Event Evolution Model Based on Random Walk Model with Hot Topic Extraction

Abstract

To identify the evolution relationship between events, this paper presents a new model which utilizes random walk model to weight the cosine similarity of events according to the chronological order. This model is proved to be effective and accurate in identifying relationship of events through comparisons with other models. This paper also puts forward an innovative method to detect hot topics which applies the concept of related events. This paper introduces parallel computing, including Spark and MapReduce, to significantly improve the efficiency of the event evolution calculation and hot topic detection.