Comparison of Text Classification Methods for Government Documents.

Classification of documents is a task which is rapidly gaining popularity in machine learning application. I can be used for such tasks as detection of spam, detection of phishing attempts or for detection of fake messages. One of the areas where document classification is of growing importance is automatic assessment of civic, legal and government documents. Using automated classification, a governmental institution might scan all its incoming correspondence, and automatically assign it to the proper departments. In this paper results of various classification approaches to classification of official documents are presented. Presented results are based upon classification of two sets of document: first, where given class of documents is derived from their common origin; and second, where class is determined by the common destination of the documents.