# On Embeddings in Relational Databases

arXiv: https://arxiv.org/abs/2005.06437 · HTML: https://mlsanity.com/paper/2005.06437/

## Key facts

- Cited by 2 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 13 May 2020.

## Built on by (most cited first, 2 of 2)

- [ML Based Lineage in Databases](https://mlsanity.com/paper/2109.06339/) (2021)
- [The Hybrid Multimodal Graph Index (HMGI): A Comprehensive Framework for Integrated Relational and Vector Search](https://mlsanity.com/paper/2510.10123/) (2025)

## Abstract (arXiv)

We address the problem of learning a distributed representation of entities in a relational database using a low-dimensional embedding. Low-dimensional embeddings aim to encapsulate a concise vector representation for an underlying dataset with minimum loss of information. Embeddings across entities in a relational database have been less explored due to the intricate data relations and representation complexity involved. Relational databases are an inter-weaved collection of relations that not only model relationships between entities but also record complex domain-specific quantitative and temporal attributes of data defining complex relationships among entities. Recent methods for learning an embedding constitute of a naive approach to consider complete denormalization of the database by materializing the full join of all tables and representing as a knowledge graph. This popular approach has certain limitations as it fails to capture the inter-row relationships and additional semantics encoded in the relational databases. In this paper we demonstrate; a better methodology for learning representations by exploiting the underlying semantics of columns in a table while using the relation joins and the latent inter-row relationships. Empirical results over a real-world database with evaluations on similarity join and table completion tasks support our proposition.