Scholarlink

<p>With the widespread adoption of the Internet and the rapid growth of information volume, sentiment analysis technology has become increasingly significant in emotion perception and opinion mining. Mobility as a service (MaaS), as an emerging and not yet widely adopted mobility platform, is highly susceptible to shifts in public perception and market acceptance due to negative online sentiment. There is a lack of systematic solutions for fine-grained sentiment understanding and negative information control targeted at MaaS platforms. To address this, the core contribution of this study is the construction and proposal of an end-to-end integrated framework for public opinion governance on MaaS platforms. This framework combines two core modules: sentiment analysis and negative information propagation blocking, thereby addressing the fragmented application of existing technologies. Within the sentiment analysis module, a hybrid model named BCA-poea, which integrates BERT and convolutional neural networks, is proposed. It enhances emotional feature extraction through an attention mechanism and employs a Softmax classifier to achieve high-precision text sentiment classification. In the negative information propagation control module, the H-DD algorithm is introduced, which combines Degree Discount and H-index to accurately assess node influence, thereby optimizing blocking strategies to curb the spread of negative content. Experimental results validate the effectiveness of the integrated framework. The BCA-poea model demonstrates improvements of 1.54–3.51% in accuracy, recall, and F1-score compared to baseline methods. The H-DD algorithm outperforms classical approaches in both influence suppression and computational efficiency. The integrated framework proposed in this study not only provides a novel technical paradigm for the cross-domain integration of sentiment analysis and information diffusion control, enriching the academic research system for MaaS platform opinion governance, but also offers platform operators comprehensive, integrated technical support for full-process public opinion management. It holds significant practical value for promoting the market adoption and sustainable development of MaaS platforms.</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

Research on sentiment analysis and influence maximization algorithms for MaaS platforms

  • 期刊论文

  • 发表日期：

    2026年3月11日

查看全文
  • Yulong Yin,
  • Xiang Zhang,
  • Jingwen Li,
  • Shenfan Zhang,
  • Ning Ma

摘要

With the widespread adoption of the Internet and the rapid growth of information volume, sentiment analysis technology has become increasingly significant in emotion perception and opinion mining. Mobility as a service (MaaS), as an emerging and not yet widely adopted mobility platform, is highly susceptible to shifts in public perception and market acceptance due to negative online sentiment. There is a lack of systematic solutions for fine-grained sentiment understanding and negative information control targeted at MaaS platforms. To address this, the core contribution of this study is the construction and proposal of an end-to-end integrated framework for public opinion governance on MaaS platforms. This framework combines two core modules: sentiment analysis and negative information propagation blocking, thereby addressing the fragmented application of existing technologies. Within the sentiment analysis module, a hybrid model named BCA-poea, which integrates BERT and convolutional neural networks, is proposed. It enhances emotional feature extraction through an attention mechanism and employs a Softmax classifier to achieve high-precision text sentiment classification. In the negative information propagation control module, the H-DD algorithm is introduced, which combines Degree Discount and H-index to accurately assess node influence, thereby optimizing blocking strategies to curb the spread of negative content. Experimental results validate the effectiveness of the integrated framework. The BCA-poea model demonstrates improvements of 1.54–3.51% in accuracy, recall, and F1-score compared to baseline methods. The H-DD algorithm outperforms classical approaches in both influence suppression and computational efficiency. The integrated framework proposed in this study not only provides a novel technical paradigm for the cross-domain integration of sentiment analysis and information diffusion control, enriching the academic research system for MaaS platform opinion governance, but also offers platform operators comprehensive, integrated technical support for full-process public opinion management. It holds significant practical value for promoting the market adoption and sustainable development of MaaS platforms.