摘要
新闻事件检测是自然语言处理任务中的一项任务。新闻事件检测旨在从新闻文本数据流中检测出新闻事件并给出事件主题。人工构建新闻事件的特征费时费力。传统的新闻事件检测方法是根据新闻事件之间的空间距离检测新闻事件,对于不同的新闻事件相似度较高时,容易误判为同一事件。针对上述问题,论文提出基于注意力机制的双向长短记忆网络构建新闻事件检测模型,通过深度学习学习新闻文本深层次的特征并且基于新闻事件检测模型构建新闻事件建模应用系统。实验表明论文方法在准确率、召回率优于传统方法,可对新闻事件准确识别。
News event detection is a task of NLP.News event detection aims to detect news events from the news text data stream and give event topics.It takes time and effort to manually build the characteristics of news events.The traditional method of detecting news events is to detect news events according to the spatial distance between news events.When different news events are similar to each other,it is easy to misjudge the same event.Aiming at the above problems,this paper proposes a attention-BiLSTM to construct a news event detection model,the deep features of news text are studied through deep learning and a news event model⁃ing application system is built based on news event detection model.Experiments show that the accuracy and recall rate of this meth⁃od is better than the traditional method,which can accurately identify news events.
作者
张秀华
云红艳
贺英
胡欢
ZHANG Xiuhua;YUN Hongyan;HE Ying;HU Huan(College of Computer Science&Technology,Qingdao University,Qingdao 266071;School of Electronic Information,Qingdao University,Qingdao 266071)
出处
《计算机与数字工程》
2021年第6期1143-1147,1280,共6页
Computer & Digital Engineering
基金
国家重点研发计划“‘云计算和大数据’专项-跨时空多源异构数据的融合、开放共享技术与平台”(编号:2016YFB1001103)资助。
