论文部分内容阅读
从跟踪词能量演化线索的角度分析潜在爆发词探测的可行性,提出一种基于词的能量和能量增长趋势的潜在爆发词探测方法。首先对词的生命周期及其演化现象进行阐述,在方法分析和词的能量积累与衰减、能量趋势变化分析的基础上,提出建模依据,设计EneTr模型,并分别针对EneTr模型中的关键问题提出相应的解决方案,实现具体的算法,最后分别针对网络新闻和科学文献两种类型的文档流进行分析和实验,验证本方法的效果。
From the perspective of energy evolution of trace words, this paper analyzes the feasibility of potential explosives detection and proposes a probabilistic approach to detect potential explosions based on word energy and energy increment. First of all, it expounds the life cycle of the word and its evolution phenomenon. Based on the analysis of the method and the energy accumulation and attenuation of the word, and the analysis of the change of the energy trend, the modeling basis is proposed, the EneTr model is designed, and the key problems in the EneTr model Propose corresponding solutions and implement specific algorithms. Finally, we analyze and test the two types of document streams, namely, news on the Internet and scientific documents, respectively, and verify the effectiveness of this method.