ORCID

0009-0009-5364-7688

Year

2024

Season

Spring

Paper Type

Master's Thesis

College

College of Computing, Engineering & Construction

Degree Name

Master of Science (MS)

Department

Computing

NACO controlled Corporate Body

University of North Florida. School of Computing

First Advisor

Dr. Indika Kahanda

Second Advisor

Dr. Xudong Liu

Third Advisor

Dr. Karthikeyan Umapathy

Fourth Advisor

Dr. Sandeep Reddivari

Department Chair

Dr. Asai Asaithambi

College Dean

William Klostermeyer

Abstract

Given the exponential growth of scientific information online, researchers often face the daunting task of detecting contradictory statements on crucial health topics. This work develops a comprehensive pipeline for automated contradiction detection that integrates an Information Retrieval (IR) system, machine learning classifiers, and Explainable AI (XAI). The Information Retrieval system is tailored for biomedical data and comprises a datastore, syntactic, and semantic components. Users can input queries, initiating a pipeline that identifies top documents through syntactic analysis and refines results via semantic examination for relevant research claims. Employing a diverse range of Large Language Models such as pre-trained Distil-BERT, BioBERT, PubMedBERT, and Bioformer, the pipeline uses these models to identify contradictions in research claims. Furthermore, an Explainable AI is utilized to further explain the results of the LLMs classifiers. Thus, a three-step development approach, first, an Information Retrieval System to retrieve right set of biomedical documents and statements, second, meticulously designing a contradiction detection classifier using Large Language Models (LLMs), and finally, the integration of Explainable AI to increase transparency by providing detailed explanations of the machine learning classifier’s decisions is developed.

Download

Share

COinS