A New Approach for Authorship Attribution

Authorship attribution is a text classification technique, which is used to find the author of an unknown document by analyzing the documents of multiple authors. The accuracy of author identification mainly depends on the writing styles of the authors. Feature selection for differentiating the writing styles of the authors is one of the most important steps in the authorship attribution. Different researchers proposed a set of features like character, word, syntactic, semantic, structural, and readability features to predict the author of a unknown document. Few researchers used term weight measures in authorship attribution. Term weight measures have proven to be an effective way to improve the accuracy of text classification. The existing approaches in authorship attribution used the bag-of-words approach to represent the document vectors. In this work, a new approach is proposed, wherein the document weight is used to represent the document vector instead of using features or terms in the document. The experimentation is carried out on reviews corpus with various classifiers, and the results achieved for author attribution are prominent than most of the existing approaches.