An integrated approach to search and representation in early Japanese writings

The present project addressed the problem of representing early Japanese writings in such a way as to permit efficient search for and use of them in a computational setting。Together with some external researchers and consultants，we worked on how to encode archaic Japanese scripts，which are often found annotated with multiple layers of information in the form of·rubi，·tochu，·，kyakuchu，·etc.A particular approach we took was based on the idea of Annotation Graph(AG)from the computational linguistics literature.We found that the AG，designed as a unified framework for representing multiply aligned texts，could be adapted to work with Japanese scripts if we assume that annotations are technically subtexts that run in parallel to the base text。We put the idea to test through experiments with actual texts from Satomi-hakkendenden，Makurano-soshi，Kogetu-sho，Wakanroeishu，with some favorable results.

The present project addressed the problem of representing early Japanese writings in such a way as to permit efficient search for and use of them in a computational setting. Together with some external researchers and consultants, we worked on how to encode archaic Japanese scripts, which are often found annotated with multiple layers of information in the form of・rubi,・・tochu,・,・kyakuchu,・etc. A particular approach we took was based on the idea of Annotation Graph (AG) from the computational linguistics literature. We found that the AG, designed as a unified framework for representing multiply aligned texts, could be adapted to work with Japanese scripts if we assume that annotations are technically subtexts that run in parallel to the base text. We put the idea to test through experiments with actual texts from Satomi-hakkendenden, Makurano-soshi, Kogetu-sho, Wakanroeishu, with some favorable results.