Improved Knowledge Graph Embedding Using Background Taxonomic Information
Abstract
Knowledge graphs are used to represent relational information in terms of triples. To enable learning about domains, embedding models, such as tensor factorization models, can be used to make predictions of new triples. Often there is background taxonomic information (in terms of subclasses and subproperties) that should also be taken into account. We show that existing fully expressive (a.k.a. universal) models cannot provably respect subclass and subproperty information. We show that minimal modifications to an existing knowledge graph completion method enables injection of taxonomic information. Moreover, we prove that our model is fully expressive, assuming a lower-bound on the size of the embeddings. Experimental results on public knowledge graphs show that despite its simplicity our approach is surprisingly effective.
Cite
Text
Fatemi et al. "Improved Knowledge Graph Embedding Using Background Taxonomic Information." AAAI Conference on Artificial Intelligence, 2019. doi:10.1609/AAAI.V33I01.33013526Markdown
[Fatemi et al. "Improved Knowledge Graph Embedding Using Background Taxonomic Information." AAAI Conference on Artificial Intelligence, 2019.](https://mlanthology.org/aaai/2019/fatemi2019aaai-improved/) doi:10.1609/AAAI.V33I01.33013526BibTeX
@inproceedings{fatemi2019aaai-improved,
title = {{Improved Knowledge Graph Embedding Using Background Taxonomic Information}},
author = {Fatemi, Bahare and Ravanbakhsh, Siamak and Poole, David},
booktitle = {AAAI Conference on Artificial Intelligence},
year = {2019},
pages = {3526-3533},
doi = {10.1609/AAAI.V33I01.33013526},
url = {https://mlanthology.org/aaai/2019/fatemi2019aaai-improved/}
}