Date of Award

12-2025

Document Type

Thesis

Degree Name

Master of Science

Department

Computer Science

Abstract

Mass-shooting events pose a significant challenge to public safety because they generate large volumes of unstructured textual data, making it difficult to conduct effective investigations and formulate public policy. The proposed knowledge acquisition system utilized Large Language Models (LLMs) with few-shot prompting to extract vital information from news articles, police reports, and social media content at high speed. The system used entity recognition to detect vital information, which included offenders, victims, locations, and criminal instruments that support legal investigations. An experimental study conducted on actual mass-shooting datasets showed GPT-4o performs best for mass-shooting NER. The o1-mini delivered competitive performance while using fewer resources more efficiently for basic NER applications. It is also observed that increasing the number of shots enhanced the performance of all models, but the gains were more substantial for GPT-4o and o1-mini, highlighting their superior adaptability to few-shot learning scenarios.

Index Terms — AI-driven justice, few-shot learning, information extraction, knowledge acquisition, knowledge graphs, large language models, mass shooting, named entity recognition (NER), prompt engineering.

Committee Chair/Advisor

Lin Li

Committee Co-Chair

Xishuang Dong

Committee Member

Md Hossain Shuvo

Committee Member

Lening Wang

Publisher

Prairie View A&M University

Rights

© 2021 Prairie View A & M University

Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.

Date of Digitization

01/13/2026

Contributing Institution

J. B . Coleman Library

City of Publication

Prairie View

MIME Type

Application/PDF

Download


Share

COinS