Lecture Series “ Stars for Tomorrow” at MSR Asia

Lecture Series “ Stars for Tomorrow” at MSR Asia

Research How to perform research ? How to write a good paper ? How to publish a (good) paper ? How to make a good presentation ?

Development The art and science of software development Patenting your invention Effective product transfer Microsoft technology roadmap

Culture Effective leadership Professionalism How to succeed in Microsoft and MSR ? Microsoft’s culture

Ya-Qin ZhangManaging Director

Microsoft Research AsiaApril 2002

How to Publish a (good) Paper?

Outline

When to write a paper ?What is a good paper?How to get a good paper published?

When to Write a Paper ?

Passion with your invention/conceptCompelled to speak and writeTruly novel concept/algorithm/procedure/architectureVision and survey that provide value for the research

communitySolid, mature, and sustainable results

What is a Good Paper ?

Right Subject MatterWell-Defined ProblemSimple and Compelling Clear Contributions Reliable and Reproducible ResultsRepeatable ProcedureGood structure and logic flow (Ref. Charles Lin’s talk)

Frequent Referrals

A few misconceptions

The more, the better Many new ideas

The bigger, the better A revolution, paradigm shift, …..

The more complex, the better Lots of math, theory, and formulas

The more selling, the better First-ever, the best, breakthrough

The more authoritative, the better Excessive use of own references and previous work

Three Steps in Publishing a Paper

Before Submissions – Choose a journal or conference Journals -> for formal evaluation and archival Conferences -> for quick presentation and interaction

After Submissions – Communicate with Reviewers/Editors Reviewers’ comments Revisions Communications with Editors Handling rejections

After the Publications – Expand the network Paper referral Follow-up work Communications w/ Readers

Step 1: Before Submissions – choose the right publications

Types of PublicationsJournals -> for archival

Correspondence; Regular paper; Invited paperConferences -> for presentation and interaction

Poster; Regular, Plenary, Keynote

Factors to ConsiderSubject MatterPrestige and ImpactExposure and VisibilityTimeliness and ResponsivenessCircle of Influence

After Submissions

Reviewers’ comments Revisions Communications with Editors Handling rejections Building a network

A Technical Journal

Sponsors and Publishers (e.g. IEEE, ACM, SPIE)Editorial Board Editor-in-Chief (1-2) Associate Editors (20-30) Publication Editor (1)

Reviewers (200-500)Authors Readers

Random Thoughts About Internet Ventures

Internet is not a bubble !

The greatest revolution ever that will profoundly transform the way we live

Tremendous opportunities awaiting for new technologies, products, markets, and ventures

The revolution just began

Editorial Board

Editor-in-Chief Appoints Associate Editors; Manages budget and operations of the journal; Resolves disputes between authors and AE; Makes final decision on paper acceptance and publications

Associate Editors Assigns reviewers Makes recommendations on the paper acceptance/rejection

Publication Editor Handling all logistics on manuscripting, proofreading, and

publications after acceptance

Review Process

1: Submit your paper to the Editor-in-Chief (EIC)2: EIC assigns a responsible Associate Editor (AE)3: AE identifies 3-5 anonymous reviewers4: AE makes a preliminary decision based on reviewers’ comments Acceptance (w/o or w/ minor revisions) Major revisions ( => Step 3) Rejection

5: AE makes final recommendation to EIC regarding the status of the paper6: EIC makes the final decision and inform the author7: Author then works with the Publication Editor (PE) to get the paper published

Reviewers

Experts and peers with in-depth technical knowledge on the subjectGives objective and professional assessment and feedback on the manuscriptTypical reviewers People who published several papers on the same subject (e.g. by AE

knowledge, your reference, …) People who have no direct conflict of interests w/ you

( not: your colleagues, your advisor/students, your relatives,…) People w/ different mix of background and seniority

(e.g. one big shot, 1-2 active researcher, and 1-2 post-PhD type) People who are within easy reach of the AE

An Example: IEEE T-CSVT Review Form

Copied

Rebuttal

When You submit a rebuttal Point-by-point detailed response to each reviewer Constructive and positive Clear and to-the-point Responsive (< 1 month)

It’s fine to disagree with the reviewers, AE may be on your side If there are many disagreements, exchange emails w/ AE in

advance, to minimize the # of rounds You need to make some compromise, but not on principles

It’s your paper !

Reviewers carefully read my paper [T][F]

[F] A review typically makes up his/her mind after 5-minute browsing: Title/author=> abstract=>conclusions => references => introductionThen spends < 1-H to justify (moving to main body of the paper)

Most readers follow the same patternYour Action: Make your points EARLY Bring up your results QUICK Highlight your contributions FAST

Reviewers are responsive [T][F]

[F] Reviewers are volunteers Reviewers have piles of papers to review Reviewers read your paper early if it’s “attractive” Reviewers read your paper early if he can learn things from it Reviewers read your paper early if his own work is related (or referred) Typically senior reviewers are less responsive but more important

Your Action: Put yourself in a reviewer’s shoes - visualize Make your paper easy to read, clear to follow, good to learn (see CL’s

part I: How to write a good paper)

Reviewers are Professional and Fair [T][F]

[T] although there are small % of exceptions Constructive critiques to improve the paper Mostly positive and constructive Do make some honest mistakes Some junior reviewers also want to establish their credibility (most AEs

come from good reviewers)Your actions: Engage a dialogue w/ reviewers via AE Make reviewers your friends Acknowledge your mistakes and make corrections Acknowledge reviewers if a good point is made Make clarifications if reviewers are wrong

Handling Rejections

Understand that most papers (> 70%) are rejected by a premier journal (e.g. IEEE Trans)No feeling of shame or losing faceThank AE/reviewers for their dedicationsAsk AE what changes I can make for resubmission, redirection to another journal, or withdrawDisplay class and style – walk away amicably

After Acceptance

Taking care of the logistics – precise and responsiveFollow up your own work if appropriatePay attention to follow-up work by othersPay attention to paper referralCommunicate w/ readersExpand your network

FAQs

Can I submit a paper to multiple journals/conferences ? ABSOLUTELY NOT ! It’s OK to have a conference presentation

followed by a journal article w/ significant enhancementsCan I recommend the AE for handling my paper ? No. But it’s OK to specify which AE to avoid under rare circumstance

What if I don’t hear from my AE for a long time (e.g. 6 months) ? Send a VERY friendly reminder, but don’t be too pushy Try not to involve EIC

What if I strongly disagree w/ AE’s decision ? It’s OK to appeal to EIC (don’t do it too often and with strong backup) What if I strongly disagree w/ EIC’s final decision ? It’s theoretically possible to appeal to IEEE TAB: never do it !

Conclusions

Content is the Key !Good writing skills are critical Communications skills are necessaryQuality > Quantity Understand why to publish

Building and expand the network of influence