Comparison of Graph-based and Term Weighting Method for Automatic Summarization of Online News

Abstract

Text summarization is one of the quickest ways to get the gist of a paragraph or story. In text summarization, there are two ways that can be used: extractive approach and abstractive approach. In this research, the summarization was conducted using extractive approach. The extraction process was conducted by taking a few sentences from a document and combining them into a short summary. The most common method used in conducting text summarization is graph-based method. The authors proposed another method for summarization, namely term weighting method. The purpose of this study is to compare between the result of graph-based method and term weighting method in order to determine the best method for text summarization. The text pre-processing phase involves omitting the stopwords and the affixes. Moreover, the researcher utilized the measurement of Precision, Recall and F-Score. Based on the experiment using the proposed method (term weighting method), the result shows that the average values on Precision and F-score for term-weighting method are 0.296 and 0.280 respectively, which are better than the values of graph-based method. In the end, the result shows that the proposed method, which is the term weighting method, produced better summary compared to graph-based method.