As in Europe, there is a strong interest in multilingual text generation and machine translation. Another central aspect of the rapid development of information and communication technologies is the ever growing demand for intelligent user interfaces in order to provide universal access to information. Hence, integrating natural language generation and hypertext as well as the generalization of text planning systems for presentation planning in multimedia human-computer interaction are also considered very important issues. Another topic where language generation comes into play are applications in automated text summarization.
Commercial applications of natural language generation technology rely on robust template-based approaches to produce standardized multiparagraph texts such as business letters or monthly reports.
The situation in Japan is somewhat different: Natural language generation does not constitute a well established research field and only a few template-based approaches have been reported. Activities related to the automatic production of natural language are focussing mainly on machine translation and the synthesis of spoken language.
(New Oct. 96) The Computer and Information Science and Engineering Directorate of the US National Science Foundation (NSF ) recently announced a new interagency initiative called STIMULATE (Speech, Text, Image, and Multimedia Advanced Technology Effort), which aims to fund basic research in human multimodal communication, including speech, text, and images, NSF [1996].
(New Sep. 97) The major Computational Linguistics conference ACL'97/EACL'97 demonstrated that not only in Europe but also in the USA concept-to-speech approaches on the one hand and automatic text summarization on the other hand constitute central topics for current research in natural language generation. Especially for the field of intelligent text summarization it is assumed that a high market potential for sophisticated solutions exists. In accordance with these promising expectations there are now significant industry-based research and development efforts in the USA involving also some of the leading software companies.
(New May 98) The UNL system for multi-lingual communication aims to enable automatic translation of online texts to provide information on the Internet in all national languages of the members of the United Nations. The UNL project, which constitutes a large scale international cooperation projected for a duration of 10 years, is led by the United Nations University in Tokyo, Japan. The Institute of Advanced Studies at United Nations University has developed the Universal Networking Language (UNL ), which serves as an intermediary representation language for the translation of text documents from one native language into another. With respect to the field of natural language generation this Universal Networking Language is of great importance since it defines a very powerful standard input format for language specific text generation components. It can be expected that the UNL will facilitate easier reuse of software components from generation systems and hence accelerate commercial exploitation of language generation technology.
(New Sep. 98) The joint International Conference on Computational Linguistics and Annual Meeting of the ACL (COLING/ACL-98) once again demonstrates the important role of the USA in this research field given the fact that the majority of the (active) participants are working in North America. With less than 15 % conference particpants from Japan the Japanese contingent is relatively small compared with Europe. The various contributions from Japan, however, indicate that the major trends in current research are quite similar to the USA. A common global trend is the high interest in natural language generation and the growing importance of multimodal systems in support of communication forms that extend natural language interaction with additional modalities. Both, Japanese and US researchers presented a broad spectrum of current research activies at the post-conference workshop dedicated to Content Visualization and Intermedia Representations . Theoretical and algorithmic aspects concerning dialog structures are being considered intensively as well. Information retrieval, extraction and classification as an area with a high commercial potential seems to be another specific focus in the USA.
(New Oct. 98)
The
15th National Conference on Artificial Intelligence
in the USA, AAAI-98, attracted also a broad spectrum of
contributions from the general field of intelligent (multimedia) user
interfaces.
Argumentation is currently a much heeded issue in natural language
generation.
Various approaches in this research area investigate computational
methods for supporting argumentative discourse in human-machine
communication.
(New Jan. 99)
The
9th International Workshop on Natural Language Generation
,
INLG-98,
indicates that the field is no longer dominated by reseachers from
North America although there is still a very active research community
both in academia and industry.
A special focus of current research seems to lie on planning and
generation with multiple media as well as on architectural questions.
Only one contribution originated from Japan. This work is related to
dialog planning.