融合句法特征和句法相似度的网络舆情突发事件识别方法研究

Research on Network Public Opinion Emergency Recognition Method Based on Syntactic Features and Syntactic Similarity

[目的/意义]快速、准确地从突发网络舆情文本中识别事件。[方法/过程]提出一种融合句法特征和句法相似度的网络舆情突发事件识别方法。结合句法特征提出面向事件的句法特征提取方法,利用事件语义标注和句法特征提取方法构造事件句法特征库,通过计算待测文本与句法库的句法相似度来识别网络舆情突发事件。[结果/结论]以新型冠状病毒肺炎疫情为例,所提出网络舆情突发事件识别方法在该舆情下的最优相似度为0.93,在此相似度下从一段新的文本中识别出160个事件和30个非事件,F1值达到了0.848。通过方法测评证明网络舆情突发事件识别方法在利用句法相似度识别事件和进行相同相邻词性合并等方面创新的有效性。

[Purpose/significance]This study aims to identify events from the text of sudden network public opinion quickly and accurately.[Method/process]This paper proposed a method to identify network public opinion emergencies by integrating syntactic features and syntactic similarity.An event oriented syntactic feature extraction method was proposed based on syntactic features.Event syntactic feature database was constructed by using event semantic annotation and syntactic feature extraction methods.The network public opinion emergencies were identified by calculating the syntactic similarity between the text to be tested and the syntax database.[Result/conclusion]Taking the novel coronavirus pneumonia epidemic as an example,the optimal similarity of the network public opinion emergency identification method proposed by the author is 0.93 in this public opinion.160 events and 30 non events are identified from a new text under this similarity,and the F1 value reaches 0.848.Through the method evaluation,it is proved that the proposed method is effective in using syntactic similarity to identify events and merge the same adjacent parts of speech.