Japan Search: Much ado about nothing

Japan Search is a searchable portal that aggregates content from digital archives from, at present, 125 different databases managed by different 70 Japanese institutions and organisations, and seeks to display this data in a “user-friendly format.” On paper the idea is very good, and therefore when it launched in late August 2020 there was much fanfare. However, although it has the potential to be very useful, I have not found the finished product particularly useful for my own research. I’d thought about writing a piece on Japan Search shortly after its launch, but disappointed I decided to wait to see how the site was improved upon as it continued to be developed.

Before I explain why I generally refrain from using Japan Search it is important to note all the things that the site does well and why other researchers may and likely do find it useful. The site is easy to use employing a simple keyword search function with options to narrow down results according to genre, database, licensing etc. A search will often allow the user to easily identify where there are substantial holdings on a certain genre or type of text, art, cultural property etc. This might be quite useful for those starting out in their academic careers, beginning to research a new theme or and looking to discover new holdings. Furthermore, unlike many of the databases that I am accustomed to using, Japan Search allows users to search through a number of non-textual categories such as animations, movies, games, and manga, alongside more traditional media such as books, maps, and documents. This is a fantastic feature, which for those of us who primarily deal with texts can facilitate the discovery of new types of data relevant to our research.

The portal does a very good job of displaying images of manuscripts, texts, and paintings when they are avaliable. I particularly like the “Fullscreen” option which provides a streamlined, almost naked, view of the image files which one can zoom in and out of easily. See for example the images below comparing a ukiyoe displayed in Japan Search and the same ukiyoe displayed in the source database. One should notice immediately, that there is a lot less clunk around the page on Japan Search. The licensing for each image is also clearly displayed. I imagine that Japan Search could become a very useful part of the research arsenal of someone working primarily with visual media.

Image on Japan Search on the left, and the source database on the right.

Whilst I have found Japan Search useful for tracking down images that I can use within my publications, it has some major flaws that make it very nearly unusable within my own research. The search function, the core part of Japan Search, simply doesn’t work very well. Japan Search suggests that the user searches for very general terms such as “Fuji San” or “Udon,” but general searches provide very general results. If I search for the term Kirishitan キリシタン, for example, I get a fantastic 4,685 hits or 235 pages of results. These are ordered by default according to “Score,” but could also be ordered according to title, contributor, or date. How “Score” orders results is unclear – on the first page I am presented with an image of an artefact, a selection of post-WWII photographs, a manuscript version Dochirina Kirishitan ドチリナ・キリシタン, and bibliographical information for several secondary sources. If one searches for the same term but in Romanised form, one is provided with a horrific 2,921,824 hits or 146,092 pages most of which are completely unrelated to the desired search. Of 20 results on the first page, only 6 are actually related to the key term. If one uses more specific search terms such as Kirishitanban キリシタン版, for example, one is provided with more specific and potentially more useful results. However, even if one conducts more specific searches with Romanised terms thousands of irrelevant results are provided. This is a problem because sometimes we may want to search for sources written in languages other than Japanese which use Romanised terms in their titles. At present, however, this may not be possible.

Searching the term Kirishitan on Japan Search.

Another problem is that although Japan Search contains a large number of image files for visual media, it does not contain many for textual sources. For example, amongst the 123 hits for Kirishitanban, only one manuscript can be viewed on the platform (others such as the aforementioned Dochirina Kirishitan exist on the site, but do not appear in the search results for Kirishitanban). In some cases, this problem stems from the poor search functionality. In others it is linked to the fact that images are not provided (due to copyright and other reasons) by the digital archives that contribute their data to Japan Search. Most entries for texts or manuscripts, therefore, contain only bibliographical information. This could be useful, but problems with the search function and the display of results mean that other databases devoted to the display of bibliographical information do a much better job and are likely more useful to those of us working with texts. Why go to Japan Search for bibliographical information when you could go to CiNii? The same could be said for those wanting to access images of texts. More often than not it is simply easier to go directly to the source databases that most of us are already using than to try to find a manuscript copy on Japan Search. That said, since Japan Search includes very clear information about image copyright and image usage, I do find myself increasingly going to the site in order to find copyright free images to use in my work.

Japan Search is a nice idea, but poorly executed. If you are using visual media in your research, it is worth taking a look especially since you can easily find freely usable images, but if you primarily work with texts it leaves a lot to be desired. If you are a textual scholar, unless you’re looking to get a general feel of where you might be able to find certain texts, I recommend going to the databases that you are already familiar with and using.