# Enhancing Domain Word Embedding via Latent Semantic Imputation

## TL;DR

This paper introduces Latent Semantic Imputation, a novel approach that enhances word embeddings by integrating external knowledge and spectral graph methods, significantly improving performance on language tasks.

## Contribution

The paper proposes a new spectral graph-based method to impute embeddings for low-frequency words, improving semantic representations in word embedding models.

## Key findings

- LSI outperforms benchmark embeddings in classification and language modeling tasks.
- The method effectively imputes embeddings for low-frequency words.
- Results are consistent across different parameter settings.

## Abstract

We present a novel method named Latent Semantic Imputation (LSI) to transfer external knowledge into semantic space for enhancing word embedding. The method integrates graph theory to extract the latent manifold structure of the entities in the affinity space and leverages non-negative least squares with standard simplex constraints and power iteration method to derive spectral embeddings. It provides an effective and efficient approach to combining entity representations defined in different Euclidean spaces. Specifically, our approach generates and imputes reliable embedding vectors for low-frequency words in the semantic space and benefits downstream language tasks that depend on word embedding. We conduct comprehensive experiments on a carefully designed classification problem and language modeling and demonstrate the superiority of the enhanced embedding via LSI over several well-known benchmark embeddings. We also confirm the consistency of the results under different parameter settings of our method.