Knowledge Graph Embeddings
Outline
Why embed the KG ?
Notations
General Algorithm
Scoring Function
Translation Based
Semantic Matching Based
KG embedding with additional resources
Entity types
Relation Paths
Textual Descriptions
Conclusion
Outline
Why embed the KG ?
Notations
General Algorithm
Scoring Function
Translation Based
Semantic Matching Based
KG embedding with additional resources
Entity types
Relation Paths
Textual Descriptions
Conclusion
Embedding the KG into a continuous space while preserving the properties and semantics of the whole KG.
Applications
Link Prediction:
Triple Classification:
Entity Resolution:
Relation Extraction:
Question Answering:
Why Embed the KG ?
Outline
Why embed the KG ?
Notations
General Algorithm
Scoring Function
Translation Based
Semantic Matching Based
KG embedding with additional resources
Entity types
Relation Paths
Textual Descriptions
Conclusion
Notations
KG contains entities
e
∈
E
e \in \mathbb{E}
e
∈
E
and relationships
r
∈
R
r \in \mathbb{R}
r
∈
R
Each KB triple (fact) is represented by
(
h
,
r
,
t
)
(h,r,t)
(
h
,
r
,
t
)
h
h
h
: head entity,
t
t
t
: tail entity and
r
r
r
: relation
Example:
(
AlfredHitchcock
,
DirectorOf
,
Psycho
)
Scoring function
f
r
(
h
,
t
)
f_r(h, t)
f
r
(
h
,
t
)
: measures the plausibility of the fact
(
h
,
r
,
t
)
(h,r,t)
(
h
,
r
,
t
)
Outline
Why embed the KG ?
Notations
General Algorithm
Scoring Function
Translation Based
Semantic Matching Based
KG embedding with additional resources
Entity types
Relation Paths
Textual Descriptions
Conclusion
General Algorithm
Outline
Why embed the KG ?
Notations
General Algorithm
Scoring Function
Translation Based
Semantic Matching Based
KG embedding with additional resources
Entity types
Relation Paths
Textual Descriptions
Conclusion
Translation Based Approaches - TransE
Head entity
h
\mathbf{h}
h
and tail entity
t
\mathbf{t}
t
of a fact
(
h
,
r
,
t
)
(h,r,t)
(
h
,
r
,
t
)
are vectors in
R
d
\mathbb{R}^d
R
d
Relation
r
\mathbf{r}
r
is a translation in
R
d
\mathbb{R}^d
R
d
Scoring function:
f
r
(
h
,
t
)
=
−
∥
h
+
r
−
t
∥
1
/
2
f_r(h,t) = -\lVert \mathbf{h}+\mathbf{r}-\mathbf{t}\lVert_{1/2}
f
r
(
h
,
t
)
=
−
∥
h
+
r
−
t
∥
1/2
Ranking Loss:
L
=
\mathcal{L} =
L
=
Resume presentation
Knowledge Graph Embeddings
Made with Slides.com