Dynamic Relational Feature
Jump to navigation
Jump to search
See: Dynamic Feature, Relational Feature, Static Relational Feature, Dynamic Relational Attribute.
References
2000
- (Neville & Jensen, 2000) ⇒ Jennifer Neville, and David Jensen. (2000). “Iterative Classification in Relational Data.” In: Proceedings of the Workshop on Statistical Relational Learning.
- Relational attributes fall into two categories which we will call static relational and dynamic relational. Any intrinsic attribute has the potential to be predicted by an SBC model; from the same company data we could predict any of the intrinsic attributes mentioned above. Static relational attributes use known intrinsic attributes of related objects and as such they can be computed without the need for inference. The values of static relational attributes remain constant over the course of classification. Dynamic relational attributes use inferred intrinsic attributes of related objects so they require that at least some related objects be classified before the attribute can be computed. The values of dynamic relational attributes may change as classification progresses and additional inferences are made about related objects.