中圖分類號: TP391.1 文獻標識碼: A DOI:10.16157/j.issn.0258-7998.200148 中文引用格式: 富雅玲,楊文忠,吾守爾·斯拉木,等. 基于重點突發(fā)詞的突發(fā)事件檢測方法[J].電子技術(shù)應(yīng)用,2020,46(11):82-86. 英文引用格式: Fu Yaling,Yang Wenzhong,Woxur Silamu,et al. Method of bursty events detection based on key bursty-words[J]. Application of Electronic Technique,2020,46(11):82-86.
Method of bursty events detection based on key bursty-words
Fu Yaling1,Yang Wenzhong1,2,Woxur Silamu1,Yang Mengmeng1,Liang Fan1
1.College of Information Science and Engineering,Xinjiang University,Urumqi 830046,China; 2.National Engineering Laboratory of Social Security Risk Perception and Prevention and Control of Big Data Application, Chinese Academy of Electronic Sciences,Urumqi 830000,China
Abstract: Because of the suddenness, crowd-gathering and destructiveness of bursty events, this paper proposes an bursty event detection method combining user influence and bursty-words for the bursty events published in weibo to avoid a series of social problems caused by bursty events. In order to extract a large number of key burst-words, we need to first calculate the bursty value of words, using two indicators: word influence and word state, taking words larger than a certain threshold as burst words; adopting cohesive hierarchical clustering method, hot topics are clustered by the co-occurrence matrix of burst word sets. After that, the results were put into the trained classifier to classify hot topics, and finally the bursty events and their types were obtained. The real microblog data were used to conduct bursty events on them. The experimental results before and after the use of the classifier were compared. This method can effectively filter common hot topics and improve the accuracy of emergency detection.
Key words : bursty event;burst word;clustering;classification;event detection