An approach to automatic text summarization using WordNet

Abstract

Text Summarization is the procedure by which the significant portions of a text are retrieved. Most of the approaches perform the summarization based on some hand tagged rules, such as format of the writing of a sentence, position of a sentence in the text, frequency of few particular words in a sentence etc. But according to different input sources, these pre-defined constraints greatly affect the result. The proposed approach performs the summarization task by unsupervised learning methodology. The importance of a sentence in an input text is evaluated by the help of Simplified Lesk algorithm. As an online semantic dictionary WordNet is used. First, this approach evaluates the weights of all the sentences of a text separately using the Simplified Lesk algorithm and arranges them in decreasing order according to their weights. Next, according to the given percentage of summarization, a particular number of sentences are selected from that ordered list. The proposed approach gives best results upto 50% summarization of the original text and gives satisfactory result even upto 25% summarization of the original text.