Scholarlink

<p>Detecting depression through sentiment classification uses natural language processing to analyze and interpret emotional text. This approach enables the early identification of depressive symptoms by examining language patterns. However, it includes challenges, like interpreting nuanced language and accounting for cultural and contextual differences. To handle these challenges, an effective model called Puzzle Aquila Optimization enabled Capsule Convolutional Neural Network (PAO_Capsule_CNN) is introduced for depression detection with sentiment classification. Initially, input review data undergo Bidirectional Encoder Representations from Transformers (BERT) tokenization to break down the text into number of tokens. These tokens are then processed by Aspect Term Frequency (ATE) to identify aspect terms within the texts. Subsequently, the identified aspect terms undergo feature extraction to extract significant features. Then, sentiment classification is conducted by utilizing Hierarchical Deep Learning for Text Classification (HDLTex) model based on these extracted features. Finally, depression detection is performed utilizing a hybrid PAO_Capsule_CNN model. This hybrid model integrates Capsule Network and CNN. The Capsule_CNN model is trained using the Puzzle Aquila Optimization (PAO) approach, which combines the Puzzle Optimization Algorithm (POA) and Aquila Optimizer (AO). The proposed PAO_Capsule_CNN has attained the accuracy of 92.298%, True-Negative Rate (TNR) of 93.238%, and True-Positive Rate (TPR) of 91.699%.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Puzzle Aquila Optimization Enabled Capsule Convolutional Neural Network for Depression Detection with Sentiment Classification

  • 期刊论文

  • 发表日期：

    2025年7月15日

查看全文
  • B. Shibi,
  • S. Ravikumar

摘要

Detecting depression through sentiment classification uses natural language processing to analyze and interpret emotional text. This approach enables the early identification of depressive symptoms by examining language patterns. However, it includes challenges, like interpreting nuanced language and accounting for cultural and contextual differences. To handle these challenges, an effective model called Puzzle Aquila Optimization enabled Capsule Convolutional Neural Network (PAO_Capsule_CNN) is introduced for depression detection with sentiment classification. Initially, input review data undergo Bidirectional Encoder Representations from Transformers (BERT) tokenization to break down the text into number of tokens. These tokens are then processed by Aspect Term Frequency (ATE) to identify aspect terms within the texts. Subsequently, the identified aspect terms undergo feature extraction to extract significant features. Then, sentiment classification is conducted by utilizing Hierarchical Deep Learning for Text Classification (HDLTex) model based on these extracted features. Finally, depression detection is performed utilizing a hybrid PAO_Capsule_CNN model. This hybrid model integrates Capsule Network and CNN. The Capsule_CNN model is trained using the Puzzle Aquila Optimization (PAO) approach, which combines the Puzzle Optimization Algorithm (POA) and Aquila Optimizer (AO). The proposed PAO_Capsule_CNN has attained the accuracy of 92.298%, True-Negative Rate (TNR) of 93.238%, and True-Positive Rate (TPR) of 91.699%.