arXiv subjects

Zihui Ma

Publications and source records attributed to Zihui Ma.

15 recordsLinked to original sources

arXiv2026

When Earth Foundation Models Meet Diffusion: An Application to Land Surface Temperature Super-Resolution

Land surface temperature (LST) super-resolution is important for environmental monitoring. However, it remains challenging as coarse thermal observations severely underdetermine fine-scale structure. In this paper, we propose Earth Foundation Model-guided Diffusion (EFDiff), a novel framework for super-resolution under extreme spatial degradation. EFDiff uses the Prithvi-EO-2.0 Earth foundation model to encode high-resolution multispectral reflectance into geospatial embeddings, which are injected into the denoising network via cross-attention to guide fine-scale reconstruction from highly degraded observations. We study two variants, EFDiff-$\epsilon$ and EFDiff-$x_0$, which offer complementary trade-offs between perceptual realism and pixel-level fidelity. We evaluate EFDiff under an extreme $32\times$ scale gap using a globally diverse benchmark comprising 242,416 co-registered Landsat thermal-reflectance patches. Results show that EFDiff consistently outperforms baseline methods and that cross-attention conditioning by EFM is more effective than HLS channel concatenation. Although we present EFDiff in the context of LST super-resolution, the framework is broadly applicable to remote sensing problems in which pretrained geospatial representations can guide generative reconstruction.

cs.CV↗

arXiv2026

Politicized Attention Shifts Amplify Polarization in the Information Ecosystem during California Wildfires

Wildfires require governments to communicate under conditions of urgency, uncertainty, and intense public scrutiny, yet such communication now unfolds within a digitally mediated environment shaped by polarization and engagement-based amplification. We analyze over 1.3 million wildfire-related social media posts from California (2016-2025) to examine how institutional actors are evaluated within this landscape. Users' stance toward government is actor-specific: individual political officials are discussed more negatively than operational agencies across federal, state, and local levels, and this gap widens during extreme wildfire events. Moreover, interaction networks become increasingly modular over time, consolidating into polarized communities in which negativity concentrates within cohesive clusters. Engagement-weighted measures show that highly interactive negative content disproportionately shapes visible discourse, while crisis periods redirect attention from emergency agencies to high-profile political figures, reinforcing reputational divergence. These findings indicate that wildfire communication operates within a polarized, engagement-ranked ecosystem in which evaluative tone, network structure, and visibility dynamics jointly shape institutional perception. Effective disaster communication should therefore account for the structural conditions of contemporary digital public communities.

cs.SI↗

arXiv2026

Bridging the Urban Divide: Adaptive Cross-City Learning for Disaster Sentiment Understanding

Social media platforms provide a real-time lens into public sentiment during natural disasters; however, models built solely on textual data often reinforce urban-centric biases and overlook underrepresented communities. This paper introduces an adaptive cross-city learning framework that enhances disaster sentiment understanding by integrating mobility-informed behavioral signals and city similarity-based data augmentation. Focusing on the January 2025 Southern California wildfires, our model achieves state-of-the-art performance and reveals geographically diverse sentiment patterns, particularly in areas experiencing overlapping fire exposure or delayed emergency responses. We further identify positive correlations between emotional expressions and real-world mobility shifts, underscoring the value of combining behavioral and textual features. Through extensive experiments, we demonstrate that multimodal fusion and city-aware training significantly improve both accuracy and fairness. Collectively, these findings highlight the importance of context-sensitive sentiment modeling and provide actionable insights toward developing more inclusive and equitable disaster response systems.

cs.SI↗

arXiv2026

Mobility Inequity and Risk Response After Hurricane Helene: Evidence from Real-Time Travel and Social Sentiment Data

Hurricanes severely disrupt infrastructure and restrict access to essential services. While the physical impacts on post-disaster mobility are well studied, less is known about how individual travel behaviors change during and after disasters, and how these responses are shaped by social and geographic disparities. This study examines mobility patterns following Hurricane Helene, a Category 4 storm that struck six southeastern U.S. states on September 26, 2024, causing over 230 fatalities. Using anonymized GPS mobility data, hurricane severity metrics, and county-level social media sentiment, we examine shifts in travel behavior and their implications for equity. We ask two questions: How do post-hurricane mobility patterns reflect community vulnerability and adaptive capacity? and How do sociodemographic conditions and public sentiment factors shape the direction and extent of mobility change? Results from robust linear and ordered logistic regressions indicate that evacuation orders increase mobility; however, severe storm conditions, particularly high wind speeds, can limit travel. Communities with lower incomes, located in rural areas, and with higher percentages of Black populations exhibit the steepest declines in mobility, suggesting resource constraints and infrastructural barriers, while wealthier, urban, and higher-education areas maintain greater flexibility. Results also show that positive social sentiment is associated with higher mobility and a greater likelihood of increased travel during the hurricane. Our findings highlight the need to address structural barriers and social conditions in post-disaster mobility and disaster response.

cs.SI↗

arXiv2025

Empowering LLM Agents with Geospatial Awareness: Toward Grounded Reasoning for Wildfire Response

Effective disaster response is essential for safeguarding lives and property. Existing statistical approaches often lack semantic context, generalize poorly across events, and offer limited interpretability. While Large language models (LLMs) provide few-shot generalization, they remain text-bound and blind to geography. To bridge this gap, we introduce a Geospatial Awareness Layer (GAL) that grounds LLM agents in structured earth data. Starting from raw wildfire detections, GAL automatically retrieves and integrates infrastructure, demographic, terrain, and weather information from external geodatabases, assembling them into a concise, unit-annotated perception script. This enriched context enables agents to produce evidence-based resource-allocation recommendations (e.g., personnel assignments, budget allocations), further reinforced by historical analogs and daily change signals for incremental updates. We evaluate the framework in real wildfire scenarios across multiple LLM models, showing that geospatially grounded agents can outperform baselines. The proposed framework can generalize to other hazards such as floods and hurricanes.

cs.AI↗

arXiv2025

LSDTs: LLM-Augmented Semantic Digital Twins for Adaptive Knowledge-Intensive Infrastructure Planning

Digital Twins (DTs) offer powerful tools for managing complex infrastructure systems, but their effectiveness is often limited by challenges in integrating unstructured knowledge. Recent advances in Large Language Models (LLMs) bring new potential to address this gap, with strong abilities in extracting and organizing diverse textual information. We therefore propose LSDTs (LLM-Augmented Semantic Digital Twins), a framework that helps LLMs extract planning knowledge from unstructured documents like environmental regulations and technical guidelines, and organize it into a formal ontology. This ontology forms a semantic layer that powers a digital twin-a virtual model of the physical system-allowing it to simulate realistic, regulation-aware planning scenarios. We evaluate LSDTs through a case study of offshore wind farm planning in Maryland, including its application during Hurricane Sandy. Results demonstrate that LSDTs support interpretable, regulation-aware layout optimization, enable high-fidelity simulation, and enhance adaptability in infrastructure planning. This work shows the potential of combining generative AI with digital twins to support complex, knowledge-driven planning tasks.

cs.ET↗

arXiv2025

A Multimodal, Multilingual, and Multidimensional Pipeline for Fine-grained Crowdsourcing Earthquake Damage Evaluation

Rapid, fine-grained disaster damage assessment is essential for effective emergency response, yet remains challenging due to limited ground sensors and delays in official reporting. Social media provides a rich, real-time source of human-centric observations, but its multimodal and unstructured nature presents challenges for traditional analytical methods. In this study, we propose a structured Multimodal, Multilingual, and Multidimensional (3M) pipeline that leverages multimodal large language models (MLLMs) to assess disaster impacts. We evaluate three foundation models across two major earthquake events using both macro- and micro-level analyses. Results show that MLLMs effectively integrate image-text signals and demonstrate a strong correlation with ground-truth seismic data. However, performance varies with language, epicentral distance, and input modality. This work highlights the potential of MLLMs for disaster assessment and provides a foundation for future research in applying MLLMs to real-time crisis contexts. The code and data are released at: https://github.com/missa7481/EMNLP25_earthquake

cs.CL↗

arXiv2025

LLMs as World Models: Data-Driven and Human-Centered Pre-Event Simulation for Disaster Impact Assessment

Efficient simulation is essential for enhancing proactive preparedness for sudden-onset disasters such as earthquakes. Recent advancements in large language models (LLMs) as world models show promise in simulating complex scenarios. This study examines multiple LLMs to proactively estimate perceived earthquake impacts. Leveraging multimodal datasets including geospatial, socioeconomic, building, and street-level imagery data, our framework generates Modified Mercalli Intensity (MMI) predictions at zip code and county scales. Evaluations on the 2014 Napa and 2019 Ridgecrest earthquakes using USGS ''Did You Feel It? (DYFI)'' reports demonstrate significant alignment, as evidenced by a high correlation of 0.88 and a low RMSE of 0.77 as compared to real reports at the zip code level. Techniques such as RAG and ICL can improve simulation performance, while visual inputs notably enhance accuracy compared to structured numerical data alone. These findings show the promise of LLMs in simulating disaster impacts that can help strengthen pre-event planning.

cs.CY↗

arXiv2024

Assessing Response Disparities in California Wildland-Urban-Interface (WUI) Cities Using the Compartmental Model

The increasing frequency and severity of wildfires pose significant risks to communities, infrastructure, and the environment, especially in Wildland-Urban Interface (WUI) areas. Effective disaster management requires understanding how the public perceives and responds to wildfire threats in real-time. This study uses social media data to assess public responses and explores how these responses are linked to city-level community characteristics. Specifically, we leveraged a transformer-based topic modeling technique called BERTopic to identify wildfire response-related topics and then utilized the Susceptible-Infectious-Recovered (SIR) model to compute two key metrics associated with wildfire responses - awareness and resilience indicators. Additionally, we used GIS-based spatial analysis to map wildfire locations along with four groups of city-level factors (racial/ethnic, socioeconomic, demographic, and wildfire-specific). Our findings reveal significant geographic and socio-spatial differences in public responses. Southern California cities with larger Hispanic populations demonstrate higher wildfire awareness and resilience. In contrast, urbanized regions in Central and Northern California exhibit lower awareness levels. Furthermore, resilience is negatively correlated with unemployment rates, particularly in southern regions where higher unemployment aligns with reduced resilience. These findings highlight the need for targeted and equitable wildfire management strategies to improve the adaptive capacity of WUI communities.

cs.SI↗

arXiv2024

Toward satisfactory public accessibility: A crowdsourcing approach through online reviews to inclusive urban design

As urban populations grow, the need for accessible urban design has become urgent. Traditional survey methods for assessing public perceptions of accessibility are often limited in scope. Crowdsourcing via online reviews offers a valuable alternative to understanding public perceptions, and advancements in large language models can facilitate their use. This study uses Google Maps reviews across the United States and fine-tunes Llama 3 model with the Low-Rank Adaptation technique to analyze public sentiment on accessibility. At the POI level, most categories -- restaurants, retail, hotels, and healthcare -- show negative sentiments. Socio-spatial analysis reveals that areas with higher proportions of white residents and greater socioeconomic status report more positive sentiment, while areas with more elderly, highly-educated residents exhibit more negative sentiment. Interestingly, no clear link is found between the presence of disabilities and public sentiments. Overall, this study highlights the potential of crowdsourcing for identifying accessibility challenges and providing insights for urban planners.

cs.SI↗

arXiv2024

Large Language Models in Biomedical and Health Informatics: A Review with Bibliometric Analysis

Large Language Models (LLMs) have rapidly become important tools in Biomedical and Health Informatics (BHI), enabling new ways to analyze data, treat patients, and conduct research. This study aims to provide a comprehensive overview of LLM applications in BHI, highlighting their transformative potential and addressing the associated ethical and practical challenges. We reviewed 1,698 research articles from January 2022 to December 2023, categorizing them by research themes and diagnostic categories. Additionally, we conducted network analysis to map scholarly collaborations and research dynamics. Our findings reveal a substantial increase in the potential applications of LLMs to a variety of BHI tasks, including clinical decision support, patient interaction, and medical document analysis. Notably, LLMs are expected to be instrumental in enhancing the accuracy of diagnostic tools and patient care protocols. The network analysis highlights dense and dynamically evolving collaborations across institutions, underscoring the interdisciplinary nature of LLM research in BHI. A significant trend was the application of LLMs in managing specific disease categories such as mental health and neurological disorders, demonstrating their potential to influence personalized medicine and public health strategies. LLMs hold promising potential to further transform biomedical research and healthcare delivery. While promising, the ethical implications and challenges of model validation call for rigorous scrutiny to optimize their benefits in clinical settings. This survey serves as a resource for stakeholders in healthcare, including researchers, clinicians, and policymakers, to understand the current state and future potential of LLMs in BHI.

cs.DL↗

arXiv2023

Thriving in a Pandemic: Lessons Learned from a Resilient University Program Seen Through the CoI Lens

In March 2020, college campuses underwent a sudden transformation to online learning due to the COVID-19 outbreak. To understand the impact of COVID-19 on students' expectations, this study conducted a three-year survey from ten core courses within the Project Management Center for Excellence at the University of Maryland. The study involved two main steps: 1) a statistical analysis to evaluate students' expectations regarding "student," "class," "instructor," and "effort;" and 2) a lexical salience-valence analysis (LSVA) through the lens of the Community of Inquiry (CoI) framework to show the changes of students' expectations. The results revealed that students' overall evaluations maintained relatively consistent amid the COVID-19 teaching period. However, there were significant shifts of the student expectations toward Cognitive, Social and Teaching Presence course elements based on LSVA results. Also, clear differences emerged between under-graduates and graduates in their expectations and preferences in course design and delivery. These insights provide practical recommendations for course instructors in designing effective online courses.

cs.CY↗

arXiv2023

Investigating disaster response through social media data and the Susceptible-Infected-Recovered (SIR) model: A case study of 2020 Western U.S. wildfire season

Effective disaster response is critical for affected communities. Responders and decision-makers would benefit from reliable, timely measures of the issues impacting their communities during a disaster, and social media offers a potentially rich data source. Social media can reflect public concerns and demands during a disaster, offering valuable insights for decision-makers to understand evolving situations and optimize resource allocation. We used Bidirectional Encoder Representations from Transformers (BERT) topic modeling to cluster topics from Twitter data. Then, we conducted a temporal-spatial analysis to examine the distribution of these topics across different regions during the 2020 western U.S. wildfire season. Our results show that Twitter users mainly focused on three topics:"health impact," "damage," and "evacuation." We used the Susceptible-Infected-Recovered (SIR) theory to explore the magnitude and velocity of topic diffusion on Twitter. The results displayed a clear relationship between topic trends and wildfire propagation patterns. The estimated parameters obtained from the SIR model in selected cities revealed that residents exhibited a high level of several concerns during the wildfire. Our study details how the SIR model and topic modeling using social media data can provide decision-makers with a quantitative approach to measure disaster response and support their decision-making processes.

cs.SI↗

arXiv2023

ChatGPT in education: A discourse analysis of worries and concerns on social media

The rapid advancements in generative AI models present new opportunities in the education sector. However, it is imperative to acknowledge and address the potential risks and concerns that may arise with their use. We analyzed Twitter data to identify key concerns related to the use of ChatGPT in education. We employed BERT-based topic modeling to conduct a discourse analysis and social network analysis to identify influential users in the conversation. While Twitter users generally ex-pressed a positive attitude towards the use of ChatGPT, their concerns converged to five specific categories: academic integrity, impact on learning outcomes and skill development, limitation of capabilities, policy and social concerns, and workforce challenges. We also found that users from the tech, education, and media fields were often implicated in the conversation, while education and tech individual users led the discussion of concerns. Based on these findings, the study provides several implications for policymakers, tech companies and individuals, educators, and media agencies. In summary, our study underscores the importance of responsible and ethical use of AI in education and highlights the need for collaboration among stakeholders to regulate AI policy.

cs.CY↗

arXiv2023

A Bibliometric Review of Large Language Models Research from 2017 to 2023

Large language models (LLMs) are a class of language models that have demonstrated outstanding performance across a range of natural language processing (NLP) tasks and have become a highly sought-after research area, because of their ability to generate human-like language and their potential to revolutionize science and technology. In this study, we conduct bibliometric and discourse analyses of scholarly literature on LLMs. Synthesizing over 5,000 publications, this paper serves as a roadmap for researchers, practitioners, and policymakers to navigate the current landscape of LLMs research. We present the research trends from 2017 to early 2023, identifying patterns in research paradigms and collaborations. We start with analyzing the core algorithm developments and NLP tasks that are fundamental in LLMs research. We then investigate the applications of LLMs in various fields and domains including medicine, engineering, social science, and humanities. Our review also reveals the dynamic, fast-paced evolution of LLMs research. Overall, this paper offers valuable insights into the current state, impact, and potential of LLMs research and its applications.

cs.DL↗