Unenhanced Sparse Vector-based Embedding Method for Sentiment Analysis

Natural language processing is one of the most trending fields in research, with sentiment analysis being one of the well-known problems in the field. Many methods have been proposed to handle text-based sentiment data, with social networks acting as one of the main data sources and research targets. An important step in designing a text-based model is the embedding method, which helps in the representation of the inputs. This study presents a novel static text embedding method to represent text inputs and compares its sentiment classification performance with some well-known text embedding methods. The results are on par with existing embedding methods, achieving a promising classification accuracy of 90.66%.