VCU Business

• INFO 202: Introduction to Information Systems Development Technologies
• INFO 320: Artificial Intelligence for Business Intelligence
• INFO 452: AI Services for Business
• INFO 462: Design of AI Systems
• INFO 474: Advanced Networking and Security
• INFO 535: Ethical, Social and Legal Issues in Computer and Information Systems Security
• INFO 544: Principles of Computer and Information Systems Security
• INFO 617: Text Analytics
• INFO 636: Securing Cloud Infrastructure
• INFO 648: Business Data Analytics
• INFO 658: Securing the Internet of Things
• INFO 602: Big Data Analytics with Cloud Platforms
• INFO 664: Information Systems for Business Intelligence
• INFO 697: Guided Study
• INFO 702: Design Science Research and Method
• INFO 740: AI-based Decision Support Systems

Course descriptions can be found in the VCU Bulletin.

The web is a key battleground in the fight against terrorism. We have developed a theory-based prototype system to address the growing law enforcement challenge posed by online personas with violent potential. The initial system is a pioneering attempt to conceive and automatically populate a knowledge representation of extremist ideologies in the form of an ontology.

In its current form, the system can analyze internet text, detect patterns in violent messages and use those patterns to identify radical individuals by calculating similarities between social media posts and known violent messages. Our ongoing project aims to enhance the core technologies behind this framework and expand its applications.

Given the profound effects of terrorism, this research has the potential for significant impact, advancing the information systems field and contributing to the community—core themes of VCU’s and VCU Business’s strategic plans.

Key Members: Ugochukwu Etudo and Victoria Yoon 

Publications:
Etudo, U., & Yoon, V. Y. (2024). Ontology-based information extraction for labeling radical online content using distant supervision. Information Systems Research, 35(1), 203-225. https://doi.org/10.1287/isre.2023.1223

In the big data era, automated data integration solutions must process high volumes of disparate data robustly and seamlessly for various analytical needs or operational actions. This study, grounded in affordance theory and the goal definition principles from the Goal-Question-Metric approach, designs and instantiates a goal-driven data integration framework for data analytics.

The proposed design automates data integration for nontechnical data users. Specifically, it demonstrates how to elicit and ontologize users’ data-analytic goals and address semantic heterogeneity, enabling the recognition of goal-relevant datasets. In a structured evaluation within the context of counterterrorism analytics, the design artifact shows promising performance in capturing diverse and dynamic user goals for data analytics and generating integrated data tailored to these goals.

Key Members: Dapeng Liu and Victoria Yoon

Publications:
Liu, D., & Yoon, V. Y. (2024). Developing a goal-driven data integration framework for effective data analytics. Decision Support Systems, 180, 114197. https://doi.org/10.1016/j.dss.2024.114197

Trust is a cornerstone of national security, yet deception poses significant risks in high-stakes situations. Understanding and predicting deceptive behavior is crucial for strengthening security strategies.

Our research leverages a unique dataset from the American game show Friend or Foe to analyze how human behaviors—such as facial expressions, body motion and language—relate to trust and deception. By applying advanced multimodal affective computing techniques, we identify key behavioral patterns of deceivers and those who trust, providing insights into real-world security challenges.

Our findings enhance the understanding of deception in critical scenarios and demonstrate the potential of AI-driven behavioral analysis for national security applications.

Key Members: Xunyu Chen

Publications:
Chen, X., Wang, X., Spitzley, L., & Nunamaker, J. (2023). Trust and deception with high stakes: Evidence from the friend or foe dataset. Decision Support Systems, 173, 113997. https://doi.org/10.1016/j.dss.2023.113997

Given the increasing use of conversational agents (CAs), such as ChatGPT, for patient communication, healthcare professionals are placing greater emphasis on the empathetic capabilities of CAs. This project aims to develop an empathetic chatbot that accurately detects patient emotions using deep learning and empirically test its effects on patient well-being.

Additionally, we are developing a GPT-powered open-domain question-answering (OpenQA) system for mental health counseling, called MentalGPT. We are also conducting qualitative research to explore the perceptions driving physicians’ intentional behavior toward AI, providing valuable insights to enhance the integration and overall quality of AI-driven healthcare delivery.

Key members: Abraham Abby Sen, Victoria Yoon, Yeongin Kim and One-Ki Daniel Lee

Robo-advisors, AI-powered cognitive agents that provide intelligent advice or recommendations, have the potential to facilitate asset management and increase asset value at a low cost, yet their acceptance remains low.

As part of this faculty-student research collaboration, we seek to answer what causes resistance to robo-advisors and how to reduce it. Additionally, we propose a new set of design policies for AI robo-advisors to improve deployment and utilization by minimizing potential sources of resistance.

Key Members: Lingyu Li, Victoria Yoon and One-Ki Daniel Lee

Nearly 1 billion people suffer from mental disorders, yet many receive little to no intervention. Generative AI (GenAI) has the potential to bridge this gap by providing supportive content, but its effectiveness remains uncertain.

Our ongoing research examines how patients perceive GenAI-generated support, drawing on the optimal matching model in mental support literature. Leveraging data from an online mental health platform, we apply a Differences-in-Differences approach and deep learning methods to assess GenAI’s performance relative to human helpers.

As we expand our analysis, we aim to advance the understanding of GenAI’s capabilities, refine its language production framework and provide actionable insights for improving its integration into online mental health platforms.

Key Members: Xunyu Chen, Lingyu Li, Xiaojin (Jim) Liu and Victoria Yoon 

As healthcare increasingly shifts to virtual platforms, online health communities (OHCs) are transforming patient-physician interactions. Our research explores how audio communication enhances innovative healthcare services by improving emotional and informational support for patients.

Analyzing extensive multimodal consultation data from a leading OHC, we examine the impact of audio messages on patient engagement and physician outcomes. While audio communication fosters deeper connections and boosts physician credibility, it also presents operational challenges, such as longer consultations and delayed responses.

By studying vocal features and their influence on patient satisfaction, our research provides critical insights into optimizing AI-driven communication tools in digital healthcare. This study informs the design of more effective virtual care models, balancing patient support with physician efficiency.

Key Members: Xunyu Chen, Seokjun Youn (University of Arizona) and Yeongin Kim

In ontology-based information systems, our research has significantly extended XBRL interoperability. The Securities and Exchange Commission (SEC) mandates that all Tier 1 public companies report financial statements in eXtensible Business Reporting Language (XBRL), now the standard for electronic financial data communication. Despite its potential benefits, several challenges prevent XBRL from reaching its full potential. Our ontology-based research program addresses these issues.

To resolve XBRL name conflict issues, we developed Financial Concept Element Mapping (FinCEM) and XBRL Indexing-based Mapping (X-IM). Additionally, we have developed several ontology-based systems:

1. An ontology of recommender system (RS) issues
2. An ontology of financial capability goals
3. A framework for a personal financial recommender system (PFRS), called FinPathlight, a financial technology (FinTech) application designed to improve users' financial capability

We are currently working on developing additional ontologies to further enhance these systems.

Key Members: Ugochukwu Etudo, Dapeng Liu and Victoria Yoon

Publications:
Liu, D., Etudo, U., and Yoon, V. (2020). X-IM Framework to Overcome Semantic Heterogeneity across XBRL Filings. Journal of Association for Information Systems 21(4). DOI: 10.17705/1jais.00626

Etudo, U., Yoon, V., & Liu, D. (2017). Financial concept element mapper (FinCEM) for XBRL interoperability: Utilizing the M3 Plus method. Decision Support Systems, 98, 36-48. https://doi.org/10.1016/j.dss.2017.04.006

This study explores how the design factors of AI chatbots, such as AI explainability and type of Human-AI collaboration, affect user’s Sense of Agency (SoA), user's perception of Human-AI Interaction Quality and ultimately user's engagement in a customer service context. Using a psychological angle through SoA as a user-related psychometric, this study investigates how different levels of Sense of Agency will impact a user's perception. 

Key Members: Samaher Aljudaibi, Victoria Yoon and  One-Ki Daniel Lee