Automatic Indonesian Text Summarization Using Vector Space Model

Abstract

Summary provides information from a text briefly to reduce the amount of effort in understanding the text. However, with the number of the existing text data, creation of a summary manually can take a long time. This process can be easily done with automatic text summarization based on Indonesian language. Automatic text summarization can be done by determining the value of similarity between sentences. This similarity value can be determined based on the vector angle sentence on the vector space model. Some sentences with the highest similarity value chosen as a representation of the text summaries. Based on testing that used data from 25 text consisted of argument, description, exposition, narration, and persuasion text that resulted in an average value of precision, recall and F-Score each 0.55, 0.49, 0.51 from expert interviewees and 0.55, 0.48, and 0.50 from common user interviewees. The summary had the same informativeness level with the expert interviewee’s summaries. Precision with the highest result obtained from argument text with an average precision of 0.52 from expert interviewees and 0.46 from common user interviewees.