GloVeInit at SemEval-2020 Task 1: Using GloVe Vector Initialization for Unsupervised Lexical Semantic Change Detection

TL;DR

This paper introduces a GloVe vector initialization method for unsupervised lexical semantic change detection, improving alignment of embeddings across time and achieving competitive rankings in SemEval-2020 tasks.

Contribution

It proposes using GloVe vector initialization for better temporal embedding alignment, a novel approach in lexical semantic change detection.

Findings

GloVe embeddings outperform SGNS for this task.

The method ranks 13th and 10th in the SemEval-2020 subtasks.

Hyperparameter tuning impacts performance significantly.

Abstract

This paper presents a vector initialization approach for the SemEval2020 Task 1: Unsupervised Lexical Semantic Change Detection. Given two corpora belonging to different time periods and a set of target words, this task requires us to classify whether a word gained or lost a sense over time (subtask 1) and to rank them on the basis of the changes in their word senses (subtask 2). The proposed approach is based on using Vector Initialization method to align GloVe embeddings. The idea is to consecutively train GloVe embeddings for both corpora, while using the first model to initialize the second one. This paper is based on the hypothesis that GloVe embeddings are more suited for the Vector Initialization method than SGNS embeddings. It presents an intuitive reasoning behind this hypothesis, and also talks about the impact of various factors and hyperparameters on the performance of the proposed approach. Our model ranks 13th and 10th among 33 teams in the two subtasks. The implementation has been shared publicly.