A study on method of noise reduction for visualizing disaster area using noun morpheme n-gram from geo-tagged tweets

Abstract

In this paper, we did not generate the noise filter in advance, and we focused on the noise reduction method which automatically generates and uses the filter using only the noise reduction targeted tweets. Specifically, in each targeted tweet of noise reduction, a morpheme n-gram analysis of only nouns was performed, those with a high combination frequency of nouns were generated as noise filters. The results suggest that there is a certain degree of effects of noise reduction method under the condition of morpheme 4-gram of noun and frequency threshold 3 or 5 in the case where the number of tweets to be processed is at least about 110 or more.