ETD

• computational narratology
• cultural heritage
• knowledge graphs
• natural language generation
• neurosymbolic AI
• semantic web

This research project harnesses Linked Open Data (LOD) as a foundation for Cultural Heritage (CH) exploration through Automatic Story Generation (ASG). Based on the rationale that discovering and reviving realistic latent historical narratives elicits more user interest than generating ungrounded ones anew, the input-2-graph phase of the ASG pipeline, heavily relying on hand-crafted LOD, is addressed in greater depth. The final Natural Language Generation (NLG) module is therefore constrained by entities, relations, patterns, and rules established in the Knowledge Graph (KG) construction modules — a configuration typical of the neurosymbolic approach.

To facilitate real-world implementation, the pipeline is designed to be modular, with self-sufficient constituent parts. Beyond countless potential applications ranging from education to entertainment, this solution transforms the user from a mere consumer to an empowered participant who not only controls the creation process but also finds within it, and not merely in the final outcome, a valuable source for intellectual growth. This work addresses a specific societal need while simultaneously filling knowledge gaps identified across related scientific domains.

In parallel to this main pipeline, a novel workflow is proposed to exploit the full potential of LOD in the domain of Historical Cryptology, as encrypted diplomatic correspondence throughout history can potentially shed new light on existing historical narratives.

Relevant by-products of the research, such as a hands-on experience in harmonizing the developed applications with other data-orchestration tools in the field of Digital History, are included in the appendix of this work.