A Novel Ensemble Model to Summarize Kannada Texts

Abstract

Automatic text summarization is the task of producing a smaller piece of text containing important sentences and all relevant important information from the original document. With the existence of abundant digital data, this technique helps in gaining quick access to the required data in native languages. In this work, we present a technique for an efficient extractive summarization of Kannada documents and articles. In the proposed novel ensemble model, each of the sentences in the input text are assigned a ‘Weighted Terms value’. This is computed by leveraging the concepts of term frequency—inverse document frequency (TF–IDF) algorithm, Galavotti Sebastiani Simi (GSS) coefficients and positional ranking of sentences. An additional mathematical function is also devised to compute weights for sentences as a whole, based on their positions in the text. The final summary is curated by coherently picking the sentences whose ‘weighted terms value’ exceeds the threshold which is set based on the size of the required summary.