Abstract
We have developed a server that analyses the morphology of agricultural documents written in Japanese. An agricultural term dictionary is combined into the server, and the server is able to identify agricultural terms within documents that may include many agricultural terms. Client applications communicate with the server using Java RMI. Software developers can incorporate the server as a remote object within their natural language processing programs, and their programs can communicate with the server through the Internet. To evaluate whether using this server improves the quality of natural language processing, we have automatically categorized the text of numerous agricultural documents. The results of these tests showed that using the server improved classification accuracy.
