- Home
- About
- News
- Resources
-
Committees & Working Groups
- Full List & Descriptions
- WORKING GROUPS
- COMMITTEES & OTHER PROJECTS
- NCC Events
- Spotlights
- Contact
The NCC is collaborating with institutions and scholars to release a monthly series on our blog entitled Japanese Studies Spotlight. These features showcase exciting online collections available to researchers and students in Japanese Studies, introducing the archive or project, describing their contents, and demonstrating how they can be usefully engaged in research or in the classroom. If you are interested in submitting something to the series, please contact Paula R. Curtis, NCC’s Digital Media Manager, at digitalmediamanager@nccjapan.org.
Dr. Asanobu Kitamoto, Director, ROIS-DS Center for Open Data in the Humanities (CODH) & National Institute of Informatics (NII)
The Center for Open Data in the Humanities (CODH) was established in April 2017 under the Research Organization of Information Systems (ROIS), part of Japan's Joint Support-Center for Data Science Research. It is aimed at advancing research on "data-driven humanities" and "humanities big data" in the era of open science. Data-driven humanities is about analyzing humanities resources using state-of-the-art technology from computer science and statistics, while humanities big data is about creating humanities datasets with that domain knowledge to be used in non-humanities research, such as natural science. Today I would like to introduce one example of this partnership between humanistic resources and data-driven technologies, our open-source software suite, the IIIF Curation Platform.
The IIIF Curation Platform is an open-source software suite, not only for viewing images but also for collecting, annotating, and processing images. IIIF (International Image Interoperability Framework) is an international community effort to standardize access to images and ensure interoperability. Since the API specifications are open, anyone can freely develop software that conforms to the API, and the results can be released as open source. Our IIIF Curation platform has been under development since 2016 and aims to extend IIIF’s functionality to enhance user-driven research and applications.
IIIF has been adopted by many libraries and museums in the world as a standard platform for high-resolution image delivery. Using IIIF, institutions can learn best practices for building institutional digital archives, including the usage of open-source software as a common tool to access many digital archives. Although two well-known IIIF image viewers, “Mirador” and “Universal Viewer,” are already in use at many universities and archives, CODH is working to push the boundaries of this technology with researchers and students in mind.
IIIF was originally designed by people in museums and libraries to deliver their collection images and authorized annotations to users. This is perfectly fine for reading books or inspecting materials. But another CODH is interested in facilitating what we call “user-generated content,” such as creating a personal image collection and publishing it on the web with one’s own annotations and commentary. Surprisingly, this is not easy to do in the current IIIF specifications, so CODH is seeking to fill this gap by expanding the concept of“curation” to account for other types of user engagement.
Our IIIF Curation Platform consists of several different tools and integrations, in particular a viewer and curation board that integrate our other AI-driven projects for the study of Japan. Below I will briefly introduce some of the CODH projects that demonstrate how these unique functionalities can be used for digital humanities research. You can also see the IIIF Curation Platform tutorial in Japanese and English to browse a variety of use cases in institutions and digital humanities research.
KaoKore
The KaoKore Project illustrates well the implementation of our IIIF Curation Viewer (ICViewer). This viewer not only offers standard functionalities of IIIF image viewers such as image zooming and page navigation but also allows for unique functionalities such as curation, and, as we shall see in the next sections, export and annotation. The goal of KaoKore is to create a theme-based collection of facial expressions across visual Japanese sources like picture books and handscrolls housed at disparate institutions. As you can see below, by navigating different collections with images that are IIIF compatible and compiling curated selections from them using IIIF’s interoperability functionality, it becomes possible to create a personalized collection of research and teaching materials.
Above, one can see a curated collection of warrior faces that are now a part of CODH’s KaoKore dataset. Through the ICViewer it is possible to select a particular area of interest from each image using what we call the “box” function, creating a kind of clipped preview of a small area of an image. Using ICViewer’s “favorite” button, one can then add that specified region into the list of favorite images, creating a specialized collection. By repeating this process, users can collect parts of images from any IIIF manifests (links on an archive’s webpage that enable IIIF compatibility) and create a theme-based personal collection. Then a user can publish the collection in the “curation JSON” format, which can be later viewed again through the ICViewer.
Because links to the original image are maintained within the ICViewer, users can go back to the original image to study it in detail or the broader context surrounding the selected regions. This is helpful for digital humanities research that combines “close reading” of a source and “distant reading” of a much larger collection of similar visual objects.
The latest tool of CODH’s IIIF Platform is the IIIF Curation Board, which was released in September of 2020. This tool enables a kind of virtual KJ-technique for idea generation, brainstorming, and project design. Using this tool for classifying face images in KaoKore dataset, this tool works effectively for art history research that analyzes styles based on the characteristics of how faces are drawn. In addition, the curation saved during the work can be used as a record of the research process, facilitating a form of open science in digital humanities research by increasing the reproducibility of research. A case study for art history research is also helpful for using IIIF Curation Board for similar purposes.
Edo Maps Beta
Our Edo Maps Beta project, which works with cartographic resources from the Edo period (1603-1868), also utilizes our IIIF Platform tools. We draw on images of old Edo maps (Edo Kiriezu) housed at the National Diet Library (NDL), which include numerous place names. The CODH implemented their ICViewer, which offers an “annotation view mode” to add a user’s annotation on the original images, to create digital counterparts to the historical labels on the maps as map markers. This is especially useful for cross-institutional digital archives, where the original image comes from the image provider and the annotation on top of it comes from the user’s service. Users can then take advantage of web-based maps using overlays of user-generated annotations.
KuroNet
Finally, the export function of our ICViewer has been integrated into an AI-driven optical character recognition (OCR) tool in our KuroNet Kuzushiji Recognition Service. There are countless archival holdings in Japan and overseas written in kuzushiji, a form of handwritten calligraphy widely used in premodern Japan. Using our IIIF Platform tools, we have combined IIIF with image processing services, allowing us to use AI to scan calligraphy images and transcribe characters therein.
As you can see below, using the ICViewer, a cropped image can be exported to an external service running kuzushiji OCR. Then the ICViewer can overlay the results on the original image using character markers in the annotation view mode. We offer two types of character markers; the standard mode allows users to control how the markers are displayed, such as the location (horizontal and vertical shifts) and the size or transparency of markers, while the overwrite mode sets the background of character markers as similar color with the image background so that a user can focus on the transcribed text. These are features that are unique to our ICViewer and will be useful for the integration of OCR services into IIIF compatible archives and sources.
The Center for Open Data in the Humanities is pioneering new uses of IIIF technologies. Through the ongoing development of our IIIF Curation Platform and other tool integrations, we are working to create a comprehensive infrastructure that supports a research and teaching workflow where users can freely collect, process, and publish images. As of May 2021, we have seven products running on web browsers, one application designed for a specific purpose, three Python programs running on the server-side, and one Docker configuration to support installation. These software products are designed to work together through APIs and are used in combination with one another to help users achieve their project goals.
In order to support widespread access to these tools and techniques, the IIIF Curation Platform is distributed under an open-source license. Some demo functions are also available for anyone to use, so please visit the website and make use of them. You can also follow our activities on our official Twitter account, where we share information on datasets, projects, and open-source software.
NCC's Comprehensive Digitization and Discoverability Program task force recently launched an informational and instructional video series that includes a video on KuroNet Kuzushiji Recognition by Tarin Clanuwat. We encourage you to explore these tools and our guides to learn more!
0 Comments.