Scholarlink

<p>The increase in the development of online products has paved the way for sharing people’s opinions on Amazon products. However, reading all the unstructured textual reviews and complicated consumer feedback is time-consuming, and to overcome this issue, an innovative deep learning model is introduced for investigating the sentiments using Amazon product reviews. At first, preprocessing is carried out on the input review data by performing lowercase conversion, stop word removal, punctuation removal, tokenization, and lemmatization processes. Then, the adaptive term frequency-inverse&#xa0;document&#xa0;frequency is introduced for transforming the text into a vector representation. In addition, the combined golden jackal optimizer with the salmon migration optimizer resulted in an improved golden jackal optimizer being introduced for tuning the weights of the adaptive term frequency-inverse document frequency technique. Lastly, the inception recurrent residual convolutional neural network is introduced to classify product reviews’ sentiment as positive, negative, and neutral. The simulation process is carried out using the Python platform, and a publicly available Amazon product review database is considered for the training process. The effectiveness of the proposed method is evaluated based on several assessment measures, including classification accuracy, Mathew’s correlation coefficient, kappa coefficient, positive predictive value, and computational time. The results demonstrate that the proposed approach achieves a classification accuracy of 94.9 percent, a Mathew’s correlation coefficient of 93 percent, a kappa coefficient of 94 percent, a positive predictive value of 94.8 percent, and a computation time of 103.93&#xa0;s.</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

Sentiment analysis of Amazon product reviews using an inception-based recurrent residual CNN approach

  • 期刊论文

  • 发表日期：

    2025年12月15日

查看全文
  • K. Balasaranya,
  • P. Ezhumalai

摘要

The increase in the development of online products has paved the way for sharing people’s opinions on Amazon products. However, reading all the unstructured textual reviews and complicated consumer feedback is time-consuming, and to overcome this issue, an innovative deep learning model is introduced for investigating the sentiments using Amazon product reviews. At first, preprocessing is carried out on the input review data by performing lowercase conversion, stop word removal, punctuation removal, tokenization, and lemmatization processes. Then, the adaptive term frequency-inverse document frequency is introduced for transforming the text into a vector representation. In addition, the combined golden jackal optimizer with the salmon migration optimizer resulted in an improved golden jackal optimizer being introduced for tuning the weights of the adaptive term frequency-inverse document frequency technique. Lastly, the inception recurrent residual convolutional neural network is introduced to classify product reviews’ sentiment as positive, negative, and neutral. The simulation process is carried out using the Python platform, and a publicly available Amazon product review database is considered for the training process. The effectiveness of the proposed method is evaluated based on several assessment measures, including classification accuracy, Mathew’s correlation coefficient, kappa coefficient, positive predictive value, and computational time. The results demonstrate that the proposed approach achieves a classification accuracy of 94.9 percent, a Mathew’s correlation coefficient of 93 percent, a kappa coefficient of 94 percent, a positive predictive value of 94.8 percent, and a computation time of 103.93 s.