Skip to content
Last Dictionary
Plane
Symbolic & Classical AIMachine LearningDeep Learning

knowledge graph embedding

nounAlso: KGE, TransE, knowledge graph completion, knowledge base completion, link prediction

In plain wordsA way of turning the things and relations in a knowledge graph into lists of numbers, so a computer can guess missing facts from patterns, such as predicting where someone was born.

  1. A method that maps each entity and relation of a knowledge graphknowledge graph, n.: A huge web of facts stored as connections between things — 'Paris is the capital of France', 'Marie Curie won a Nobel Prize' — that computers can search, combine and reason over. to a low-dimensional vector (or matrix), trained so that a scoring function gives high scores to true triples and low scores to corrupted ones. In TransE, a relation acts as a translation: for a true triple (h, r, t), the head's embeddingembedding, n.: A list of numbers that stands for a word, picture or other item, arranged so that similar items get similar lists. It lets a computer measure similarity of meaning as distance. plus the relation's should lie close to the tail's. Other models use rotations in complex space (RotatE), bilinear products (RESCAL, DistMult, ComplEx), or convolutional and graph neural networkgraph neural network, n.: A neural network for data shaped like a web of connections, such as molecules, social networks or road maps, in which each point repeatedly gathers information from its neighbors. encoders. Learned embeddings support link prediction (knowledge-graph completion), entity resolution, recommendation and question answering, and inject relational structure into neural models. Evaluation usually ranks the true answer among all candidates, reporting mean reciprocal rank and Hits@10 on benchmarks such as FB15k-237 and WN18RR.

f(h,r,t)=−∥h+r−t∥f(h, r, t) = -\lVert \mathbf{h} + \mathbf{r} - \mathbf{t} \rVert

Origin

Antoine Bordes and colleagues introduced TransE in 2013, following tensor-factorization approaches such as Maximilian Nickel, Volker Tresp and Hans-Peter Kriegel's RESCAL (2011). Kristina Toutanova and Danqi Chen (2015) and Tim Dettmers and colleagues (2018) built the harder FB15k-237 and WN18RR benchmarks after finding test leakage in earlier ones.

Related

See also
  • knowledge graphknowledge graph, n.: A huge web of facts stored as connections between things — 'Paris is the capital of France', 'Marie Curie won a Nobel Prize' — that computers can search, combine and reason over.
  • embeddingembedding, n.: A list of numbers that stands for a word, picture or other item, arranged so that similar items get similar lists. It lets a computer measure similarity of meaning as distance.
  • graph neural networkgraph neural network, n.: A neural network for data shaped like a web of connections, such as molecules, social networks or road maps, in which each point repeatedly gathers information from its neighbors.
  • neuro-symbolic AIneuro-symbolic AI, n.: AI that combines neural networks, which learn patterns from data, with symbolic methods, which follow explicit rules and logic, aiming for systems that both learn flexibly and reason reliably.
  • WikidataWikidata, prop. n.: A free, shared collection of facts about millions of things — people, places, books, chemicals — kept by volunteers in the Wikipedia family and laid out so programs can use it as easily as people.

Sources

  1. Bordes, A., Usunier, N., García-Durán, A., Weston, J. & Yakhnenko, O. (2013). Translating Embeddings for Modeling Multi-relational Data. Advances in Neural Information Processing Systems 26.
  2. Nickel, M., Tresp, V. & Kriegel, H.-P. (2011). A Three-Way Model for Collective Learning on Multi-Relational Data. Proceedings of ICML 2011.

Mentioned in

knowledge graphknowledge graph, n.: A huge web of facts stored as connections between things — 'Paris is the capital of France', 'Marie Curie won a Nobel Prize' — that computers can search, combine and reason over., WikidataWikidata, prop. n.: A free, shared collection of facts about millions of things — people, places, books, chemicals — kept by volunteers in the Wikipedia family and laid out so programs can use it as easily as people., ConceptNetConceptNet, prop. n.: A free, crowd-built network of everyday facts — 'a knife is used for cutting', 'rain makes things wet' — in many languages, which AI systems use to make sense of ordinary situations.