Text summarization

Abstract

Text summarization has always been a necessity for the common people to read, write and present the content in a brief and understandable way. This requires the person to manually go through the available content and use their own approach to effectively reduce the content to a short summary of readable size. This work is to reduce the manual burden in making the content brief and automate the process effectively. This work will find its use case among a wide range of people ranging from students who need to summarize their work for a presentation to content writers publishing their content on a website. This model gets the text document as an input and provides a summarized text document in the text format. All of these processes happen in an offline mode and can be considered as an added advantage. As a whole, the desired output of this work is to provide relevant simplified information for fast intake and it summarizes the most important points of the document. The interface to this work only accepts text documents in English as an input and the summarization algorithm used is cosine similarity. The proposed algorithm for ranking of the summarized sentences is text rank algorithm. Cosine similarity is a similarity measure that compares two non-zero vectors and also the cosine of angle between the two vectors aforementioned. The text rank algorithm is a graph based algorithm which is used for ranking sentences accordingly with the priority being the most important point on top.