How to Read, Write, Present PapersCaveatsSlide 3OmissionsSummaryReading a PaperWhy read papersWhy not to read papersRead or not to read, that is the questionWhat to readWhat’s in a paperHow to read a paper?What to noteSo this paper is in print ...How to WriteHow to write a paperSlide 17Slide 18Readability a mustDo not irritate the readerSlide 21How to write a theory paperHow to write a systems paperDiscuss related workTell them your shortcomingsHow to write weak resultsMiscellaneousTechnical reportsHow to PresentHow to present a paperHow to present a paper (at a conference)How to present a paper (at a conference)Talk outline or not ?TextSlide 35PowerPoint, but not excessivelyPicture is worth 1000 wordsShort talksSlide 39How many slides?Slide 41If English is your second languageNo substitute for experienceSlide 44Useful referencesThanks !How to Read, Write, Present PapersNitin H. VaidyaUniversity of Illinois at [email protected]© 2002 Nitin VaidyaCaveatsStatutory warning : Your advisor may not agreeOnly my opinions. Random thoughts, often in no particular orderUse advise at your own riskI do not necessarily follow the advise all the timeCaveatsThis presentation ignores some of its suggestionsOmissionsReferences at the end of the talk provide many suggestions not included in this talkSummaryUse common senseLearn from experienceReading a PaperWhy read papersSo you know what’s happeningAvoid reinventing the wheel does happen commonly, too many wheels alreadyFind interesting research topicsWhy not to read papersCannot read everythingShould not read everythingCan suppress innovationonce you see solutions using a particular theme, often hard to think differentlyRead or not to read,that is the questionRead, of courseKnow what’s importantKnow what can be ignored without significant loss of informationWhat to readMajor conferencesJournals are a few years behind, but still can be usefulTech reports from active research groupsneed to know which groups to look upSurvey / overview papersACM Computing SurveysCACM, IEEE Computer, Spectrummore technical - IEEE Personal Communications, …newsletters - ACM SIGCOMM, ACM SIGMOBILE, ...What’s in a paperAbstractIntroductionMotivationProblem descriptionSolution...Performance AnalysisConclusionsFuture WorkHow to read a paper?Know why you want to read the paperTo know what’s going on (e.g., scanning proceedings)title, authors, abstractPapers in your broad research areaintroduction, motivation, solution description, summary, conclusionssometimes reading more details useful, but not alwaysPapers you may want to improve onread entire paper carefullyWhat to noteAuthors and research groupNeed to know where to look for a paper on particular topicTheme of the solutionShould be able to go back to the paper if you need more infoApproach to performance evaluationNote any shortcomingsSo this paper is in print ...Be skepticalIf it sounds too good to be true, it often isHow to WriteHow to write a paperDo unto others as you would have them do unto youHow to write a paperWhen you have truly exceptional resultsP == NPProbably doesn’t matter how you write, people will read it anywayHow to write a paperMost papers are not that exceptionalGood writing makes significant differenceBetter to say little clearly, than saying too much unclearlyReadability a mustIf the paper is not readable, author has not given writing sufficient thoughtTwo kinds of refereesIf I cannot understand the paper, it is the writer’s faultIf I cannot understand the paper, I cannot reject itDon’t take chances. Write the paper well.Badly written papers typically do not get readDo not irritate the readerDefine notation before useNo one is impressed anymore by Greek symbolsIf you use much notation, make it easy to findsummarize most notation in one placeDo not irritate the readerAvoid Using Too Many AcronymsAUTMA ?!You may know the acronyms well. Do not assume that the reader does (or cares to)How to write a theory paperUnreadability is not the same as formalismReader should be able to understand contributions without reading all detailsIf some proofs are not too important, relegate them to an appendixProofs are not as worthy as new proof techniquesHow to write a systems paperProvide sufficient information to allow people to reproduce your resultspeople may want to reproduce exciting resultsdo not assume this won’t happen to your paperbesides, referees expect the informationDo not provide wrong informationSometimes hard to provide all details in available spacemay be forced to omit some informationjudge what is most essential to the experimentscite a tech report for more informationDiscuss related workExplain how your work relates to state of the artDiscuss relevant past work by other people tooRemember, they may be reviewing your paper.Avoid: The scheme presented by Vaidya performs terriblyPrefer: The scheme by Vaidya does not perform as well in scenario X as it does in scenario YAvoid offending people, unless you mustTell them your shortcomingsIf your ideas do not work well in some interesting scenarios, tell the readerPeople appreciate a balanced presentationHow to write weak resultsIf results are not that great, come up with better onesDo not hide weak results behind bad writingBe sure to explain why results are weaker than you expectedIf you must publish: write well, but may have to go to second-best conferenceOnly a few conferences in any area are worth publishing inToo many papers in poor conferences bad for your reputationJust because a conference is “IEEE” or “ACM” or “International” does not mean it is any goodIf results not good enough for a decent conference, rethink your problem/solutionMiscellaneousRead some well-written papersaward-winning papers from conferencesAvoid long sentencesIf you have nothing to say, say nothingdon’t feel obliged to fill up space with useless textif you must fill all available space, use more line spacing, greater margins, bigger font, bigger figures, anything but drivelTechnical reportsUseful to get early feedback from other researchersPuts a timestamp on your workCan include more
View Full Document