Scholarlink

Identifying crisis callers on psychological assistance hotlines is crucial for providing timely and effective support. To achieve this, the Crisis Psychological Hotline Corpus (CPHC) dataset was developed for suicidal crisis recognition. Statistical analysis reveals that non-crisis callers have significantly longer speech durations compared to operators, while there is no significant difference in speech duration between mild and severe crisis situations. The results show that the model achieves high accuracy and consistency in recognizing callers at different crisis levels, with an F1 score and accuracy of about 0.92 on the test set. Future research could expand the dataset, optimize the automated screening and identification system, and explore the detection of other mental health conditions to enhance the model’s effectiveness in psychological assistance and suicide prevention.

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

Psychological Crisis Detection in Crisis Hotline Chats Based on Weighted-Bidirectional Long Short-Term Neural Network

  • 会议论文

  • 发表日期：

    2025年

查看全文
  • Zhong Ding,
  • Yang Zhou,
  • Baoliang Zhong,
  • Chenling Liu,
  • Zhentao Liu

摘要

Identifying crisis callers on psychological assistance hotlines is crucial for providing timely and effective support. To achieve this, the Crisis Psychological Hotline Corpus (CPHC) dataset was developed for suicidal crisis recognition. Statistical analysis reveals that non-crisis callers have significantly longer speech durations compared to operators, while there is no significant difference in speech duration between mild and severe crisis situations. The results show that the model achieves high accuracy and consistency in recognizing callers at different crisis levels, with an F1 score and accuracy of about 0.92 on the test set. Future research could expand the dataset, optimize the automated screening and identification system, and explore the detection of other mental health conditions to enhance the model’s effectiveness in psychological assistance and suicide prevention.