Enhanced Frequency-Based Word Embedding Approach For Text Analysis: Comparative Study With Existing Method

Text classification is vital in natural language processing (NLP) and has diverse applications. A new approach, Relative Word Frequency Ratio (RWFR), is introduced for text classification. RWFR utilizes positive and negative word frequencies in a text corpus to create a feature vector for logistic regression. The method's effectiveness is compared to Term Frequency-Inverse Document Frequency (TF-IDF) and CountVectorizer on Reuters-21578 and 20 Newsgroup datasets. RWFR, TF-IDF, and CountVectorizer-derived feature vectors are used to train classifiers, with performance assessed using standard metrics.Experimental results show RWFR achieves notable classification accuracy, akin to TF-IDF and CountVectorizer. RWFR exhibits promising F1-score and recall for text classification. This suggests RWFR as a viable alternative for feature extraction and logistic regression-based classification. RWFR's uniqueness lies in considering relative positive and negative word frequencies, capturing sentiment information for accurate classification. In comparison to TF-IDF and CountVectorizer, RWFR demonstrates competitive performance, highlighting its potential for various NLP applications.