基于中文文本相似度评估的情感勒索话语检测系统  

Emotional blackmail utterance detection system based onChinese text similarity evaluation

作　　者：林文晟 杨观赐[1] 钟世昊 Lin Wensheng;Yang Guanci;Zhong Shihao(Key Laboratory of Advanced Manufacturing Technology of the Ministry of Education,Guizhou University,Guiyang 550025,China)

机构地区：[1]贵州大学现代制造技术教育部重点实验室,贵阳550025

出　　处：《计算机应用研究》2024年第10期3073-3080,共8页Application Research of Computers

基　　金：国家自然科学基金资助项目(62373116,62163007);贵州省科技计划资助项目(黔科合支撑[2023]一般118,黔科合平台人才[2020]6007-2)。

摘　　要：情感勒索是通过情感压力迫使周围人听从自己要求的交流方式,容易导致对方的负面情绪甚至心理问题从而影响交流效果。为了检测日常交流场景下的情感勒索话语进而改善交流效果,开发了基于中文文本相似度评估的情感勒索话语检测系统。首先,基于Susan Forward的情感勒索理论标注收集到的数据,构建情感勒索语料库和测试集;其次,分析情感勒索的表达方式,分别基于词性和语义词设计文本相似度评估方法,进而形成基于中文文本相似度评估的情感勒索话语检测算法;然后,在构建的数据集上开展实验,该算法获得的平均recall与F 1-score分别为95.21%和79.95%,优于对比算法;最后,基于该算法开发情感勒索话语检测原型系统,系统在不同测试条件下的平均recall为87.24%,展现出良好的鲁棒性和检测性能。Emotional blackmail is a way of communication that forces people around them to listen to their requirements by emotional pressure.It can easily lead to negative emotions and even psychological problems of the other side,then affect the communication effect.In order to detect emotional blackmail utterance in daily communication scenarios and improve the communication effect,this paper developed an emotional blackmail utterance detection system based on Chinese text similarity evaluation.Firstly,this paper labelled the collected data based on Susan Forward’s emotional blackmail theory,and constructed an emotional blackmail corpus and a test set.Secondly,this paper analyzed the expression modes of emotional blackmail,and designed text similarity evaluation methods based on part of speech and semantic words respectively,then formed an emotional blackmail utterance detection algorithm based on Chinese text similarity evaluation.Thirdly,this paper carried out experiments on the constructed datasets.The average recall and F 1-score of the proposed algorithm are 95.21%and 79.95%respectively,which are better than the compared algorithms.Finally,this paper integrated an emotional blackmail utterance detection prototype system based on the proposed algorithm.Under different testing conditions,system test results show that the average recall is 87.24%,which shows good robustness and detection performance.

关 键 词：智能系统 情感勒索话语检测 文本表达方式 相似度评估 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]