Choosing a lab

Every doctoral thesis in Japan is required to be published online. Almost nobody applying has read one.

When you choose a graduate programme in Japan you are really choosing a laboratory, and usually a single supervisor. The information you are given to decide with is a faculty page and a list of publication titles. There are two much better documents. One exists for every doctorate awarded in this country because an ordinance requires it. The other is not required by anything, and most laboratories publish it anyway — twenty years of their students' thesis titles. Both are free, and you can read them this afternoon.

A Japanese university website will tell you that a laboratory works on, say, energy materials. It will not tell you whether anyone in it finished a doctorate in the last four years, what they actually built, how long it took them, who examined it, or whether the work was written in Japanese or English. All of that is a matter of public record. Very few applicants know it is there, because the thing that makes it public is a Japanese ministerial ordinance that has never been of any interest to anyone writing in English.

What the rule says

The Degree Regulations — a Ministry of Education ordinance dating from 1953 — carry two separate publication duties, and they fall on two different parties.

Article 9 has an escape hatch, and it is worth understanding rather than ignoring. Where there is an unavoidable reason — commercially sensitive work, a patent in progress, a book contract — the author may, with the university's approval, publish a summary in place of the full text. In that case the ordinance requires the university to make the full text available for inspection on request.

So the floor is high. For any doctorate awarded in Japan you are entitled to find, at minimum, an abstract, a summary of what the examiners concluded, and either the whole thesis or a summary plus a route to the whole thesis.

Where to read them

The national aggregator is CiNii Research, run by the National Institute of Informatics. It has a search restricted to dissertations, it is free, it needs no account, and it works in English.

A single record returns the thesis title, the author, the university that awarded the degree, the degree name and field, the date, the subject keywords, and a direct link into that university's own repository, where the file itself sits. Here is a real record, taken from the database on 5 August 2026, chosen because it is entirely ordinary:

Search in both languages — the answers are different

The database indexes Japanese and English text, and the same field returns different sets depending on which you type. We ran a handful on 5 August 2026:

Note the third row against the second. Searching in English and searching in Japanese for the same concept returns overlapping but different sets — 924 against 839. If you only search in one language you are seeing part of the field. Paste the Japanese term in even if you cannot read the results; the titles come back with the repository links attached, and the abstracts of Japanese theses are frequently written in English.

You can also restrict by year, which is how you find out whether a field is alive here. Theses matching 機械学習, by year of award: 5 in 2010, 9 in 2013, 11 in 2016, 51 in 2020, 134 in 2024.

⚠️ What the ordinance does not cover — and what exists anyway

The publication duty applies to doctorates only. Nothing in it touches master's theses or undergraduate dissertations, and you will not find the full text of those in any national database. If someone tells you every thesis written in Japan is online, they are wrong, and the boundary is exactly here.

But the titles are a different matter, and for judging a laboratory they are arguably the more useful half.

A large share of Japanese laboratories and departments publish a running list of their master's and undergraduate thesis titles on their own websites — not because a rule requires it, but because it is the normal way a Japanese lab shows what it has been doing. The lists are often very long-running. One laboratory at a national university publishes titles going back to 2005, twenty years of them, each year separated into master's and undergraduate. Another has an index of every year since 2002.

The catch is that there is no central search for these. The doctoral database is national; the title lists are scattered across individual laboratory websites, in Japanese, and no aggregator collects them. You find them by going to the laboratory's own site and looking for a menu item such as 業績 (output), 論文 (papers) or 卒業論文・修士論文題目 (undergraduate and master's thesis titles). Searching the laboratory or department name in Japanese together with 修士論文 題目 will usually get you there directly.

Two further limits on the doctoral side. Where an author used Article 9's summary provision, what is online is the summary rather than the whole thing — the full text comes from the university on request, which is slower and involves a human. And the requirement dates from a 2013 amendment, so older theses tend to sit in libraries and the National Diet Library rather than on the open web.

What you can actually learn from this

Reading one thesis is interesting. Reading the last five years of a laboratory's output is a different exercise, and it answers questions that no prospectus will.

• Does the lab finish people? Count the doctorates, then count the master's titles — a group with no doctorates in four years may still be graduating five master's students a year, which is a different diagnosis entirely.
• What is it really working on? Faculty pages describe a research area. A thesis describes a problem that someone actually solved, at the level of detail where you can tell whether you want to spend three years on it.
• What language does it work in? If recent theses from a group are written in English, an international student has been through it. If none are, you will be the experiment.
• What does it consider good enough? The examiners' review summary is published under Article 8. It is short and it is candid in a way that almost nothing else in academic publishing is.
• Who examined it. The review names the committee, which tells you who else in the department is adjacent to the work — useful if your supervisor retires or moves, which happens.

And it costs nothing. The reason to mention this at all is that the natural sciences hold 1.7% of the international students in Japanese higher education, and one plausible part of that is simply that it is hard to tell, from outside Japan and in English, what any specific Japanese laboratory does. This is the document that answers it, and it has been sitting there since 2013.

What to take from this

• Every doctorate awarded in Japan has a published abstract and a published examiners' summary, by ordinance, within three months of the award.
• The full text is due within a year, published by the author with the university's cooperation. Where a summary was published instead, the university must supply the full text on request.
• Search CiNii Research, in both English and Japanese, and restrict by year to see whether a field is growing here.
• Read the last five years of your target lab before you write to a prospective supervisor. It is the single highest-value hour available to you, and almost no applicant spends it.
• Then find the lab's own list of master's and undergraduate thesis titles. The ordinance does not cover those, but most labs publish the titles anyway, often twenty years of them. That list tells you how many people the group actually graduates each year, which the doctoral record alone will not.
• Search for the title lists in Japanese — laboratory or department name plus 修士論文 題目. There is no central database for them; they live on individual lab websites.

Sources. Degree Regulations (学位規則), Ministry of Education Ordinance No. 9 of 1953, Articles 4, 8 and 9, retrieved from the e-Gov statutory database on 5 August 2026. Article 8 places the duty to publish the abstract and the examiners' review summary on the university, within three months. Article 9 places the duty to publish the full text on the person awarded the degree, within one year, with provision for a summary in its place where the university approves, and a corresponding duty on the university to make the full text available for inspection on request. On "course doctorates" and "thesis doctorates". Article 4 sets out two routes: award to a person who has completed a doctoral programme, and award to a person who passes the thesis examination and is confirmed to have equivalent academic ability without having completed the programme. Repository records often carry a certificate number distinguishing the two; the labels used for those numbers are institutional convention rather than something the ordinance specifies, so we have not translated them here. Search counts. All figures in the table are result counts from CiNii Research's dissertations endpoint, run by us on 5 August 2026. These are our searches, not a published statistic. They reflect what that database has indexed on that day, which is not the same as every doctorate awarded in Japan, and they will drift. Treat them as an indication of scale, not as a count. The example record is reproduced from the same database on the same date and is quoted for illustration; we have no connection to the author, the laboratory or the university. On the thesis title lists. The description of what those lists contain is based on laboratory pages we read on 5 August 2026, including one publishing titles by year since 2005 (Iwate University) and one indexing every year since 2002 (Ritsumeikan University). Practice varies by laboratory and there is no requirement to publish these at all, so we cannot tell you what share of Japanese laboratories do — only that it is common enough to be worth checking for any group you are considering, and that the per-year counts quoted above are from those specific pages rather than a national figure. What this article is not. It is a description of a publication rule and how to use it. It is not advice about admissions, and finding a thesis you like is not the same as being accepted by the group that produced it.