# Representing Sentences as Low-Rank Subspaces

arXiv: https://arxiv.org/abs/1704.05358 · HTML: https://mlsanity.com/paper/1704.05358/

## Key facts

- Cited by 7 papers in the map (data through 14 Oct 2026).
- First posted on arXiv 18 Apr 2017.

## Built on by (most cited first, 7 of 7)

- [Fine-Grained Image Analysis with Deep Learning: A Survey](https://mlsanity.com/paper/2111.06119/) (2021)
- [Effects of Pre- and Post-Processing on type-based Embeddings in Lexical Semantic Change Detection](https://mlsanity.com/paper/2101.09368/) (2021)
- [Low-Rank Subspaces for Unsupervised Entity Linking](https://mlsanity.com/paper/2104.08737/) (2021)
- [A prototype-based model for set classification](https://mlsanity.com/paper/2408.13720/) (2024)
- [Aligning Instance-Semantic Sparse Representation towards Unsupervised Object Segmentation and Shape Abstraction with Repeatable Primitives](https://mlsanity.com/paper/2503.06947/) (2025)
- [Socially Fair Center-based and Linear Subspace Clustering](https://mlsanity.com/paper/2208.10095/) (2022)
- [Improving Imbalanced Text Classification with Dynamic Curriculum Learning](https://mlsanity.com/paper/2210.14724/) (2022)

## Abstract (arXiv)

Sentences are important semantic units of natural language. A generic, distributional representation of sentences that can capture the latent semantics is beneficial to multiple downstream applications. We observe a simple geometry of sentences -- the word representations of a given sentence (on average 10.23 words in all SemEval datasets with a standard deviation 4.84) roughly lie in a low-rank subspace (roughly, rank 4). Motivated by this observation, we represent a sentence by the low-rank subspace spanned by its word vectors. Such an unsupervised representation is empirically validated via semantic textual similarity tasks on 19 different datasets, where it outperforms the sophisticated neural network models, including skip-thought vectors, by 15% on average.