Scholarlink

<p>The World Wide Web facilitates rapid global information dissemination, underscoring the need for digital preservation. While existing frameworks emphasize methods like emulation and migration, they often overlook the credibility of preserved content. This study addresses this gap by identifying and analyzing 14 key credibility factors (CFs) in digital news using a hybrid methodology involving multivocal literature review, expert survey, interpretive structural modeling (ISM), and BERT-based semantic analysis. The ISM revealed a highly interconnected structure where all CFs exhibit mutual influence, with factors such as “reputation of the content creator” and “frequent sharing” demonstrating high driving power. Matrix impact cross-reference multiplication applied to a classification analysis further categorized CFs into driving, dependent, and linkage types. Complementarily, bidirectional encoder representations from transformers (BERT) uncovered semantic relationships among CFs, identifying strong correlations like those between “number of views” and “frequent sharing.” A comparative analysis confirmed the complementarity of ISM's structural insights and BERT’s contextual understanding. These findings offer a robust framework for enhancing digital credibility evaluation and inform future preservation strategies.</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

A dual-modal analysis of credibility in integrating interpretive structural modeling (ISM) and BERT for enhanced fake news detection

  • 期刊论文

  • 发表日期：

    2025年8月21日

查看全文
  • Muhammad Faisal Abrar,
  • Ali Alferaidi,
  • Tarqi S. Almurayziq,
  • Muhammad Saqib,
  • Raza Uddin,
  • Wilayat Khan,
  • Jawad Khan,
  • Mohammad Alsaffar

摘要

The World Wide Web facilitates rapid global information dissemination, underscoring the need for digital preservation. While existing frameworks emphasize methods like emulation and migration, they often overlook the credibility of preserved content. This study addresses this gap by identifying and analyzing 14 key credibility factors (CFs) in digital news using a hybrid methodology involving multivocal literature review, expert survey, interpretive structural modeling (ISM), and BERT-based semantic analysis. The ISM revealed a highly interconnected structure where all CFs exhibit mutual influence, with factors such as “reputation of the content creator” and “frequent sharing” demonstrating high driving power. Matrix impact cross-reference multiplication applied to a classification analysis further categorized CFs into driving, dependent, and linkage types. Complementarily, bidirectional encoder representations from transformers (BERT) uncovered semantic relationships among CFs, identifying strong correlations like those between “number of views” and “frequent sharing.” A comparative analysis confirmed the complementarity of ISM's structural insights and BERT’s contextual understanding. These findings offer a robust framework for enhancing digital credibility evaluation and inform future preservation strategies.