Deriving Word Vectors from Contextualized Language Models using Topic-Aware Mention Selection

TL;DR

This paper introduces a novel method for deriving word vectors from contextualized language models by partitioning contexts with a topic model and selecting vectors with task-specific supervision, resulting in improved semantic prediction.

Contribution

It proposes a new approach combining CLMs and topic models to generate more semantically meaningful word representations, differing from standard embeddings.

Findings

Word vectors are more predictive of semantic properties.

Topic-aware vectors outperform standard embeddings.

Supervised selection enhances vector quality.

Abstract

One of the long-standing challenges in lexical semantics consists in learning representations of words which reflect their semantic properties. The remarkable success of word embeddings for this purpose suggests that high-quality representations can be obtained by summarizing the sentence contexts of word mentions. In this paper, we propose a method for learning word representations that follows this basic strategy, but differs from standard word embeddings in two important ways. First, we take advantage of contextualized language models (CLMs) rather than bags of word vectors to encode contexts. Second, rather than learning a word vector directly, we use a topic model to partition the contexts in which words appear, and then learn different topic-specific vectors for each word. Finally, we use a task-specific supervision signal to make a soft selection of the resulting vectors. We show that this simple strategy leads to high-quality word vectors, which are more predictive of semantic properties than word embeddings and existing CLM-based strategies.