Reflection-based Word Attribute Transfer

TL;DR

This paper introduces a reflection-based method for transferring word attributes in embeddings, avoiding costly analogy operations and effectively changing attributes like gender without altering unrelated words.

Contribution

The paper proposes a novel reflection mapping approach for word attribute transfer that does not rely on analogy operations, simplifying the process and reducing knowledge requirements.

Findings

Effective attribute transfer without changing unrelated words

Outperforms analogy-based methods in attribute transfer tasks

Requires less prior knowledge about word attributes

Abstract

Word embeddings, which often represent such analogic relations as king - man + woman = queen, can be used to change a word's attribute, including its gender. For transferring king into queen in this analogy-based manner, we subtract a difference vector man - woman based on the knowledge that king is male. However, developing such knowledge is very costly for words and attributes. In this work, we propose a novel method for word attribute transfer based on reflection mappings without such an analogy operation. Experimental results show that our proposed method can transfer the word attributes of the given words without changing the words that do not have the target attributes.