Implementation of an Automatic, a Posteriori, Hierarchical Classification System
This paper describes a total system which provides the capability to semi-automatically index and classify any given file of information. The semi-automatic indexing method assigns key terms to each "document" in the file . These key terms may be modified, (i.e., added, deleted, or changed) by the user. The indexed documents are then assigned to categories by an automatic classification algorithm. The classification assignments are a posteriori. Classification dictionaries are also produced which can be used as an aid in browsing through the data base and in retrievals from the data base. Samples are given of the results obtained while indexing and classifying an experimental data base containing the texts of 1669 radio messages.