AI-Driven Ensemble for Enhanced Sentiment Polarity Detection in Movie Reviews

This paper presents a novel AI-driven ensembleapproach for discerning sentiment polarity in text documents,specifically consumer reviews. We address the binary classifica-tion problem of identifying positive versus negative sentimentby proposing a uniquely hybrid framework that integratesgenerative, discriminative, and deep embedding-based models.Our key contribution is a carefully designed, optimized weightedvoting mechanism that leverages cross-validation to assignmodel-specific weights, effectively harnessing the complementarystrengths of its diverse constituents. This ensemble strategy isevaluated on a widely recognized movie review dataset, whereit demonstrates robust and superior performance compared tostate-of-the-art standalone models. The findings confirm that ourmulti-paradigm fusion leads to significant gains in accuracy, ad-vancing the capabilities of automated sentiment analysis systemsby mitigating the individual limitations of each model family.