Summarization of Odia Text Document Using Cosine Similarity and Clustering

Abstract

Automatic text summarization a subfield of Natural Language Processing (NLP) aims at producing precise and non redundant text aided by machine learning techniques. Using varied methods, especially machine learning techniques has enhanced its performance from different perspectives. The proposed work efficiently utilizes hierarchical clustering by using cosine similarity measure for segregating sentences. The model adopts an extractive method for summarizing Odia text document. It focuses on minimizing redundancy and achieves it through cosine similarity matrix. Though the methods employed are primitive for European languages like English, Odia language which is computationally in a passive state and has complex morphological structure, is a novel work. The results obtained can modestly be considered satisfactory.