Notes On Getting Into Grad School Naacl 2021

Notes on Webminar: Getting into grad school (NAACL 2021)

What do you think sets apart a strong application for graduate study?

Vlad Nicule) Differs so much

Motivation / Recommendation letters matter so much

Good Letter? Relevant papers that shows your field of NLP

What are the criteria to keep in mind while choosing a grad school, a research group and a PhD advisor?

Elizabeth Nielsen) 2 things to look for:

1. How they are to work with? -> more important, hard to figure out
2. How well do they do research?

Is your supervisor overfed with work? what kind of dynamics are they included? competitive? Find single-author papers / How well do they explain? It’s a good indicator of great advisor(?)

How do you structure statement of purpose / research proposal?

Nick McKenna) Statement of purpose is very important. To convince reader that you have an idea of project that is useful / can do that. Motivation of problem - State of field - How you would approach your problem. Not have to be exact, but more thoughtthrough it is, more convincing it would be. It’s just like who you are, because each of us have a different story. And definately tell your story!

What were the most useful resources in getting into grad school?

Salomon Kabongo Kabenamualu) Twitter is valuable! (GOOD~~)

How did you face imposter syndrome, during grad school and your application process?

Sian Gooding)

1. External ) You compell yourself to people around you
2. Internal ) You doubt your skills <- Effected more highly

For 1 - Anti-savi <- Do it reversely - If they didn’t get this intern, didn’t get this paper accepted…

You’re seeing the best path of the people you look on twitter. You keep that on mind (being inspired v.s. being intimidated) <- Parallelized on how am I going to get to this

For 2 -Try not to be a perfectionist <- become one of it because you’re so self-critical. Be kind to yourself.

Nick) I face that myself a lot. Other than papers & application process. There are other things you can do that you can show that you are good at research. (e.g. participating at a reading group. Able to show up every week, do discussion, talk about papers - good way to get a feel for the field. And got good recommendation letters by that.)

Even if it doesn’t make it to paper, it is still a good progress.

What aspects of the labor environment should I keep an eye out for?

Vlad Niculae) Group dynamics, healthy environment with collegues…

Also, environment is important (very different from country to country and school to school)

Having a contract <- expectations are clear. Cannot do what is not asked

Are you going to be an employee in the University? or not? This is very important.

Pension? Visa aspects? Technical error in a corner? Immigrant? Salary’s differnt. Is there a grad student union? If you face injustice, where can you go to? <- So different. & Important

Can you change advisors? Is there a history of this?

Mental health resources at university?

How do you conduct or manage online research collaborations, especially under current circumstances?

Sebastian Ruder ) Just be kind to your collaborators.

Regarding broad research collaboration - most collabo was remotely done. I think communications is more important. Writing one more email / one more chat is important (to sync with each other)

How does one keep up with the latest research and papers?

Nick McKenna) 3 things you can do

1. Going out and find research papers yourself
2. Mining by google scholar - targeted searches. Getting researches sent to you (news feed by arxiv, semantic scholar on keywords)
3. Form a peer group, share some interests. Meet together and discuss a paper.

When and how did you get into NLP research?

Salomon Kabongo Kabenamualu) Opportunity - wasn’t able to search for a mathmatician job - got ML job - … Why didn’t I see this option before?!?!

How to approach potential research mentors?

Elizabeth Nielsen) Read their website first and read the instructions.

Reach them first (e.g. conference and so on) and then email.

Make your email easier to read. Good sharp students that can help students they can enjoy advising. Don’t ask super open-ended questions….

Stay away from open-ended questions / vague questions / long emails…. <- This doesn’t make people think that you can make their life easier..

How was your first publication process looked / experience like?

Sian Gooding) Had no idea what I was doing. Once done once, it becomes so much easier next time. Checking it hundred times because I was so nervous. It is very normal to get papers rejected. That is NOT the end of life. Peer review is amazing and you can learn a lot from it. It’s okay for it to be rejected. It can make the paper a lot better.

What pre-doctoral research experiences you think could better prepare a student for graduate study?

Reading a bunch of research. Learn to judge papers. Learn to look at paper and see what you can do after this research. How can I actually come up with research questions that I can test? (It’s not like I want to study NLG.)

….. My hand hurts…

What career experiences do you think are most fruitful for building a professional identity in NLP?

internal identity / research taste / What are your personal preferences? (beautiful theory / Useful practice / … / writing beautiful papers?) <- how you work yourself. Useful in finding your collaborators - who can do complementary things

Figure out what your overall research area is. Communicate more on surveys/blog posts.

How do you grapple with the ethical implications of developing NLP, including biases in algorithms and data?

Careful not just on relying on GLEU score.