A Graph Based Ranking Strategy for Automated Text Summarization


Text summarization is a process of capturing the idea and line of thought from an original text and inculcating the same into a short coherent text. Automated text summarization aims to meet this objective of retaining all the key ideas instilled in the text while skipping upon the redundant and repetitive bits of information. The reduced text thus compiled must be coherent in itself in order to meet the semantic and syntactic organization of the language. This work presents an extraction based automatic text summarization algorithm. The methodology proposed involves constructing of a directed weighted graph out of the original text wherein each sentences is taken to be a node. The weights for each of the edges are determined by using a suitable distortion measure which analyses the semantic relation between the two adjacent nodes/sentences. A ranking algorithm is used to compute the most important sentences in the text and that should be present in the summary based on the weighted graph. This technique has been employed on multiple data sets and has performed well on the evaluation parameters laid down for such applications.

DU Journal of undergraduate research and Innovation 1(1), University of Delhi
Nitin Agrawal
DPhil Student in Computer Science