Working with a supervisor who is not expert in my research field

Question

I am a PhD student, and I am working with a supervisor who is not expert in my research field. I am wondering what I should do to remedy this.

When I just started out my PhD program, as I was applying for a major government scholarship, I approached my supervisor to give me some guidance on my general research path that I am expected to take through my PhD program, because I needed to write my research proposal to apply for the scholarship. My supervisor gave me a list of papers to read, and the papers were heavily oriented towards deep neural natural language processing models. So I assumed that my supervisor was heavily into the application of deep learning in natural language processing, and so I picked my research direction to be deep learning / natural language processing. I identified my specific research topic, ran the experiments, and my supervisor and I tried to submit my paper to a computational linguistics conference, which was rejected.

However, as I worked with my supervisor, I realized that my supervisor's field of expertise is application of machine learning in open-ended questions in survey, and that he is not at all an expert in the deep neural language models. He in general lacks the knowledge in deep neural language modeling, and when I worked with him on my first publication, he was not able to provide any guidance/feedback on the design of my experiment or my general approach to the problem. Now I am scared that I picked a research direction that is too deviated from my supervisor's research, and I am keep wondering whether I have to tell my supervisor that I'd rather do the type of research that he is doing. I haven't talked to my supervisor about this, but I am getting an impression that he would most likely to suggest just continue working on the revision of my rejected paper.

Now I am a bit frustrated that my supervisor made me to read deep natural language processing papers when I applied for the scholarship. I think it would have been more reasonable for him to recommend me to read something that is related to his research, which is the statistical analysis of open ended survey questions. When I told him the type of research problem that I would like to work on, he encouraged me to pursue it, but I think it would have been much better if he stopped me at that stage.

Can I still be a successful PhD student if I pursue deep natural language processing on my own without getting much help from my supervisor? I feel very insecure because I suddenly feel like I am taking up on this all by myself without getting any guidance from anyone. Usually students who publish their paper at top computational linguistics conferences/journals are a part of natural lnaguage processing lab lead by a renowned professor in that field. I don't think I can compete against these students who are a part of active natural language processing lab.

Should I try finding external collaborator? or should I just tell my supervisor that I'd rather do the research that is closer from his domain? I feel like I am doomed and I am depressed. If I am to find an external collaborator, I am not sure how to do this since I do not have any personal contact in this field.

Any advice should be highly appreciated,

Thank you,

Answer 1

Can I still be a successful PhD student if I pursue deep natural language processing on my own without getting much help from my supervisor?

The core task of a PhD supervisor is not necessarily to be an expert on the content of your research. Sure, it will help if there is some overlap, but even if you pursue a research direction that is well aligned with your PhD supervisor, it is very likely that you will surpass their knowledge within the first year of your PhD. Key is, that your supervisor can provide you with a tremendous amount of help on how to do research productively. If you get stuck, which strategies are available for overcoming the obstacle? How does one balance reading, programming, writing (everyone prefers some of these tasks over others, but you'll need to overcome your instincts to spend the bulk of your time on your preferred tasks because the other tasks also require time investment) such that publications result? Once the experimental results are in, how does one convert the programming and experimental results into a publishable manuscript? How many citations are enough?

Your questions is best answered when taking two distinct approaches to two halves of the question. "Can I still be a successful PhD student if I pursue deep natural language processing on my own?" Yes. "[...] without getting much help from my supervisor?" It is probably most productive to reframe your thinking about this: you will get much help from your supervisor, just not in the way you imagined.

---

Should I try finding external collaborator?

This depends a bit on your personal style, and the preferred working style of your supervisor. I am always more than happy to have my PhD students interact with more experts in the field; it's good for the network. In fact, your supervisor may have people in their own professional network that may be helpful in this matter. Consider discussing this with your supervisor: the first paper that the two of you have worked on has been rejected, and you think that the work might benefit from bringing in outside experience. Would your supervisor be open to opening up the collaboration to more NLP researchers?

---

What should I do?

This is impossible to answer, since it involves you caring for your personal wellbeing in the specific circumstances in which you find yourself. You write that you feel doomed and depressed; you should definitely talk to people, whether those people are mental health professionals, or 'just' friends and family. Asking the question here is a healthy thing to do, but you may also consider finding out whether your university has support groups for PhD students; social events, mailinglists, other ways to interact informally with your peers. Sharing experiences can be very illuminating.

Answer 2

It seems you have unrealistic expectations of what a PhD is supposed to involve. Who cares if you don't publish in a top journal? It seems your goal is not to do actual research.

You are the one who is supposed to pick a research topic. Your advisor simply advises you how to attack it.

You are doing original research, which, by definition means nobody is an expert on it! If the advisor was an expert on the topic, there would be no research since the topic would have already been solved.