Ant Colony Optimisation for Stylometry: The Federalist Papers

This paper describes the use of Ant Colony Optimisation for the classification of works of disputed authorship, in this case the Federalist Papers. Classification accuracy was 79,1%, which compares reasonably well with previous work on the same data set using neural networks and genetic algorithms. Although statistical approaches have performed much better than this, the advantage of a rule-based approach is the ability to produce readily intelligible criteria for the classification decisions made.

@InProceedings{Oak04,
 	 author =  {Oakes, Michael Philip},
 	 title = {Ant Colony Optimisation for Stylometry: The Federalist Papers},
 	 booktitle = {International Conference on Recent Advances in Soft Computing},
 	 journal = {None},
 	 editors = {}
 	 number = {}
 	 volume = {}
 	 pages = {86--91},
 	 year = {2004},
 	 month = {}
 	 publisher = {Nottingham Trent University},
 	 doi = {}
 }