A novel LSTM-CNN-grid search-based deep neural network for sentiment analysis.

Area of Science:

• Natural Language Processing
• Machine Learning
• Artificial Intelligence

Background:

• The rapid growth of internet users generates vast amounts of user-generated content.
• Analyzing opinions, sentiments, and emotions in text is vital for social media, brand monitoring, customer service, and market research.
• Sentiment analysis is key to understanding user attitudes and emotions expressed online.

Purpose of the Study:

• To develop a novel deep neural network model for enhanced sentiment analysis.
• To evaluate the proposed model against various baseline algorithms using multiple datasets.
• To demonstrate the effectiveness of hyperparameter optimization in improving sentiment analysis performance.

Main Methods:

• Proposed a novel deep neural network model integrating Long Short-Term Memory (LSTM) and Convolutional Neural Networks (CNN) with grid search for hyperparameter optimization.
• Evaluated baseline algorithms including Convolutional Neural Networks (CNN), K-nearest neighbor, LSTM, neural networks, LSTM-CNN, and CNN-LSTM.
• Assessed model performance using metrics such as accuracy, precision, sensitivity, specificity, and F-1 score across multiple datasets.

Main Results:

• The proposed LSTM-CNN grid search-based deep neural network model significantly outperformed baseline algorithms.
• Achieved an overall accuracy exceeding 96% on multiple sentiment analysis datasets.
• Hyperparameter optimization through grid search was crucial for the model's superior performance.

Conclusions:

• The novel LSTM-CNN grid search model offers a highly effective approach to sentiment analysis.
• This method provides superior accuracy in understanding user sentiment from diverse online text sources.
• The findings highlight the importance of advanced deep learning architectures and optimization techniques for sentiment analysis tasks.