• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文会议>International Conference on Orange Technologies >HAPPINESS/SUFFERING factors recognition based on point-wise mutual information
【24h】

HAPPINESS/SUFFERING factors recognition based on point-wise mutual information

机译：基于逐点互信息的幸福/受难因素识别

获取原文
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

This work presents a novel approach for happiness improvement based on two psychological factors of redefined individual alphabets on ¿¿¿HAPPINESS¿¿¿ and ¿¿¿SUFFERING¿¿¿ (See Section II-A), of which the factor recognition based on Point-wise Mutual Information (PMI) method is treated as baseline. To further enhance PMI performance, a modified baseline namely Keyword PMI (K-PMI) approach is proposed by using POS tags as features for the keywords selection, and calculates the association between the factor category and the keyword in the training lexicon. To reduce the scale gap of PMI value, normalizing PMI value is essential and adjusted by the significant score of the keyword for the factors category classification. The experimental results have shown that the proposed K-PMI approach can achieve an average recognition accuracy of 96.72% in the inside test. Whereas the outside test, the average precision rate is able to achieve 73.5% accuracy, which is significantly higher than the baseline accuracy of 43.46%, such results further to prove the proposed K-PMI approach outperforms the baseline in the experiment, and demonstrates the efficiency and the feasibility of the proposed approach.
机译：这项工作提出了一种基于两个心理因素的幸福改善方法，该心理因素是在¿¿¿HAPPINESS¿¿¿和¿¿SUFFERING¿¿上重新定义了单个字母（请参阅第II-A节），其中基于Point的因素识别互惠信息（PMI）方法被视为基线。为了进一步提高PMI性能，提出了一种改进的基线，即关键字PMI（K-PMI）方法，该方法使用POS标签作为关键字选择的特征，并计算出训练词典中因子类别与关键字之间的关联。为了缩小PMI值的规模差距，归一化PMI值是必不可少的，并通过针对因子类别分类的关键字显着性分数进行调整。实验结果表明，所提出的K-PMI方法在内部测试中可以达到96.72％的平均识别精度。外部测试的平均准确率能够达到73.5％的准确度，明显高于基线准确度43.46％，这些结果进一步证明了所提出的K-PMI方法在实验中优于基线，并证明了效率和建议方法的可行性。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号