Neuro-Symbolic AI for Deep Analysis of Social Media Big Data
Neuro-Symbolic AI for Deep Analysis of Social Media Big Data
Abstract
This tutorial introduces a neuro-symbolic AI framework to analyze big data from social media platforms. Integrating human-curated knowledge through symbolic AI with the pattern recognition capabilities of neural networks enhances the adaptability and efficiency of traditional neural network approaches. Knowledge-guided zero-shot learning techniques enable swift adaption to new linguistic contexts and emerging events [6]. Participants will explore how to design, develop, and utilize these models in specific domains, such as public health surveillance that require dynamic adaptation to new terminologies. The tutorial aims to equip attendees with practical skills and a deep understanding of how to apply neuro-symbolic AI to manage and analyze large-scale social media datasets effectively.
Goals of the Tutorial
Focus of the Tutorial
Foundational Principles:
Lab Exercise Using Real-World Data and Knowledge:
Knowledge Graph Integration:
Relevance to Attendees
This tutorial addresses the pressing challenges associated with vast, dynamic datasets, such as those generated by social media platforms. The tutorial highlights how neuro-symbolic AI technology facilitates adaptation and computational demand, crucial for contemporary applications.
Targeted Audience
Data Scientists and AI Researchers: Those exploring advanced data analysis techniques within AI.
Public Health Officials and Social Media Analysts: Professionals seeking to enhance real-time monitoring and analysis.
Advanced Students and Educators: Individuals seeking knowledge in cutting-edge AI technologies.
Content Level
Length of the Tutorial
We plan a 2-hour tutorial organized in four parts. Each part builds on the previous one, combining theoretical concepts with practical applications of neuro-symbolic AI. The tutorial includes the following segments:
(Total Duration: 2 hours, including interactive segments and breaks.)
General Description of Tutorial Content
Session 1: Theoretical Foundations (30 minutes)
Overview: The evolution of neuro-symbolic AI and its critical role in analyzing dynamic social media data.
Content:
Learning Outcomes:
Session 2: Coding Demonstrations (40 minutes)
Overview: Live coding demonstrations to build neuro-symbolic models that integrate large language models and dynamic knowledge graphs for analysis of social media data during health-related events such as COVID-19.
Content: Participants will implement a neuro-symbolic AI model using Python:
Learning Outcomes:
Session 3: Knowledge Graphs and LLMs in Health Applications (30 minutes)
Overview: This session will explore the application of knowledge graphs and large language models in health contexts, focusing on the unique challenges of social media data.
Content: How neuro-symbolic AI models, enhanced with a knowledge graph, can be tailored to interpret health-related data from unstructured social media data. We address sentiment analysis, trend detection, and the identification of health misinformation.
Learning Outcomes: Participants will understand the critical role of neuro-symbolic AI in enhancing the reliability of health data analysis on social media platforms.
Session 4: Interactive Case Studies and Q&A (20 minutes)
Overview: An interactive wrap-up session allowing participants to engage with real-world applications of neuro-symbolic AI across various data scenarios.
Content: Presentation of diverse case studies where neuro-symbolic AI has been applied, ranging from public health surveillance to analyzing shifts in public sentiment on social media. A live Q&A will address practical challenges, gather feedback, and discuss potential future developments in the field.
Learning Outcomes: Participants will synthesize knowledge from the tutorial, enhancing their ability to apply neuro-symbolic techniques in their work, and leave with a comprehensive understanding of these approaches' practical challenges and advantages in real-time data analysis.
Expected Background and Prerequisites of Audience
This tutorial will combine lecture-style content with hands-on practice in Python. Participants are expected to have a foundational understanding of artificial intelligence and machine learning principles, along with some proficiency in Python programming. Basic knowledge of neuro-symbolic AI concepts and experience with social media data (e.g., from platforms like Twitter or Reddit) will be helpful but is not required.
The tutorial will guide attendees through practical exercises, including using APIs to fetch and analyze real-world data, integrating a knowledge graph, and building neuro-symbolic AI models. Attendees should bring their laptops; details on required tools and background materials will be provided prior to the tutorial. By the end of the session, participants will understand how to leverage neuro-symbolic AI for real-time data analysis, enhancing adaptability, utility, interpretability, and performance in social media applications.
Tutorial Presentation
Note: The websites and slides preview shown are for illustrative purposes only.
Presenters' Biographies
Artificial Intelligence Institute, University of South Carolina
He is a Doctoral Student at the University of South Carolina, specializing in generative AI, knowledge graphs, and reinforcement learning. His research is focused on developing foundational models for pathfinding problems and safe conversational agents for mental health applications. He has worked extensively on integrating Large Language Models and Knowledge Graphs to build robust AI systems. Vedant's notable projects include analyzing the psychological impacts of pandemics using social media data and developing AI conversational agents to support mental health. He has published his research in prominent journals and conferences such as the AAAI Conference on Artificial Intelligence and IEEE Internet Computing.
Department of Computer Science and Engineering, University of Maryland, Baltimore County, Maryland
He is an Assistant Professor at the University of Maryland Baltimore County, Department of Computer Science and Electrical Engineering. His research primarily focuses on the intersection of Neurosymbolic AI, knowledge graphs, and machine learning, strongly emphasizing health informatics and cybersecurity. He has contributed to the field through various roles, including Senior Research Scientist at Samsung Research America and Visiting Researcher at the Alan Turing Institute in London. Notable publications include works on trustworthy NeuroSymbolic AI systems and the application of AI in cybersecurity and privacy. He has co-organized high-profile tutorials and workshops on Neurosymbolic AI at major conferences.
Institute for Insight, Georgia State University, Atlanta, Georgia
He is an Assistant Professor at the Robinson College of Business, Georgia State University, and a former Postdoctoral Fellow at the AI Institute, University of South Carolina. His research spans human-centered social computing, AI for social good, and knowledge-infused learning, specifically focusing on natural language processing and machine/deep learning. He has a robust teaching portfolio, including graduate courses on data programming and research methods for analytics. Dr. Kursuncu has significantly contributed to his field through publications in high-impact journals and conferences and has organized several workshops and tutorials related to cybersecurity and AI.
Contact| @valerie-shalin-a4619b71
Department of Psychology, Wright State University, Dayton, Ohioa
She is a Full Professor in the Department of Psychology at Wright State University and an Adjunct (Affiliated) Faculty at the Department of Computer Science \& Engineering and AI Institute at the University of South Carolina. Her research spans cognitive psychology, human-centered computing, and knowledge-infused learning, significantly contributing to understanding human-machine interaction. Dr. Shalin's work includes developing tools for NASA's Mars Exploration Rover and collaborating on numerous grants from NSF, NASA, and other agencies. She has co-authored impactful publications in top-tier journals and conferences, addressing complex problems in natural language understanding and the cognitive aspects of AI. Her recent research focuses on integrating deep learning and knowledge graphs for enhanced cognitive analytics in healthcare and social media contexts.
Artificial Intelligence Institute, University of South Carolina
He is the founding director of the AI Institute, University of South Carolina (AIISC). His current core re- search includes knowledge-infused learning and explanation- ability. He is a fellow of IEEE, AAAI, AAAS, and ACM. He has co-organized $>$100 international events and tutorials. He has founded three companies by licensing his university research outcomes, including the first Semantic Web company in 1999 that pioneered technology similar to what is found today in Google Semantic Search and KG.
References