Natural Language Processing (NLP) took advantage of the apparition of powerful computers together with the development of new tools and methods, involving linguistics, mathematics and computer science to make important progress recently. Even though many challenges remain to be taken up, like automatic translation or virtual operators communicating with humans in natural language, it appears that the advances in this field are of huge economical interest. As a result, large companies are investing a lot of money and trying to create research networks in this field, in order to obtain results as quickly as possible. Institutions like the DARPA in the United States, the MITI in Japan or the European Commission are also funding many large-scale programs to foster research.
This report gives an overview of the research carried out in Japan in NLP, and limits itself to written language processing. It stresses the research topics of interest in 1998 and presents the laboratories involved. It was achieved after the study trip made by a student of the Corps National des Mines (France) who had the opportunity to meet many Japanese researchers and to visit their laboratories.
This work was made possible by a grant from the French Ministry of Economy, Finance and Industry, and the extreme kindness and open-mindedness of all the Japanese laboratories that were solicited.The author thanks them all.
It begins with a presentation of the tools developed to analyse written language, including morphological and syntactic analysis. It then focuses on the semantical representation of language and the main applications currently under development. A section is also devoted to the linguistic resources available, followed by a quick presentation of the laboratories who contributed to this report by inviting its author to visit them.
The goal of this report is to give an overview of the research. Therefore the author chose to write a report for a non-specialist, often at the expense of the scientific precision that a deeper study of any research topic would need. The interested reader is therefore invited to visit the web sites of the laboratories indicated in the last part, which often contain a bibliography, or to contact the author of this report at the following e-mail:
jean-philippe.vert@mines.org