Abstract
We have focused on local assembly minutes as a large-scale language resource, and have developed 'Giimiru,' a cross-searching and visualization system for local assembly minutes, based on the corpus construction. This paper outlines the feature visualization using XAI technology newly implemented in 'Giimiru,' and describes the prospects for contributing to quantitative research on the Japanese language through observation and discussion of the visualization results. While many place names and councilor names were highlighted in the visualization results, not a few expressions that characterize other regions were also found, suggesting the possibility of discovering features that cannot be captured by human visual confirmation alone.
https://ror.org/04nt8b154 
