Abstract
Automatic text classification was considered as a series of information processing, and a new classification technique, namely the "Accumulation Method," was proposed in my previous paper. This proposed method has the property of unlimited use of feature terms even though it is a simple technique. The use of "character N-gram" and "word N-gram" as feature terms are proposed using this property of the classification technique. Next, the technique is evaluated through some experiments. In these experiments, the newspaper articles of Japanese "CD-Mainichi 2002" using the Naive Bayes method (baseline method) and this proposed method were classified. As this result, the classification accuracy of the proposed method is shown to improve greatly compared to the baseline method. That is, it is 88.7%. Thus, the proposed method has very high performance. The proposed method has a new viewpoint, so it can be expected to be developed further in the future.
