Enhancing Sentiment Analysis: A Novel Machine Learning Framework Integrating Textual Features and Document Metadata

Abstract

Introduction: In general, sentiment evaluation has focused on text content analysis. Traditional sentiment analysis (SA) excludes document metadata and focuses on text. The lack of research in this area prevents textual material and metadata from working together to create synergies that could lead to reliable sentiment assessment solutions.
Method: We exhibit the potential uses of our integrated methodology across multiple areas and prove its efficacy through testing and evaluation on a different datasets. Data preprocessed using a multiple techniques, such as “stemming, tokenization, and stop word removal”. The “term frequency-inverse document frequency (TF-IDF)”methodology is utilized for feature extraction.
Result: The outcomes of our experiments reveal that our suggested method is successful in collecting sensitive sentiment such as” accuracy, recall, precision, and f1-score”. This is a promising development in the field of sentiment assessment exploration.
Conclusion: In conclusion, this study proposed a novel approach, the DDO-SVM, which addresses the limitations of conventional SA by incorporating both textual and metadata features. These components integrate well, improving sentiment evaluation across measurements. SA becomes more reliable and effective in actual applications with this breakthrough.