Duong Tung Ta PHD student

Research Experience

•  LARGE LANGUAGE MODELS I am presently conducting research at the Coral Lab , which is supervised by Dr. Tim Oates, a Professor of Computer Science and Electrical Engineering at the University of Maryland, Baltimore County. My specific area of focus involves enhancing the reasoning abilities of large language models through auto-metacognitive prompting, with the goal of improving understanding in these models.



•  NATURAL LANGUAGE PROCESSING  As part of my role at CMC Institute of Technology (CMC CIST), I had the opportunity to contribute to the NL2SQL project, which focused on translating natural language questions into SQL commands leveraging advanced natural language processing techniques. Within this project, I conducted experiments utilizing the RoBERTa model in multilingual mode, specifically aimed at addressing low-resource languages like Vietnamese. My approach demonstrated real-time efficiency while achieving acceptable accuracy in comprehending Vietnamese queries.



•  KNOWLEDGE GRAPH  In addition to my work on the NL2SQL project at CMC CIST, I delved into knowledge graph techniques and explored similar search algorithms. Specifically, I employed the TransE model for knowledge graph embedding of company features and utilized the HNSW similar search algorithm. Subsequently, I enhanced the accuracy by 5% through the incorporation of the ComplEx knowledge graph model and the scaNN search algorithm. These endeavors significantly contributed to the successful development of our business-to-business recommendation system product.