Enhanced Word Embedding with CNN using Word Order Information For Sentiment Analysis

On several e-commerce platforms, internet users share opinions. Understanding people's sentiments and opinions is necessary. GloVe, which uses word contexts and matrix vectorization, is an effective word vector learning method. Vector-learning systems have improved with this method. However, the GloVe model ignores word order in context. This paper presents Positional Embedding, which integrates word order into GloVe word embeddings. We found that our Word Order Vector (PEWOVe) word embeddings method outperforms GloVe in sentiment classification. Amazon data was used to test the proposed technique. The proposed model is 2.7
^%
, 2.5%, and 2.1 % more accurate than baseline models on GE, GEC, and GE+PE.