Date of Award

2024-05-01

Degree Name

Doctor of Philosophy

Department

Computer Science

Advisor(s)

M. Shahriar Hossain

Abstract

Storytelling is a set of algorithms used to create narratives by connecting documents in a sequencethat accurately reflects the evolution of events and entities within a particular topic or theme. Early storytelling algorithms face challenges in encoding the progression and interconnections of information between consecutive texts, given that the conventional approaches rely primarily on connecting document pairs based on content overlap. They often neglect critical linguistic features, such as word contexts, semantics, the roles words play across different documents, and attention to the historical contexts of the underlying documents. Many existing storytelling models frequently produce story chains that, while connected by keywords, lack meaningful coherence in the chains. My dissertation introduces innovative LLM-driven storytelling algorithms to overcome the challenges traditional storytelling algorithms face and significantly enhance downstream tasks. In my research, I propose a role-based contextual embedding algorithm using a large language model that provides a rich understanding of text in a document by considering the different roles of the same word in other documents. I also employ a generative diffusion model to seamlessly link documents within a narrative, even with gaps in the data, to ensure a smoother and more logical story progression. In my dissertation, I introduce a novel distributed attention similarity mechanism designed to control the narrative output of storytelling algorithms locally and globally. The techniques I have designed ensure that the generated stories are not merely connected by keywords but are also coherent and meaningfully sequenced. The experiments in my dissertation indicate that the proposed storytelling models generate more coherent, cohesive, and contextually rich narratives than existing approaches. In addition, I demonstrate that my proposed storytelling model has immense potential in vector data preparation for conventional machine-learning tasks.

Language

en

Provenance

Received from ProQuest

File Size

148 p.

File Format

application/pdf

Rights Holder

Alireza Pasha Nouri

Download

Share

COinS