Redefining Context Windows for Word Embedding Models: An Experimental Study

2017/04/19 by Lison, Pierre, Kutuzov, Andrey
#Computation and Language (cs.CL) #FOS: Computer and information sciences

paper · doi:10.48550/arxiv.1704.05781

Abstract

Distributional semantic models learn vector representations of words through the contexts they occur in. Although the choice of context (which often takes the form of a sliding window) has a direct influence on the resulting embeddings, the exact role of this model component is still not fully understood. This paper presents a systematic analysis of context windows based on a set of four distinct hyper-parameters. We train continuous Skip-Gram models on two English-language corpora for various combinations of these hyper-parameters, and evaluate them on both lexical similarity and analogy tasks. Notable experimental results are the positive impact of cross-sentential contexts and the surprisingly good performance of right-context windows.

Related

• The Role of Context Types and Dimensionality in Learning Word Embeddings
• Syntactic Interchangeability in Word Embedding Models
• How Contextual are Contextualized Word Representations? Comparing the\n Geometry of BERT, ELMo, and GPT-2 Embeddings
• Exploring Context Window of Large Language Models via Decomposed Positional Vectors
• Low-dimensional Semantic Space: from Text to Word Embedding