A Survey on Spatio-Temporal Knowledge Graph Models

TL;DR

The paper addresses the fragmentation in spatio-temporal knowledge graph modeling by providing a systematic review of static, temporal, and spatial graph models and proposing a modeling guideline for STKGs. It traces foundational and contemporary STKG approaches, analyzes key design choices (edge semantics, temporal and spatial annotation, and dimensionality), and highlights open challenges across modeling, construction, provenance, integration, and evaluation. The authors argue that many STKGs are domain-specific and struggle with reusability and long-term preservation, emphasizing the need for generalized abstractions and provenance-aware designs. By outlining a structured modeling workflow and identifying cross-domain challenges, the work aims to catalyze more interoperable and durable STKGs that support broader analyses and cross-domain insights.

Abstract

Many complex real-world systems exhibit inherently intertwined temporal and spatial characteristics. Spatio-temporal knowledge graphs (STKGs) have therefore emerged as a powerful representation paradigm, as they integrate entities, relationships, time and space within a unified graph structure. They are increasingly applied across diverse domains, including environmental systems and urban, transportation, social and human mobility networks. However, modeling STKGs remains challenging: their foundations span classical graph theory as well as temporal and spatial graph models, which have evolved independently across different research communities and follow heterogeneous modeling assumptions and terminologies. As a result, existing approaches often lack conceptual alignment, generalizability and reusability. This survey provides a systematic review of spatio-temporal knowledge graph models, tracing their origins in static, temporal and spatial graph modeling. We analyze existing approaches along key modeling dimensions, including edge semantics, temporal and spatial annotation strategies, temporal and spatial semantics and relate these choices to their respective application domains. Our analysis reveals that unified modeling frameworks are largely absent and that most current models are tailored to specific use cases rather than designed for reuse or long-term knowledge preservation. Based on these findings, we derive modeling guidelines and identify open challenges to guide future research.

Figures (12)

• Figure 1: From spatio-temporal data to a spatio-temporal knowledge graph. In the example scenario, data is collected at multiple sites over several points in time (left). The spatio-temporal knowledge graph represents these data points as nodes, incorporating their spatial information, temporal attributes and possible relationships (right).
• Figure 2: Overview of STKG origins. The simple graph model has been extended by additional characteristics and combines temporal and spatial properties.
• Figure 3: Temporal annotations in a temporal knowledge graph. In a node-annotated graph (A) the time is stored at the nodes, i.e. the nodes $u$ and $v$ are valid at time $t$. An edge-annotated graph (B) stores the time at the edges, i.e. the edge between $u$ and $v$ exists at time $t$, the nodes remain static. In a node-edge-annotated graph (C) the time can be stored at the nodes and edges. In a graph-annotated approach (D) neither the nodes nor the edges store time. The graph $G_t$ represents the nodes and edges at time $t$.
• Figure 4: Temporal semantics in a temporal knowledge graph. A duration-labeled semantic (A) defines the traversal time from one node to another. An interval-labeled semantic (B) indicates the periods during which the nodes or edges are valid. A timestamp-labeled semantic (C) reflects discrete timestamps at which a node or edge is valid.
• Figure 5: Obtaining edges between nodes with spatial information. Sometimes (A) edges are inherently given between nodes, e.g. as power lines between power poles. Inference methods (B) try to infer missing or unknown edges between entities, e.g. by defining a threshold for a distance.

Theorems & Definitions (3)

• Definition 1: Simple Graph
• Definition 2: Property Graph angles2018property
• Definition 3: RDF Graph Tamer_2016