Data Science view of the KDD 2014
Download as PPTX, PDF24 likes15,239 views
The KDD '14 conference saw significant growth with 151 research papers accepted out of 1036 submissions, resulting in a 14.6% acceptance rate. Key insights into the review process indicate that collaboration with a senior author and limiting submissions to five papers improve acceptance chances. Additionally, the conference emphasizes the need for clearer reviews and adherence to submission deadlines to enhance the quality of the peer review process.
Data Science view of the KDD 2014
- 1. Data Science of the KDD ‘14 Review Process Jure Leskovec (Stanford) and Wei Wang (UCLA) Joint work with Jason Hirshman and David Zeng (Stanford)
- 2. KDD 2014 Research Track Statistics
- 3. KDD 2014 Program Largest KDD program ever: • 151 research papers (20% growth over KDD’13) • 43 industry & govt. papers (30% growth) • 26 workshops (75% growth) • 11 tutorials (83% growth) Program highlights: • Paper spotlights early morning (8:15am) • Oral presentations (Mon-Wed) • Posters at the reception (Tue night)
- 4. KDD 2014 Research Track • 1036 submissions from 2600 authors – 42% increase over KDD ’13 • 151 papers: – Acceptance rate 14.6% 0 200 400 600 800 1000 1200 2000 2005 2010 2015 KDD year Numberofsubmissions
- 5. KDD Reviewing Process 46 Senior PC members + 340 PC members • 2971 reviews in total (Rough) Acceptance rule: • Raw review score AND Standardized review score AND Raw meta-review AND Standardized meta-review score ≥ Weak Accept • 110 papers matched (immediate accepts) • Remaining papers were discussed with meta-reviewers and final decisions were made
- 7. Acceptance Rate per Country
- 8. Acceptance by Subject Area
- 9. Predicting Paper Acceptance Features Used Accuracy Random Guessing 0.50 Paper Abstract 0.57 Author Status (Past paper counts) 0.64 Author Status (DBLP graph connectivity) 0.61 Author Status (Counts + Graph) 0.65 Reviewer (Similarity, Graph distance to authors) 0.60 All (Abstract, Author Status, and Reviewer) 0.65
- 10. Predicting Paper Acceptance from the Review Text Features Used Paper: Accepted? Review: Score > 0? Random Guessing 0.50 0.50 Review Text 0.68 0.72 Review Text + Numeric Score (Novelty, Presentation) 0.77 0.77 Human Reading of Review Text 0.88 0.73
- 11. I’m submitting a paper: What correlates with acceptance?
- 12. Academia + Industry Papers do Better
- 13. Submissions per Author: 5 is best!
- 14. No benefit in submitting >5 papers!
- 15. Having more authors (seems to) help
- 16. It is the most experienced author that matters!
- 17. What insights can we gain on the review process?
- 18. Most reviews are Weak Rejects
- 19. More granularity is needed at the Weak Reject / Weak Accept level Reviewagreeswiththefinaloutcome
- 20. Review length is a good determinant of a review’s influence/quality Reviewagreeswiththefinaloutcome
- 21. Shorter reviews are used for clear accepts and rejects
- 22. Never review co-author’s papers
- 23. The Curse of the Review Submission Deadline
- 24. Over 50% reviews submitted in the last 5 days Over 20% reviews submitted in the last 24 hours 10% of reviews submitted late
- 25. Ratings increase near the deadline Weak Rejects increase while Rejects decrease
- 26. Reviews submitted late are less likely to agree with final outcome
- 27. Late reviews are shorter
- 28. Review quality drops: Accuracy of predicting score from review text
- 29. Conclusions • To get your papers accepted to KDD: – Collaborate in multidisciplinary teams – Have a senior author on board – Do not submit more than 5 papers • To improve KDD community standards: – Avoid Weak Reject/Weak Accept scores – Write longer and clearer reviews – Submit reviews early!
Editor's Notes
- #8: Country of the paper is given by the mode author nationality. Only countries with more than 10 submissions are shown (except South Korea, which had 0 acceptance)
- #9: Subject areas were based on a field that authors tagged their papers with. Only subject areas with more than 50 submissions are shown.
- #10: On balanced dataset (we subsampled negative examples)
- #11: On balanced dataset (we subsampled negative examples)
- #14: Academic: All authors of paper are affiliated with a university Industry: All authors of paper are affiliated with industry Mixed: Paper has authors affiliated with both universities and industry.