I have been trying to deploy named entity extraction and constructing linked data from digital-humanities-related data for a couple of years. This repository shares you how to do them.
Ogi Nikki is business records of Ogi Domain, called Ogi Han, at Edo era (1600 - 1868) in Saga, Japan. Ogi-han was one of feudal domains based on a tax system in Edo era and settled as an affiliation of Saga Domain. The number of titles that represents the records is assumed about 74,000. The records contain when, where, who and what happened in Ogi-han. The original records are hand-writing, called Kuzushi-ji. The Center for Regional Culture and History, Saga University, has been transcripting into Japanese texts. The grammar of the texts is called "Sourou bun", quite different from the recent Japanese one. Sourou bun was mainly used for business records and official papers in Edo era. You can browse the results of the transcription at a digital archive which is also constructed by the center.
By 2019, more than 30,000 titles are converted into texts and stored in the archive. I have tried to convert the data of the archive into Linked Data.
I suppose linked data is a future-oriented data. The important merit is that linked data can include information of structured relationships about the data's attributes and annotations, not only the values. This means even if your archives have gone for some reason, you can restore your archives much better and easier again because you don't miss "information" of fields of data. In other words, when you lose "meanings" of the fields, you lose how to "use" the data.
Linked Data can also describe relationships on an instance level. You can connect some values in data to other information on the Web. This means you can get new semantic information more than original content.
Technical costs for converting conventional data into Linked Data is the biggest issue. Lack of technical information and pragmatic use cases of modeling and converting data in Digital Humanities is also an issue.
OgiNikki Projects are supported by JSPS KAKENHI Grant Number 19K20630, 2019-2022.