As a passionate AI researcher at the University of South Florida (USF), I have had the opportunity to explore cutting-edge technologies in the field of Artificial Intelligence, focusing on advanced machine learning models and information extraction systems. My research efforts have primarily been directed toward building intelligent systems that can understand and interpret complex data relationships, which is an exciting frontier in AI development.
I am currently part of the Knowledge Extraction Group at USF, working closely with Professor John Licato and PhD student Onur Bilgin. My research aims to advance the capabilities of AI models, specifically in the realm of information extraction and spatial reasoning.
Information Extraction: My work involves developing AI systems that can accurately extract and interpret data from unstructured sources. This includes methods such as TF-IDF, BM25, and cosine similarity, which are used to process text data and extract meaningful information.
Embedding Models: I have worked with embedding models from Ollama for tasks related to search and retrieval augmented generation (RAG). These models play a crucial role in understanding the context and relationships within datasets, enabling more intelligent decision-making.
Spatial and Relational Data Interpretation: One of the exciting challenges I’ve been tackling is the development of algorithms that can understand and analyze spatial relationships between data points. For example, my current project involves interpreting datasets like 'facts_2.tsv,' which require understanding relative positions and other spatial attributes. I’m responsible for labeling this data and ensuring the accuracy of spatial relationships in AI models.
COLIEE Competition 2024: As part of my research group, I participated in the COLIEE competition, which focuses on legal information extraction and entailment. This experience helped me understand the complexities of domain-specific AI applications and further strengthened my skills in building systems capable of reasoning and decision-making.
FAIRIS-Lite Framework: I am also working on a project involving the FAIRIS-Lite framework, a platform for implementing navigational control logic for robots. My contributions to the project include the development of a Webots Robot Controller, which I built by following detailed documentation and integrating with the WebotsSim libraries from the public FAIRIS-Lite GitHub repository. This controller allows robots to make autonomous decisions based on environmental data.
Python: The primary language used in my research, allowing me to efficiently implement algorithms and data analysis pipelines.
Ollama Embedding Models: Used for enhancing AI's understanding of text and relational data.
TF-IDF, BM25, Cosine Similarity: These techniques are essential in our information retrieval projects, helping the AI to rank and retrieve relevant documents from large datasets.
Webots: For simulating robot environments and implementing control algorithms in the FAIRIS-Lite project.
Through my research, I aim to contribute to the development of AI systems that can reason, learn, and adapt to new environments. I am particularly interested in creating systems that can process and interpret complex data relationships, allowing AI to have a deeper understanding of the world around it.
Looking forward, I plan to expand my research to explore generative AI models, natural language processing (NLP) applications, and autonomous decision-making systems in both digital and physical environments.