Text Summarization using Feature Similarity based K Nearest Neighbor

This article proposes the modified KNN (K Nearest Neighbor)algorithm which considers the feature similarity and is applied tothe text summarization. The words which are given as features forencoding words into numerical vectors have their own meanings andsemantic relations with others, and the text summarization is ableto be viewed into a binary classification where each paragraph isclassified into summary or non-summary. In the proposed system, atext which is given as the input is partitioned into a list ofparagraphs, each paragraph is classified by the proposed KNNversion, and the paragraphs which are classified into summary areextracted ad the output. The proposed KNN version is empiricallyvalidated as the better approach in deciding whether each paragraphis essential or not in news articles and opinions. The significanceof this research is to improve the classification performance byutilizing the feature similarities.