Content deleted Content added
→See also: removed redlink |
added Knowledge graph embedding |
||
Line 62:
An embedding of a graph into three-dimensional space in which no two of the cycles are topologically linked is called a [[linkless embedding]]. A graph has a linkless embedding if and only if it does not have one of the seven graphs of the [[Petersen family]] as a [[minor (graph theory)|minor]].
==Knowledge graph embedding==
[[directed graph|Directed]] [[graph labeling|labeled graphs]] are graphs where relationships have a direction (i.e., a subject and an object), each node has a (unique) name and each edge has a name associated with a relation. These graphs are usually referred to as [[knowledge graph|knowledge graphs]]. Some graph models even feature [[literal (computer programming)|literals]], such as the [[Resource Description Framework]].
In order to preserve the [[semantics (computer science)|semantics]] behind each relation, the problem of knowledge graph embedding is often tackled using approaches based on translations on [[hyperplane|hyperplanes]]<ref name="BordesGlorot2013">{{cite journal|last1=Bordes|first1=Antoine|last2=Glorot|first2=Xavier|last3=Weston|first3=Jason|last4=Bengio|first4=Yoshua|title=A semantic matching energy function for learning with multi-relational data|journal=Machine Learning|volume=94|issue=2|year=2013|pages=233–259|issn=0885-6125|doi=10.1007/s10994-013-5363-6}}</ref>, [[tensor]] [[factorization]]<ref name="NickelTresp2013">{{cite journal|last1=Nickel|first1=Maximilian|last2=Tresp|first2=Volker|title=Tensor Factorization for Multi-relational Learning|volume=8190|year=2013|pages=617–621|issn=0302-9743|doi=10.1007/978-3-642-40994-3_40}}</ref> and [[holography|holographic projections]]<ref>{{citation|first1=Maximilian|last1=Nickel|first2=Lorenzo|last2=Rosasco|first3=Tomaso|last3=Poggio|title=[[AAAI|Thirtieth AAAI Conference on Artificial Intelligence]]|year=2016|chapter=Holographic Embeddings of Knowledge Graphs}}.</ref>.
==See also==
| |||