期刊名称：IAENG International Journal of Computer Science
印刷版ISSN：1819-656X
电子版ISSN：1819-9224
出版年度：2019
卷号：46
期号：2
页码：377-383
出版社：IAENG - International Association of Engineers
摘要：We propose a method of estimating the importanceof figures and tables in scientific papers by propagatingimportance beyond media; from language to image. In scientificpapers, language and image information coordinately enable thereader to easily understand the complicated contents in detail.The proposed method propagates this importance from thesentence to figures/tables in which the position of the sentencereferring to a figure/table and surrounding sentences are usedto evaluate the importance of figures/tables. We conductedan experiment on estimating the importance of figures/tablesby assuming figures/tables used in a presentation poster areimportant. The experimental results indicated that the proposedmethod exhibited the highest mean of the average precision(MAP) compared to comparative methods focusing on thesize and caption of the figures/tables. Moreover, the proposedmethod exhibited more MAPs than the comparative methodwhich did not update the sentence importance using similarityamong sentences. We believe that the proposed method iseffective in supporting the creation of scientific presentationposters from papers.
关键词：importance propagation; application of natural;language processing; creative support; poster design
