A Study on an Automatic Categorization of the Report of Judgement on Marine Accident Inquiry Agency : A Categorization of Documents of Cause of Collision Accidents by Term Weighting

Abstract

This paper presents an automatic categorization of documents of cause of collision accidents from the Report of Judgement on Marine Accident Inquiry Agency by term weighting. Term weighting is performed by term frequency and kinds of terms. WIDF (Weighted Inverse Document Frequency) is adopted as a term frequency method. Kinds of terms which are registered in a dictionary for parsing the documents are discussed from technical and general prospects. The results obtained are as follows. (1) 5 kinds of terms which parse the documents are defined. They are terms defined in law, techinical terms, semi-technical terms, complex terms and nouns. (2) Terms defined in law, techinical terms and semi-technical terms plays an important role of categorization. (3) The accuracy of a right answer is about 60% at the maximum.