Writing Research Papers Andreas Paepcke adapting and augmenting

Writing Research Papers Andreas Paepcke adapting and augmenting notes by Jennifer Widom 1

Decision Dynamics Impact on Overall Score Rating Impact of 1 pt increase Innovation <=. 5 pt Technical quality <=. 5 pt Presentation <=. 5 pt 2

Decision Dynamics Impact on Overall Score Impact on Odds of Acceptance Rating Impact of 1 pt increase Innovation <=. 5 pt Innovation >200% Technical quality <=. 5 pt Presentation 70% <=. 5 pt Technical quality 40% Presentation 3

Outline It! • • • Title Abstract Intro [Related Work] System Experiment Discussion [Related Work] Conclusion 4

Outlining: Not Just For High School • What do you want to communicate? • What goes where? • Figures 5

Title Options: • Descriptive: “A Tool for Measuring Interaction Speeds on Small Devices Under Varying Lighting Conditions” • Snappy: “Interaction Speeds on Small Devices” • With hook: “Floosy: Twofold Increase in PDA Reading Speeds” 6

Abstract • • • Don’t just grab text from paper State problem Your approach/solution Main contribution Very little background/motivation Include the main numbers/results 7

Introduction: Five Elements 1. What is the problem? 2. Why is it interesting and important? 3. Why is it hard? (e. g. , why do naive approaches fail? ) 4. Why hasn't it been solved before? (Or, what's wrong with previous proposed solutions? How does yours differ? ) 5. What are the key components of your approach and results? Include any specific limitations. 8

Example Intro: Cursor Trails What? 1. People lose sight of cursor Why bad? 2. Efficiency cost; confusion Why hard? 3. Large cursors are found, Why now? but disrupt work How solved? 4. Problem worse on new large displays 5. Cursor leaves visual echo while moved Significant gain in re-acquisition performance. 9

Other Intro Pitfalls • Don’t start with Adam and Eve • No cliché in first sentence: V 1: “With the explosion of the Web. . . ” V 2: “Forty-five million news pages are read online every weekday. ” • Don’t rile the reviewer • Be interesting 10

Related Work • Pro Early: – Critical to be defensive – Short enough – (Do mention most significant in Intro) • Pro Late: – Want reader to reach meat – Comparison understood only w/ body of paper 11

Related Work (cont) • Group similar approaches: “Several approaches animate the cursor continuously or on demand [5, 6, 7, 8, 9]. Others do away with cursors and magnify selected targets [10, 11, 12]. ” • Good rel work section contrasts with your work. Often: no space. • Check coverage of program committee 12

Main Body • • Briefly introduce required foreign material Introduce a running example Organize around figures Guide reader smoothly across thoughts; reader should want each new section. “These initial results suggested the need to verify that the red color does not harm. . . ” EXPERIMENT. . . 13

Experiments: What to Measure? Options: • • Completion times Latency Sensitivity to important parameters Scalability for: – screen size, – #of collaborators, . . . • Subjective preference • Others? 14

Experiments: What to Show Options: • Performance usable? • Relative performance to naive approaches • Relative performance to previous approaches • Relative performance among different proposed approaches • Others? 15

Experiment Setup Section • Goal (should be mostly clear by now) • What did you do? – Type of experiment (within/between) – Task • Data/Equipment used: How big, how fast, how much, preprocessing/cold-start? • People: how many, age range, gender, social distribution, . . . • Don't mention IRB unless relevant 16

Example Experimental Section • “We compared cursor trails against blinkies. . . ” • “In a within-participants design we had participants. . . ” • “On a 2560× 1600 30” display participants acquired. . . ” • “Ten women and twelve men ranging in age from. . . ” Goal What? Equipment Participants 17

Results • • • Only facts The numbers Failure cases Statistics Active vs. Passive Voice 18

Discussion • No new facts • Implications – Describe – How convincing? More studies needed? 19

Conclusion • Short summary • Drive home your main results • Future work – Say what you’re planning next – List promising directions 20

Bibliography • Use a tool for consistency • Check result carefully 21

Selected Challenges • That and Which: – That defines: The experiment that failed was well designed. – Which elaborates: The experiment, which failed, was well designed. • And others Et aliter Et al. • Use etc. only when blatantly clear • Avoid: for various reasons. . . 22

Selected Challenges Continued • Vary your non-technical words: “We see that the row that differs most from the other rows is the third one. ” “The third row maximally differs from others in the table. ” • Avoid thing and other globalities: “The thing we notice is that. . . ” “Observe that. . . ” 23

Selected Challenges Continued • Avoid far-referential use of this, that, these. “All participants committed many errors. That shows how fragile cognitive. . . ” Those errors show. . . This fact shows. . . We conclude from this outcome that. . . • Try it! 24

Finally. . . Don’t start the paper too late 25

References • The Elements of Style. Strunk and White (Grammar; sentence-level) • Made to Stick—Why Some Ideas Survive and Others Die. Chip Heath & Dan Heath. (Presentation level) http: //www. madetostick. com/ • Jennifer Widom’s paper writing page: http: //infolab. stanford. edu/~widom/paper-writing. html 26