SHADOWS: A Computational Knowledge Graph of Mythological and Literary Characters for Cross-Cultural Narrative Analysis

This poster presents Shadows, an ongoing digital humanities project developing a computational knowledge graph of mythological and literary characters designed to enable large-scale comparative analysis of narrative figures across cultures. Positioned at the intersection of computational literary studies, comparative mythology, and knowledge graph modeling, the project explores how graph-based methods can support the identification of structural correspondences among narrative characters drawn from diverse traditions. The current phase of Shadows focuses on constructing a structured corpus of mythological entities from multiple cultural areas. Each character is modeled as a node within a knowledge graph enriched with attributes including domain, associated objects and animals, character traits, physical characteristics, key narrative events, birth and death types, gender, and family roles. These attributes enable the formal encoding of narrative and symbolic features traditionally addressed qualitatively in comparative mythology and literary studies. The project employs a mixed-method data acquisition strategy combining AI-assisted extraction, structured manual encoding, and collaborative scholarly validation. AI-based tools assist in identifying and extracting relevant character attributes from encyclopedic and textual sources, accelerating corpus construction at scale. This process is complemented by manual curation and scholarly sourcing: contributors can link encoded attributes to specific passages in mythological or literary texts, ensuring traceability and interpretability. The platform also supports controlled participatory contributions, allowing users to propose new characters or attributes subject to review. This hybrid workflow balances scalability with scholarly rigor. Using graph modeling and exploratory network analysis, the platform enables visualization of clusters, proximities, and correspondences among characters across traditions through interactive visualization and correspondence analysis tools. The scholarly argument underlying Shadows is that computational graph modeling enables the simultaneous comparison of far more narrative characters than is possible through traditional qualitative approaches. Structuring large corpora within a unified relational framework makes it possible to identify patterns of similarity, clustering, and divergence across dozens or hundreds of figures at once, revealing structural regularities difficult to detect through close reading alone. Shadows is conceived as a multi-phase knowledge graph integrating mythological and literary fictional characters. The first phase focuses on mythological figures, followed by literary characters from canonical fairy tales and global fiction (e.g., Snow White, Aladdin, Bluebeard, Dracula, Cthulhu, Harry Potter). Inclusion follows a reproducible pipeline based on automated extraction and encyclopedic notability, ensuring transparency and scalability. Shadows aims to provide both a research tool and a methodological framework for computationally modeling narrative figures at scale.