Scholarlink

<p>As hidden information discovery is fundamental in AI research, natural language researchers have become interested in extracting meaningful information from social networks. One source of such hidden information is irony in text, which has grown significantly on these platforms. Since social media allows anyone to publish any content, understanding irony is crucial for grasping the true meaning of a text. Detecting irony and identifying authors of ironic texts can be instrumental in analyzing customer reviews, gauging public sentiment about brands or events, and uncovering implicit feedback. This research proposes a novel method that utilizes clustering of user writings with importance-based weighting, along with transformer-based models, to detect authors with a penchant for ironic expression. Recognizing the conversational nature of social media text, the method incorporates pre-processing techniques to enhance detection accuracy. Evaluation on a Twitter dataset demonstrates the efficacy of the proposed method, achieving a 98% accuracy rate in classifying authors as ironic or non-ironic.</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

Improving irony speech spreaders profiling on social networks using clustering & transformer based models

  • 期刊论文

  • 发表日期：

    2025年2月14日

查看全文
  • Leila Hazrati,
  • Alireza Sokhandan,
  • Leili Farzinvash

摘要

As hidden information discovery is fundamental in AI research, natural language researchers have become interested in extracting meaningful information from social networks. One source of such hidden information is irony in text, which has grown significantly on these platforms. Since social media allows anyone to publish any content, understanding irony is crucial for grasping the true meaning of a text. Detecting irony and identifying authors of ironic texts can be instrumental in analyzing customer reviews, gauging public sentiment about brands or events, and uncovering implicit feedback. This research proposes a novel method that utilizes clustering of user writings with importance-based weighting, along with transformer-based models, to detect authors with a penchant for ironic expression. Recognizing the conversational nature of social media text, the method incorporates pre-processing techniques to enhance detection accuracy. Evaluation on a Twitter dataset demonstrates the efficacy of the proposed method, achieving a 98% accuracy rate in classifying authors as ironic or non-ironic.