WOVe: Incorporating Word Order in GloVe Word Embeddings

Word vector representations open up new opportunities to extract useful information from unstructured text. Defining a word as made it easy for the machine learning algorithms understand text and from. have been used in many applications such synonyms, analogy, syntactic parsing, others. GloVe, based on contexts matrix vectorization, is an effective vector-learning algorithm. It improves previous algorithms. However, GloVe model fails explicitly consider order which words appear within their contexts. In this paper, multiple methods of incorporating embeddings are proposed. Experimental results show that our Order Vector (WOVe) approach outperforms unmodified natural language tasks analogy completion similarity. WOVe with direct concatenation slightly outperformed similarity task, increasing average rank by 2%. greatly improved baseline achieving 36.34% improvement accuracy.