SentiFuse: Deep Multi-model Fusion Framework for Robust Sentiment Extraction

2026-02-01

AI Summary

This work addresses the limitations of existing sentiment analysis approaches, which lack a unified framework for effectively integrating heterogeneous models, thereby constraining performance and robustness. To overcome this, the authors propose SentiFuse—a model-agnostic, deep multimodal fusion framework that employs a standardized interface layer to accommodate arbitrary heterogeneous sentiment models. SentiFuse integrates feature-level, decision-level, and adaptive fusion strategies within a flexible and unified architecture. Experimental results on three benchmark datasets—Crowdflower, GoEmotions, and Sentiment140—demonstrate that SentiFuse significantly outperforms both individual models and naive ensemble methods, achieving up to a 4% absolute improvement in F1 score through feature-level fusion. Notably, the framework exhibits enhanced robustness, particularly in handling complex emotional expressions.

Technology Category

Intelligent Robots: Multimodal Perception & Sensor FusionMachine Learning: Multimodal LearningNatural Language Processing: Interpretability, Analysis, and Evaluation of NLP Models

Application Category

Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMsWeb Mining and Content Analysis: Sentiment analysis and opinion miningGraph Algorithms and Modeling for the Web: Algorithms and analysis for heterogeneous, signed, attributed, multi-relational, temporal, higher-order, and annotated Web-related graphs

Abstract

Sentiment analysis models exhibit complementary strengths, yet existing approaches lack a unified framework for effective integration. We present SentiFuse, a flexible and model-agnostic framework that integrates heterogeneous sentiment models through a standardization layer and multiple fusion strategies. Our approach supports decision-level fusion, feature-level fusion, and adaptive fusion, enabling systematic combination of diverse models. We conduct experiments on three large-scale social-media datasets: Crowdflower, GoEmotions, and Sentiment140. These experiments show that SentiFuse consistently outperforms individual models and naive ensembles. Feature-level fusion achieves the strongest overall effectiveness, yielding up to 4\% absolute improvement in F1 score over the best individual model and simple averaging, while adaptive fusion enhances robustness on challenging cases such as negation, mixed emotions, and complex sentiment expressions. These results demonstrate that systematically leveraging model complementarity yields more accurate and reliable sentiment analysis across diverse datasets and text types.

Problem

• sentiment analysis
• model fusion
• heterogeneous models
• robust sentiment extraction
• multi-model integration

Innovation

• multi-model fusion
• sentiment analysis
• feature-level fusion
• adaptive fusion
• model-agnostic framework