Information Terra: A Narrative-Anchored Semantic-First Projection of Document Embeddings
Authors
Brian Keith Norambuena (Universidad Católica del Norte), Fausto German (Virginia Tech), Chris North (Virginia Tech)
Presentation
- Session
- Lost in Dimensions
- Time
- Thursday, Nov 12, 15:18 – 15:27 (US/Eastern) · session 15:00 – 16:30
- Location
- Hall Essex north
Links
Sign in to access the preprint PDF.
Sign in- Download Supplemental Material
Keywords
Narrative visualization, information landscapes, document embeddings, spherical projection
Abstract
We introduce Information Terra, a narrative-anchored semantic-first projection that places a document corpus on an Earth-like globe whose poles are two user-chosen endpoint documents and whose prime meridian is the great-circle geodesic between them on the embedding hypersphere—so latitude encodes narrative progress and longitude thematic deviation. Land features are recovered from document density via kernel density estimation and labeled by theme. A narrative trail built from the underlying narrative coherence graph, and constrained to be monotone in geodesic progress, provides a readable storyline. The projection's axes are semantically grounded in the user's chosen narrative endpoints, and the globe metaphor affords rotation and antipodal reading. We demonstrate the method on a 540-article Cuban Protests corpus, showing a storyline from Obama's 2016 visit to the 2021 International Aid during the protests.
For Practitioners
Who would be interested: Information Terra is most relevant to practitioners who make sense of large text collections by tracing how a topic develops between two reference points. Data journalists and investigative reporters can follow a documented storyline from an originating event to a later one across a news archive. Policy, intelligence, and competitive intelligence analysts can reconstruct how a situation evolved between two key documents. Digital humanists, historians, and computational social scientists can survey the thematic structure of a corpus and find the clusters that sit opposite a main storyline. Builders of text analytics and sensemaking tools are a secondary audience, since the projection can be embedded as a component. How they would apply it: A practitioner picks two endpoint documents that frame a question of interest and reads the resulting globe, where latitude shows progress from one endpoint toward the other, longitude shows thematic divergence from the connecting path, the recovered landmasses group related documents by theme, and the narrative trail gives a coherent chain of documents to follow or to hand off as a reading list. Because the projection is deterministic and quick to recompute, analysts can revise the endpoints and compare alternative framings interactively, which makes the technique a lightweight exploration step within existing workflows rather than a separate heavyweight tool.