An Enhanced Knowledge Graph Embedding for Small-Scale Sparse Knowledge Graph

Poor KGE on small sparse graphs: Existing knowledge graph embedding techniques perform poorly on small-scale datasets because small knowledge graphs are incomplete and sparse, leading to overfitting.

Temporal dependency preservation in attack KGs: Applying KGE to the Cybersecurity Attack Knowledge Graph requires handling temporal attributes to preserve temporal dependencies of attack steps.

Poor KGE on small sparse graphs: Existing knowledge graph embedding techniques perform poorly on small-scale datasets because small knowledge graphs are incomplete and sparse, leading to overfitting.

Temporal dependency preservation in attack KGs: Applying KGE to the Cybersecurity Attack Knowledge Graph requires handling temporal attributes to preserve temporal dependencies of attack steps.

Enhanced small-graph KGE: The paper proposes an enhanced knowledge graph embedding method for small-scale knowledge graphs that combines rule-based data augmentation with neighborhood-based embedding enhancement.

Temporal attribute abstraction: The method abstracts temporal attributes into knowledge to preserve temporal dependencies in cybersecurity attack steps.

Spatial-rule ranking correction: The method corrects KGE ranking by using spatial rule scores.

Enhanced small-graph KGE: The paper proposes an enhanced knowledge graph embedding method for small-scale knowledge graphs that combines rule-based data augmentation with neighborhood-based embedding enhancement.

Temporal attribute abstraction: The method abstracts temporal attributes into knowledge to preserve temporal dependencies in cybersecurity attack steps.

Improved KGE link prediction metrics:

Improved KGE link prediction metrics: