摘要
针对现有的弹幕文本敏感词识别方法查准率不高的问题,提出了一种基于情感计算与深度学习的弹幕文本敏感词识别方法。该方法可对数据进行预处理,对弹幕文本进行分词;可根据文本主题采用情感计算法提取敏感关键词,组成敏感词汇集;可基于深度学习理论,采用情感计算方法构建弹幕文本敏感词识别模型;可采用DHT方法训练模型的敏感度,对弹幕文本进行敏感等级定级,以此完成敏感词识别。实验结果表明,该方法查准率为92%,查全率为85%,既提高了查准率,也减少了敏感词识别时间,从而达到保护数据安全的目的。
Aiming at the problem that the precision rate of the existing recognition method of sensitive words in bullet screen text is not high,a recognition method based on emotional computing and deep lear-ning is proposed.First,the data is preprocessed and the bullet screen text is classified.Then,according to the text theme,the sensitive keywords are extracted by means of emotional computing to form a collection of sensitive words.Based on the deep learning theory,a model to recognize the sensitive words in bullet screen text is constructed by using the emotional computing method.Finally,the sensitivity of the model is trained by DHT method,and the sensitivity of the bullet text is graded to complete the recognition of sensitive words.The experimental results show that the precision rate of this method is 92%,and the recall rate is 85%,which not only improves the precision rate,but also reduces the recognition time of sensitive words,thus achieving the purpose of protecting data security.
作者
叶海燕
YE Haiyan(College of Information Engineering, Chaohu University, Hefei 238024)
出处
《常州工学院学报》
2022年第3期29-33,共5页
Journal of Changzhou Institute of Technology
基金
安徽省质量工程教学研究一般项目(2020jyxm1253)。
关键词
情感计算
弹幕
敏感词
识别
emotional computing
bullet text
sensitive words
recognition
