基于多种特征融合的微博突发事件检测方法

利用微博数据检测突发事件具有重要意义.针对以往检测方法特征不够丰富、准确率不高等问题,提出了一种基于多种特征融合的微博突发事件检测方法.该方法根据情感符号构建情感特征模型,对微博数据进行情感分类,并采用Kleinberg算法对情感特征进行突发期检测;在突发期内根据词频特征、词频增长特征和话题标签特征融合加权提取突发词,...

Many-objective flexible job shop scheduling optimization based on INSGA-II

Using micro-blog data to detect bursty events is of great significance. Existing main methods still have the problems of insufficient features and low accuracy, so we propose a micro-blog bursty event detection method based on multi-feature fusion. The method constructs an emotional feature model based on emoticons, classifies micro-blog data, and uses Kleinberg algorithm to detect the bursty periods of emotional features. In the bursty period, the bursty words are extracted according to the fusion weighting of word frequency characteristics, word frequency growth characteristics and topic label characteristics, and the similarity distance between emergent words is calculated based on the co-word analysis method to construct the word co-occurrence matrix. The condensed hierarchical clustering algorithm is used to cluster the bursty words to obtain the bursty events detection results. The experiment shows that the accuracy rate is 0.836 1, the recall rate is 0.879 3, and it proves that the proposed method can detect the bursty events in micro-blog effectively.