Abstract :
The goal of the Style Change Detection task is to detect the stylistic changes in a document and exploit them to determine the number of authors. This study reviewed nineteen (19) state of the art papers and articles on writing style change detection. The papers were identified and selected based on study area, year of publication and the technique proposed for writing style change detection. The focus of this study was to investigate the features used, the techniques and the results obtained by these state of the art studies. Three categories were defined and all papers placed in one of the groups based on the problem it was solving. The study found out that the most commonly used feature category was the lexical features although using feature combinations yields better results. In addition, simple distance measures were shown to outperform other state-of-the art techniques in authorship clustering and style change detection. The use of ensembles of algorithms is recommended for style change detection tasks when the text length is short and the dataset is large.
Keyword :
Authorship, Clustering algorithms, Multiple authorship, Stylometry, Style change detection