Bengali text summarization using TextRank, Fuzzy C-Means and aggregate scoring methods

Abstract

In this world, it is very difficult and time consuming for humans to summarize large documents, reports, news and research articles. Multiple text summarization techniques play vital roles in picking the important points and sentences thus reducing the time and effort required to read a whole article. Numerous summarization techniques have been applied to the English language but works on Bangla text summarization is still limited. Furthermore, in our country, Bangladesh, all summarization is mainly done by humans. Keeping that in mind we aim to find a simple way of summarizing Bengali texts with the technology at hand. Text summarization can be of two types, abstractive and extractive. In this paper, we will use extractive text summarization to summarize Bengali passages, using Fuzzy C-Means, TextRank and Aggregate Sentence Scoring methodologies. We have also done a comparative study, among the three methodologies to find out that Fuzzy C-Means out performs the other two methods to generate a more concise and accurate summary.