Top Takeaways from an ACL 2020 Mentoring Session on Career planning + Becoming a research leader

This year’s top NLP conference, ACL 2020, is the hub for many fruitful discussions. Among all the events, I really, really appreciate the mentoring sessions, conducted by more experienced members to help junior NLP researchers.

This post aims to spread the key takeaways to a broader audience. We hope people from all backgrounds can get first-hand notes on one of this year’s ACL mentoring sessions — Long-term career planning + Becoming a research leader: Building your professional identity, held on 8 July, 2020.

Word Cloud made from all the discussions in the mentoring session. Credits to this word cloud maker.

Credits: Many thanks to Professors Mohit Bansal, Hanna Hajishirzi, Heng Ji, Rada Mihalcea (alphabetically ordered by family names), and all the active participants in the session!

Theme: In this session, the mentors (NLP professors) shared their answers to many of the most frequently asked questions from mentees (junior NLP researchers). In this post, we compiled the notes regarding the most upvoted four Q&As by the mentees.

So here we go!

Question 1 (Q1): Key takeaways to succeed in NLP research

Here are the top suggestions:

1.Be passion-driven!

People will approach research differently if they are passionate. Think of something you did that makes you really excited — you will talk to many friends about it and think over it all the time!

2. Care about *quality* more than quantity

Think of how you want people in the field to perceive you — you want to establish yourself as the top people to think of in a specific area.

3. Be practical (E.g., Pick things you can finish in five years but don’t overshoot.)

4. Feel free to change topics if you discover your real interest

5. Acknowledge your personality

E.g., if you don’t like crowded places, don’t force yourself to stick in a crowded research area.

6. Good research balance: 1 focused direction + some side project for fun

Extension of Q1’s “personality” discussion

Question: How do you think of the use of Twitter for NLP researchers? There is so much PR on Twitter, but not all people feel comfortable with Twitter…

Answer: Don’t get overstressed. It can even take years for professors to establish a Twitter account, and years to decide to use it once they established it (e.g., Mohit and Rada). You are perfectly fine if you want to focus on research first, and tweet less.

• You can think of it as long-term versus short-term gains. If you care a lot about long term, then just do great work. If you care about the instant reception of the work, feel free to Tweet.
• On the other hand, some people use Twitter for professional reasons. Professors can choose to support their students by posting about their work.
• Social media can also be a platform for research conversations that may not happen elsewhere.
• Don’t worry too much if you are a student. Things will happen naturally as your career goes on. E.g., when you become a PI, you need to be responsible for some PR work, and you will naturally use Twitter.
• Always take a break from social media, and refresh your mind :)

Q2: How to find the right collaborators

• First of all, many collaborations will come naturally!
• If you want to actively reach out, check what types of work interests you, and make sure there is mutual understanding between you and the collaborators. Do not force yourself into collaboration just because of the title of that person you want to reach out to.
• Choose collaborators because you want to learn from them. They can provide you with valuable knowledge that is out of your domain.

Remember, it is a lot about the person:

• Focus on the person but not the publication itself. -> Collaboration can turn into long-term friendships!
• Choose collaborators with compatible work styles as yours.
• Collaboration works when you truly like the collaborator :)

Tips on initiating collaboration:

• Even as a student, you can take the initiative to contact your interested collaborators, and cc’ your supervisor.
• If you don’t have particular project ideas in mind, but really likes the work style of a person: You can set up a monthly chat with the person, and just brainstorm without pressure. Collaboration ideas will pop out!
• Good collaborations take time to establish :)

Collaboration makes NLP community prosper. Image credit to REF2021.

Q3: Breadth versus Depth when doing research?

• One clear focus & at the same time some explorations

*The reason is that you want to be remembered for being really strong at something (echoing with the 2nd suggestion of Question 1).

• Sometimes it is good to reach out for breadth near the end of your PhD (so that you already established your name as an expert in your focused field)
• A perspective to think about this is from your end goal.

E.g., at the end of your PhD, you need to have (for example) three really good, first-authored papers, and three strong recommendation letters. So you might have a clear solution in mind on how to balance breadth and depth, based on these end goals.

• Research diversity can happen very naturally

E.g., many people have internships during their PhD, so during internship, you will work in slightly different areas and extend your breadth.

Or e.g., choosing a good postdoc position is also another way if you want to work on a new topic. You can do work different from your original PhD theme.

Q4: What are the top pros and cons of staying in academia instead of industry

Advantages:

• Freedom: You choose your own topic, goal, and work-life balance. You can create your own problem instead of aligning with the company’s interest. You have the ability to work on not-for-profit, crazy problems :).
• Mentorship: Mentoring students is very very rewarding, and very different experience from mentoring interns in the industry.

Middle area:

• Salary concerns: In academia, you can surely have a good life — you can afford your house and car, but definitely not living in a castle. But if you wish to be more financially comfortable, you can start collaborating with industry, or have your own startup.

Disadvantages:

• You have more peer pressure. You need to do frequent self-adjustment and reflection in academia. (Nonetheless, you should be kind to yourself :)!)
• In academia, you have to multitask a lot, whereas in industry you are rather focused. Also, you have to be really self-organized.

Additional Suggestions:

• You can try to do a postdoc, or research scientist, and these experiences will help you get a better flavor of academia vs. industry.

My favorite quote from the discussion:

“People in the industry want to make an impact by building products. But for us who work in academia, our proud products are the students :).”

Thanks again to Prof Mohit Bansal (UNC Chapel Hill), Hanna Hajishirzi (University of Washington), Heng Ji (University of Illinois at Urbana-Champaign), and Rada Mihalcea (University of Michigan) for organizing the wonderful mentoring session! Their sharing of opinions on the four important topics are very inspiring for other researchers. Surely, this note cannot include all the anecdotes and fun stories in the meetup, but we really wish to spread the sparkles of our discussions to a broader audience.

Have fun reading :)!

Note-taker: Zhijing Jin (Incoming Ph.D. at Max Planck Institute, Germany)