# [Paper Study]KGE: A Survey of Approaches and Applications
https://persagen.com/files/misc/Wang2017Knowledge.pdf
### Details
- KG embeddings steps:
1. Entity, relation represenstaions
2. score function
3. learning, optimize representations
### Embedding techniques:
#### Translational distance models
==distance-based scoring functions==

- TransE:
- distance between h+r, t
- flaw: 1-N, N-1, N-N
- TransH:
- model relation as vector distance on a hyperplane.

- TransR: project to relation space
- TransD: decompsose projection matrix into 2 vectors.
- TransM: manifold
- TransA: adaptive Mahalanobis distance
- TransF: ...

#### Semantic Mathing Models
==similarity based scoring function==
latent representation


- semantic matching models
- RESCAL
-