How to publish results of master theses

Question

This post is from the perspective of the management of a master’s degree in computer science (not from the perspective of a single student).

We have a 2 years master’s degree (= 4 terms). In 3rd term students use about 15 to 20% of their time to find and plan their master thesis topic. In the final 4th term the students then use 100% of their time on carrying out the research and writing up their thesis (70-80 pages). The master thesis defence is at the end of the 4th term. The hand-in deadline for the thesis is about a month before that.

Once the thesis is handed in the students are asked to write a 3-5 pages (scientific)paper about their thesis and handed it in 1 week before their master thesis defence (so they have 3 weeks time to write it and potentially get feedback from their supervisor). Usually about 25% of these papers are (i) interesting enough in terms of findings and (ii) of high enough quality to be worth a shot for a journal/conference publication.

The problem is however that most of the time a few iterations between student and supervisor on the paper would be necessary to make the paper ready for a submission to a journal and even if this is achieved the journal review process takes months at which point the students have already graduated. At this time most students have already found well-paid jobs and show little interest in getting the paper published e.g. doing the revisions the reviewers ask for.

What’s the best way to get the (often very interesting) results of the students still published? Just uploading the papers as pre-prints on e.g. the arxiv yields relatively low visibility. Writing the master paper before the actual master thesis write-up usually yields lower quality papers. Another student continuing the paper submission is also not really an option as each student needs an independent topic and the topic cannot be something like “I got the results of Smith at al. published”. The supervisor is usually not deep enough in the code etc to do additional analysis during the revision(s).

What is the best way to get the results of master theses published? How do other universities approach this problem?

Answer 1

You are seeing the obvious problem, but you are not writing it in the correct terms. Let me rephrase your problem.

How can I retain students, so they can perform the necessary additional work to prepare and revise a paper?

There are many workaround, but the only obvious solution is this one:

to get a job done, pay a professional to complete the tasks needed

After the student completed their studies they are professionals, whatever work done in terms of research/literature study/paper writing is work (and research work, and personal growth, and science progress, like the work of any PhD/PostDoc/Professor/Research Assistant/etcetc, of course) and work must be paid.

Find the way to fund 6 months of additional work for the student that is doing a good job and for which you can reasonably see a scientific publication.

Answer 2

I'd guess that it is pretty hard to get publishable results in most fields, including CS, with only one term's work. It is hard to even get started in that much time, much less produce results and write them up properly. You point out other obstacles as well.

But one option, though it takes work, is to have the department start a local journal with an editorial board consisting of professors and possibly others (faculty from other places, graduates, etc.).

Students are given the option (or are required) to publish in that journal unless they submit elsewhere. Depending on the reputation of the institution this could actually be a valuable resource generally if you can find a way to make it visible.

This is easier if the faculty is large. It is also conceivable that such a journal might be multi-disciplinary, increasing the options for both papers and editors.

When I was in grad school we had such a publication. A couple of faculty (out of about 50 in math) made it happen and kept it interesting. It was for more than MS theses, however, and some interesting but short mathematical papers were published there. It was good practice for us, the students. Some faculty also had papers there.

Note that Law Review publications by some major law schools are actually student run and a position on the editorial board is considered a career booster.

Answer 3

What is the best way to get the results of master theses published? How do other universities approach this problem?

The three universities I worked and studied in all published a certain fraction of master's theses successfully, and it always worked in the same, boring principle. A paper, by and large, does not get published based on thesis work unless there is somebody available that is willing to continue pushing for it after the student graduates. This can be PhD students that collaborated with the student or a supervisor (faculty), but it needs to be somebody with a sufficiently long clock on their contract to see the paper through to the end. Expecting students to commit to any non-trivial work after graduation is a fool's errand - there is no way to force them, and very little incentives for them to voluntarily deprioritise their new paying job over writing a paper.

As you observe, this does indeed require the responsible person to be more involved with the work than simply grading a finished report - some familiarity with the data, analyses, and maybe even code will often be required to do the kind of revisions that are often necessary (not only because of journal revision requests - it happens quite frequently that student work is decent upon submission but is missing one or two more advanced analyses to be actually publishable). Herein actually lies a key advantage of the model to the student - supervisors that are looking to potentially publish a paper often are much more involved because they have to be, and consequently in a position to provide much more detailed feedback than a supervisor that is mainly evaluating a report as it develops.

My personal model, which has led to around 10 published papers based on thesis work so far, works like this:

• When the student is about 50% done I evaluate if I see the potential for a publishable paper in this work.
• If yes, I ask the student if that's something they would be interested in, at the same time explaining that I have no hard expectation on them to do anything on top of what they have to do anyway to graduate.
• If the student agrees, I start to observe the project (much) more closely. I ask to be added to all repositories, check out if I can understand and reproduce what the student is doing. Essentially, I try to make sure that I can do at least simple extensions in case they are needed for the paper.
• After the project is over, I start drafting the paper. As much as I can I try to re-use what the student has already produced (text, figures) etc., but more often than not significant re-writing is necessary, some or all of the figures need to be re-drawn, and sometimes additional analysis is required. I add the student as first author of the paper, and keep them in the loop what I'm doing. The only thing I require from them is that they give a final ok to the paper I produced based on their work before I hand it in. Sometimes they volunteer to do more themselves (like conduct additional analysis or revising some text), but quite frankly it's rare.

This means that these papers aren't "free" - there is still non-trivial work from my side that goes into each and every one of them. But even with this model the amount of work from my side per paper is much less than doing an entire paper from scratch. On the other hand, it does happen (more often than I'd like) that publishable work never gets submitted, either because the students do not agree (rare, but not unheard of) or because neither me nor one of my students finds the time to go through the steps outlined above.

Answer 4

Been there, done that. Being on both sides of the process I have seen basically three approaches:

• A first and obvious way is indeed to have the master's thesis available on some university hosted website or open repository. The master's thesis would be referenced and cited much in the same way as a PhD thesis. If the work has merit and is well referenced on search engines, why wouldn't it be cited?

• If you want to have the student publish themselves in a reviewed conference or journal, one approach is to help the student find a position in academia after the master's thesis. Typically they would work towards a PhD thesis. If the PhD is related to the master's and you are involved, publishing the paper is logical and the student would be expected to do that. However, the student may also work on another research topic, in another lab. This is basically not different to the situation where a reasearcher switches lab (for instance for a postdoc) and continues collaboration with their previous lab. In this situation the student only works for glory. Publication is not guaranteed and is a delicate balance between will, interests, curtesies between labs, and some actual involvement of the advisor in the writing process (see next point).

3. The best option is to publish the work yourself. If only one student's work is published, put their name as the first author except if you contribute significantly with your own additional work. What works best in my opinion, is to publish one paper after a series of master's theses done on the same subject. In this case, having the advisor as the first author and actual writer of the paper also makes sense since they promoted the theses and have an overview of the work done; the paper would also cover a larger scope, which makes it more interesting. There are some curtesy rules to follow:

• You must list the students as co-authors, after getting their approval to do so;
• You must give the students an opportunity to contribute to the act of writing the paper;
• You must respect the fact that the students may not be able to contribute to the writing and by default expect to do the additional coding, debugging, simulations, analyses, yourself. The students should still get their names on the paper.

In fact, I have the feeling the whole discussion boils down to this:

The supervisor is usually not deep enough in the code etc to do additional analysis during the revision(s).

The work published by a lab should be owned by the permanent members of the lab, otherwise the ideas are vaporware. You, as a researcher and advisor, should ensure reuse of the code and ideas --even if it means re-coding a bad student's code yourself to make it useable-- and know enough about the work to put it into persepective. In CS, a good way to ensure reuse and citation is to integrate the student's work in a library published by the lab, and well curated by the permanent members of the lab. This also promotes involvement of the permanent members, yourself included, in the actual code.

You cannot expect to get papers published with your name on it, just by sitting in the back and without getting your hands dirty.