Scholarlink

This work explores multivariate functional graphical models, focusing on their application to dynamic social media analysis, specifically Twitter debates such as the UK Brexit discourse. By introducing a novel extension considering spatially and temporally correlated random functions, the study constructs semantic networks through graphical representation of conditional dependence. Incorporating spatial and temporal information, the methodology captures semantic changes during significant events. Employing PARAFAC decomposition on the estimated semantic networks reveals latent factors in word usage over time, enhancing interpretability and identifying key themes in debates. The effectiveness of the methodology will be demostrated analyzing the Brexit debate on Twitter.

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

Functional Graphical Models for Corpus Linguistics in Social Media

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Nicola Pronello,
  • Emiliano del Gobbo,
  • Lara Fontanella,
  • Rosaria Ignaccolo,
  • Luigi Ippoliti,
  • Sara Fontanella

摘要

This work explores multivariate functional graphical models, focusing on their application to dynamic social media analysis, specifically Twitter debates such as the UK Brexit discourse. By introducing a novel extension considering spatially and temporally correlated random functions, the study constructs semantic networks through graphical representation of conditional dependence. Incorporating spatial and temporal information, the methodology captures semantic changes during significant events. Employing PARAFAC decomposition on the estimated semantic networks reveals latent factors in word usage over time, enhancing interpretability and identifying key themes in debates. The effectiveness of the methodology will be demostrated analyzing the Brexit debate on Twitter.