# Low-Rank Subspaces for Unsupervised Entity Linking

arXiv: https://arxiv.org/abs/2104.08737 · HTML: https://mlsanity.com/paper/2104.08737/

## Key facts

- Cited by 4 papers in the map (data through 14 Oct 2026).
- Builds on 20 of its 81 references that are in the map.
- First posted on arXiv 18 Apr 2021.

## Builds on (most cited first, 20 of 20)

- [Inductive Representation Learning on Large Graphs](https://mlsanity.com/paper/1706.02216/) (2017)
- [Distributed Representations of Words and Phrases and their Compositionality](https://mlsanity.com/paper/1310.4546/) (2013)
- [DeepWalk: Online Learning of Social Representations](https://mlsanity.com/paper/1403.6652/) (2014)
- [Simplifying Graph Convolutional Networks](https://mlsanity.com/paper/1902.07153/) (2019)
- [Zero-Shot Entity Linking by Reading Entity Descriptions](https://mlsanity.com/paper/1906.07348/) (2019)
- [End-to-End Neural Entity Linking](https://mlsanity.com/paper/1808.07699/) (2018)
- [Deep Joint Entity Disambiguation with Local Neural Attention](https://mlsanity.com/paper/1704.04920/) (2017)
- [Improving Entity Linking by Modeling Latent Relations between Mentions](https://mlsanity.com/paper/1804.10637/) (2018)
- [Joint Learning of the Embedding of Words and Entities for Named Entity Disambiguation](https://mlsanity.com/paper/1601.01343/) (2016)
- [NetSMF: Large-Scale Network Embedding as Sparse Matrix Factorization](https://mlsanity.com/paper/1906.11156/) (2019)
- [Benchmark and Best Practices for Biomedical Knowledge Graph Embeddings](https://mlsanity.com/paper/2006.13774/) (2020)
- [Distant Learning for Entity Linking with Automatic Noise Detection](https://mlsanity.com/paper/1905.07189/) (2019)
- [Learning Distributed Representations of Texts and Entities from Knowledge Base](https://mlsanity.com/paper/1705.02494/) (2017)
- [Improving Broad-Coverage Medical Entity Linking with Semantic Type Prediction and Large-Scale Datasets](https://mlsanity.com/paper/2005.00460/) (2020)
- [Representing Sentences as Low-Rank Subspaces](https://mlsanity.com/paper/1704.05358/) (2017)
- [Joint Multilingual Supervision for Cross-lingual Entity Linking](https://mlsanity.com/paper/1809.07657/) (2018)
- [Geometry of Polysemy](https://mlsanity.com/paper/1610.07569/) (2016)
- [Representation Learning of Entities and Documents from Knowledge Base Descriptions](https://mlsanity.com/paper/1806.02960/) (2018)
- [Distant Supervision for Entity Linking](https://mlsanity.com/paper/1505.03823/) (2015)
- [Word Embeddings for Entity-annotated Texts](https://mlsanity.com/paper/1902.02078/) (2019)

## Built on by (most cited first, 4 of 4)

- [GenIE: Generative Information Extraction](https://mlsanity.com/paper/2112.08340/) (2021)
- [Entity Insertion in Multilingual Linked Corpora: The Case of Wikipedia](https://mlsanity.com/paper/2410.04254/) (2024)
- [Strong Heuristics for Named Entity Linking](https://mlsanity.com/paper/2207.02824/) (2022)
- [Unsupervised Named Entity Disambiguation for Low Resource Domains](https://mlsanity.com/paper/2412.10054/) (2024)

## Abstract (arXiv)

Entity linking is an important problem with many applications. Most previous solutions were designed for settings where annotated training data is available, which is, however, not the case in numerous domains. We propose a light-weight and scalable entity linking method, Eigenthemes, that relies solely on the availability of entity names and a referent knowledge base. Eigenthemes exploits the fact that the entities that are truly mentioned in a document (the "gold entities") tend to form a semantically dense subset of the set of all candidate entities in the document. Geometrically speaking, when representing entities as vectors via some given embedding, the gold entities tend to lie in a low-rank subspace of the full embedding space. Eigenthemes identifies this subspace using the singular value decomposition and scores candidate entities according to their proximity to the subspace. On the empirical front, we introduce multiple strong baselines that compare favorably to (and sometimes even outperform) the existing state of the art. Extensive experiments on benchmark datasets from a variety of real-world domains showcase the effectiveness of our approach.