Performance analysis of keyword extraction algorithms assessing extractive text summarization

Automatic text summarization is the task of deriving a meaningful and concise brief from a given text while retaining the concept and key information conveyed by the original text. So far, numerous approaches and algorithms have been devised to achieve this goal with certain accuracy and effectiveness. One key aspect of text summarization is accurate identification of keywords from the given textual content. In this paper, the relative performance of three popular algorithms, namely TextRank, LexRank and Latent Semantic Analysis for keyword extraction were investigated by measuring their effectiveness in identifying keywords from set of articles. The performance of each of these algorithms were contrasted with those of handwritten summaries of the same articles. The most effective algorithm was identified from the empirical results.