Report
  1. HOME
  2. REPORT
  3. Natural Language Processing for a Geospatial Exploration of ...
  4. Overview

 [書込]

Overview
April 19, 2021(Mon)

The proposed project aims at digital geospatial exploration of the sites depicted in Japanese early modern ukiyo-e landscape prints, and focuses on a large dataset of prints available through the database of the ARC. The information on the locations depicted in the prints will be extracted by applying the Natural Language Processing technology (e.g. Name Entity Recognition) on their meta-data. This will be followed by visual analysis of the images performed with the use of visual encoders (pre-trained to classify/caption generic images) to identify possible connections between visual characteristics of the images and certain locations. These analyses will generate a new information on geo-temporal distribution of the sites and their visual depiction contributing to a better understanding of Japanese landscape prints in general.