Microblog Hot Topic Detection Based on Box-plot
Guo Deqin · Journal of Shanxi University · 2014
A new kind of method was proposed to detect the hot topic which used the principle of box-plot to recognize burst words.It counted the noise characteristics of Microblog,and filtered the Microblog according to these characteristics.Moreover,the principle of box-plot was used to recognize burst words.Moreover,it summated the weight of burst words to detect the microblog hot topics.The experiment result shows that compared with SSM,EC and ELM,the proposed scheme increases by an average of 13.8%in F1value,and it is insensitive to the noise.