Scholarlink

<p>Emotion detection stands as a pivotal approach for extracting unbiased insights from textual documents. This process empowers business experts to identify areas of improvement and bolster key facets of their operations. Given that every online customer post holds the potential to influence the market and consumer behaviour, the significance of emotion recognition has surged. This article introduces a cutting-edge deep learning (DL) approach for emotion detection, specifically focusing on analyzing tweet text sourced from the CARER dataset. The proposed method employs a two-step process, utilizing sentence-transformer for text embedding and Legendre Memory Unit (LMU) based DL framework for emotion classification. The LMU is a novel memory cell designed for recurrent neural networks (RNNs). The key innovation of LMU lies in its ability to achieve “orthogonalization" of its continuous-time history. This means that the memory unit effectively processes input data in a way that retains important patterns while reducing interference from irrelevant or noisy information. The study has undertaken an exhaustive comparison with various state-of-the-art classification models, including RNN, long short term memory (LSTM), bidirectional-long short term memory (Bi-LSTM), gated recurrent unit (GRU), and bidirectional gated recurrent unit (Bi-GRU). Notably, the proposed method distinguishes itself by demonstrating superior performance, surpassing established benchmarks in the field.</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

Emotion detection in text using a legendre memory unit based deep learning framework

  • 期刊论文

  • 发表日期：

    2025年5月7日

查看全文
  • Abrar Khan,
  • Prabhat Kumar

摘要

Emotion detection stands as a pivotal approach for extracting unbiased insights from textual documents. This process empowers business experts to identify areas of improvement and bolster key facets of their operations. Given that every online customer post holds the potential to influence the market and consumer behaviour, the significance of emotion recognition has surged. This article introduces a cutting-edge deep learning (DL) approach for emotion detection, specifically focusing on analyzing tweet text sourced from the CARER dataset. The proposed method employs a two-step process, utilizing sentence-transformer for text embedding and Legendre Memory Unit (LMU) based DL framework for emotion classification. The LMU is a novel memory cell designed for recurrent neural networks (RNNs). The key innovation of LMU lies in its ability to achieve “orthogonalization" of its continuous-time history. This means that the memory unit effectively processes input data in a way that retains important patterns while reducing interference from irrelevant or noisy information. The study has undertaken an exhaustive comparison with various state-of-the-art classification models, including RNN, long short term memory (LSTM), bidirectional-long short term memory (Bi-LSTM), gated recurrent unit (GRU), and bidirectional gated recurrent unit (Bi-GRU). Notably, the proposed method distinguishes itself by demonstrating superior performance, surpassing established benchmarks in the field.