Emotion Analysis Based on Neural Network under the Big Data Environment.

Area of Science:

• Natural Language Processing
• Machine Learning
• Artificial Intelligence

Background:

• Traditional sentiment analysis methods struggle with nuanced emotional tendencies and effective utilization of syntactic information.
• Big data presents unique challenges for accurate sentiment classification due to volume and complexity.

Purpose of the Study:

• To develop an advanced big data sentiment analysis method using neural networks.
• To improve the accuracy and efficiency of sentiment prediction, especially for ambiguous texts.
• To enhance the utilization of syntactic and semantic information in sentiment analysis.

Main Methods:

• Utilized the BERT model for data vectorization to minimize semantic loss.
• Employed a bidirectional LSTM encoder to extract features from word vectors.
• Incorporated an attention layer to generate a final feature vector, reducing noise from irrelevant data.

Main Results:

• Achieved high accuracy, recall, and F1 scores in sentiment classification tasks.
• Demonstrated significant improvement in fine-grained sentiment classification for ambiguous texts.
• Effectively reduced the influence of irrelevant data through the attention mechanism.

Conclusions:

• The proposed neural network-based method offers a robust solution for big data sentiment analysis.
• The integration of BERT, bidirectional LSTM, and attention mechanisms enhances the prediction of emotional tendencies.
• This approach effectively addresses limitations in existing methods, particularly for complex and ambiguous textual data.