photoresponsivity. PdNPs can be incorporated into LIG in various ways, including laser doping, physical vapor deposition, and electroless deposition. In this study, we directly compare the effectiveness of two methods: conventional solution processing and deposition during laser fabrication. We expect photosensors with PdNPs added during laser fabrication to have significantly higher photoresponses than those created with conventional solution processing methods. To create the solution-processed photosensor, we irradiate fluorinated polyimide with a CO2 laser to make LIG, then apply a solution of 20nm PdNPs to the graphene. This approach is compared with an alternative technique where PdNPs are incorporated during the laser fabrication of LIG, creating a nanocomposite of graphene and palladium nanoparticles. Due to palladium’s affinity for hydrogen, this platform also demonstrates potential in functioning as a hydrogen detector with further development
Analyzing Visualization Practices in Modeling and Simulation
No full textAs data becomes more complex, it is imperative that it is communicated in a manner that is equally advanced. Poorly made figures in papers focused on data collection and representation can obscure meaning and mislead readers, pointing to an opportunity for the field to evolve its visual standards alongside its models. Prior to advancements in computer graphics and specialty tools, data visualization was limited by technological constraints. This has changed with the emergence of more powerful, affordable computers, giving researchers more diverse, polished figures to support their findings. Digestible visualizations are increasingly important as researchers aim to reach wider audiences, especially in modeling and simulation. This study evaluates visualization quality over the span of 60 years to understand the evolution of data visualization practices and investigate potential challenges. We cover seven prominent journals and conferences in the field of modeling and simulation, analyzing over ten thousand papers published between 1963 and 2025. First, we utilized and fine-tuned a transformer-based document layout Artificial Intelligence (AI) model to locate and extract figures and metadata from these papers. We then trained a couple of AI classifiers to distinguish images of data visualizations from other images using a training-test-validation set that we curated. We achieved an accuracy of 99% on our test set using a vision transformer (ViT), improving on the accuracy of 95% that we measured using a convolutional neural network (CNN). The next step is to evaluate figures that are classified as data visualization with respect to their conformity with modern visualization practices. We believe that our results will shed light on the development and current state of the modeling and simulation field with respect to data visualization and offer practical insight to researchers
Enhancing Agent-Based Model Comprehension through Virtual Reality
No full textAgent-based models (ABMs) are a valuable tool for simulating complex systems as they represent the interactions of autonomous individual components. However, their interpretability is often limited by the two-dimensional (2D) visualizations of spatially located static images on a grid. This makes it challenging for inexperienced users to understand how emergent behavior arises and how agents function. To address this challenge, we developed two agent-based models in the game engine Roblox Studio enhanced with virtual reality (VR). Using Roblox’s toolkit, we created a wolf-sheep predation model adapted from NetLogo, and a Kuramoto-FBM firefly model simulating blinking and movement. We designed these models interactive with support for both VR and keyboard and mouse, each with unique features including (1) spectating the agents from their point of view, (2) possessing the agents with tailored controls for each implementation, (3) capturing the agents, and (4) user-tunable parameters. In addition, custom three-dimensional (3D) models for the agents and scenery were made in Maya and Blender, along with animations, sound design, and visual effects. This allows the user to experience the simulation environment in an engaging and immersive way, while also producing accurate behavioral, population, and environmental data. Overall, these VR implementations leverage the users' understanding beyond other 2D visualizations by enabling users to spatially and temporally experience agents' decisions, bridging the gap between agent rules and emergent behavior. We hope this new implementation of ABMs in Roblox and VR serves as a reference for future research focused on expanding the accessibility, intuition, and interactivity in the modeling and simulation field. Future work comprises investigating other domains in this field, increasing the immersion with improved multiplayer support and visuals, and bettering accessibility to such platforms
Analyzing Tokenization of U.S. Treasuries as On‑Chain Real‑World Assets: Product Structure, Market Implications, and Regulatory Concerns
No full textThe tokenization of real-world assets (RWAs)—particularly U.S. Treasuries—is rapidly bridging the gap between traditional finance (TradFi) and decentralized finance (DeFi), offering fractional ownership, 24/7 settlement, and reduced reliance on intermediaries. Tokenized Treasuries, one of the fastest-growing RWA sectors, provide low-risk exposure to U.S. government debt with added benefits of liquidity, transparency, and programmable automation. Despite their appeal, the market is hindered by fragmented disclosures, jurisdictional inconsistencies, and regulatory ambiguity. To assess these challenges, we conducted due diligence on 47 tokenized Treasury products listed on RWA.xyz, extracting on-chain activity and issuer data via custom Python scripts. We analyzed key stakeholders, legal frameworks, peg mechanisms, transaction patterns, and investor eligibility. Our research shows that tokenized U.S. Treasuries offer yield-bearing exposure backed by government debt and are accessible to KYC-compliant wallets. Compared to most native crypto assets, they provide greater efficiency and stability.
However, the market remains highly concentrated, with 69% of assets issued by just three entities (BlackRock, Franklin Templeton, and Ondo), introducing systemic dependencies and custodial risks. The lack of consistent legal protections and dispute resolution mechanisms, especially across international jurisdictions, raises additional concerns about long-term resilience. Regulatory uncertainty, including the risk of classification as unregistered securities by the SEC, further complicates adoption and may hinder broader institutional engagement. Yet the model holds promise. Tokenized Treasuries deliver reliable yield, broad wallet-based access, and superior liquidity versus native crypto assets. Their continued growth will depend on improved legal clarity, diversification of issuers, and deeper institutional trust
Expression of MMP-9 within EVs of HPV-related Cancer Cells
No full textHuman papilloma virus (HPV) is an infection affecting millions of people globally. According to the CDC, more than 42 million Americans are infected with types of HPV that are known to cause disease. Over 200 strains of HPV are known to affect humans, but HPV-16 and HPV-18 are considered the most common high-risk infections that often lead to cancer, specifically cervical cancer. Extracellular vesicles (EVs) are have the potential to diagnose and track the progression of many diseases and cancers, including HPV-induced cervical cancer. Proteins transported by EVs were shown in prior studies to influence the growth and invasion of the virus within cells, and by extension, increase the likelihood of cancerous lesions in the future. Matrix metalloproteinases (MMPs) are a set of proteins involved in the degradation of the extracellular matrix (ECM) and are dysregulated during the development of cervical cancer. Increased expression of MMPs allows for further disruption of the ECM and migration of cancer into epithelial tissue. Prior studies have demonstrated that MMP-9 is a major player in the proliferation of cervical cancer, and its expression can indicate the prognosis of precancerous and cancerous lesions. This study aimed to compare MMP-9 levels within the EVs of HeLa (HPV-16 infected), U937, and CEM cells. ExoMax was used to isolate EVs and the protein expression of CD63, CD81, GAPDH, and MMP-9 were analyzed using Western blot. Preliminary results show U937 and CEM expressed all four proteins. However, they were not detected in HeLa due to improper EV isolation
The Choice between M1 or M2 macrophages for U937
No full textOne of the most well-known innate immune cells is the M1 and M2 macrophages both derive from monocytes, which are white blood cells. M1 macrophages are activated by IFN-γ or lipopolysaccharide products and induce a pro-inflammatory response, which triggers the inflammation in multiple areas such as the brain or lungs. This macrophage contains extracellular vehicles (EVs) that carry pro-inflammatory cytokines such as TNF-α. M2 macrophages are activated by IL-4 and IL-13 products, and this macrophage induces anti-inflammatory responses and influences wound healing. This macrophage can reduce inflammation in the brain and lungs. In this study, we were given U937 which is a cell line derived histiocytic lymphoma and CEM which are t-cells. These cells can differentiate into macrophages or dendritic cells, and the aim is to determine which macrophage U937 would fall into, which is M1 or M2. The CEM and U937 cells were treated with lipopolysaccharide (endotoxin), and the cells were observed under a microscope for 2 days. EVs were isolated from both cells, followed by a Western blot. After transferring the proteins onto a membrane via gel electrophoresis, primary and secondary antibodies such as CD63-rabbit and secondary anti-rabbit were introduced. CD81-RB and secondary rabbit were introduced reveal whether any CD81 proteins present. TNF- α, an M1 marker for macrophages, and CD11b, an M2 marker, were introduced to the membrane and then imaged. Imaging revealed CD63, CD81, CD11B were detected, while TNF- α was inconclusive. Overall, the U937 leans towards an M2 marker rather than an M1 marker