Automated Text Summarization for Indonesian Article Using Vector Space Model

In a scientific work, an abstract always contains main information of an article including at least a researched problem, aim(s), methodology, and result of the study. Writing an abstract requires a conscientious analysis since the contents would affect both the readers' interestedness and disinterestedness on a particular or overall research topic. However, people generally write manually by summarizing the article. The aim of this study is constructing automation for summarizing Indonesian articles as an alternative approach to an abstract. This is involving two methods to summarize an article. A Term Frequency-inverse Document Frequency is used to get a keyword and weight terms, and a Vector Space Model is utilized to represent abstract text into a vector that used to identify the linkage of documents. From this method, the result of the summary can be generated from documents. Supporting this research, we used several journal articles written by a manual abstract. The results of this application show that the automatic summarization produces a paragraph which consists of more than three same sentences constantly as compared to manual paragraphing.

Перевод пока недоступен