Visualizing author attribution using Blobby objects

Abstract

In this paper, we propose a set of distinct features that endeavors to recognize the author of the text documents, independent of the theme of the document. The features are used to distinguish the unique style of every author. The features proposed are independent of the theme, so the features are suitable for any type of text document. Thus, it will be useful for systems that train themselves to particular users in order to identify the author automatically. Blobby objects are modeled based on the author attribution information and the objects depict authors features, author feature weight ages and the author identification for each and every document. The visual display of author attribution is useful in depicting the differences or similarities in the style of author, based on the features as a measure. The blobby objects are implicitly modelled and they move and join together showing the author of the document with the authors feature set values. Gestalt laws were applied to the perceived visual output to visualize author attribution.