1 / 15

WING Monthly Meeting SIGIR 2014 Debrief

WING Monthly Meeting SIGIR 2014 Debrief. 25 th July 2014 By Jovian Lin. Stats. 387 full paper submissions 6% increase from last year 82 (21%) were accepted. Stats: Top Countries (Full Papers). Stats: Top Topics (Full Papers). Selection Process (1/2).

doyle
Download Presentation

WING Monthly Meeting SIGIR 2014 Debrief

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. WING Monthly MeetingSIGIR 2014 Debrief 25th July 2014 By Jovian Lin

  2. Stats • 387 full paper submissions • 6% increase from last year • 82 (21%) were accepted.

  3. Stats: Top Countries (Full Papers)

  4. Stats: Top Topics (Full Papers)

  5. Selection Process (1/2) • Two-tier double-blind review process. • At least 3 reviewers for each paper. • Then Primary Area Chair leads a discussion; produces a meta-review. • Secondary Area Chair (assigned to each paper) double checks the reviews and discussion, and may provide additional reviews

  6. Selection Process (2/2) • The PC chairs ranked the submitted papers by the meta-review score and then by the average score of the three reviewer scores, carefully examined the reviews and associated discussion. • The PC chairs first identified "clear accepts" and "clear rejects". • Then the undecided papers were carefully discussed in a face-to-face PC meeting held in Amsterdam, which involved all available Area Chairs.

  7. Additional Info (1/2) • Tutorials: 7 out of 17 accepted. • Workshops: 7 out of 11 accepted. • Demos: 16 out of 34 accepted. • Doctoral Consortium: 8 out of 11 accepted.

  8. Additional Info (2/2) • The short papers track received 263 submissions (3% increase over last year)… • …and accepted 104 of them (40% acceptance compared to 34% last year). • For the second year, the short papers are 4 pages long.

  9. The Future… SIGIR'15: Santiago, Chile (Due: Jan 28, 2015) SIGIR'16: Pisa, Italy SIGIR'17: Tokyo, Japan CIKM'15: Melbourne, Australia CIKM'16: Indianapolis, US WSDM'15: Shanghai, China JCDL'15: Tennessee, US

  10. Best Papers • The SIGIR 2014 Best Paper Award was presented to Giuseppe Ottaviano and RossanoVenturini for their paper “Partitioned Elias-Fano indexes.” • The Best Student Paper Award was awarded to Dmitry Lagun, Chih-Hung Hsieh, Dale Webster, and VidhyaNavalpakkam for their paper “Towards better measurement of attention and satisfaction in mobile search.”

  11. Learning Similarity of Functions for Topic Detection in Online Reputation Monitoring • Problem: • Reputation management experts have to monitor Twitter constantly to see what is being said. • Real-time online opinions are now key to understand the reputation of organizations/individuals. • Managing it manually is costly and unfeasible.

  12. Learning Similarity of Functions for Topic Detection in Online Reputation Monitoring • Solution: • Seeing “reputation monitoring” as a topic detection task • Task Definition: • Given an entity (e.g. Yamaha) and a set of tweets relevant to the entity, the task consists of identifying tweet clusters, where each cluster represents a topic/event/issue/conversation being discussed in the tweets… • … just like how it’d be identified by a reputation management expert.

  13. Learning Similarity of Functions for Topic Detection in Online Reputation Monitoring • What they did: • Learn a pairwise tweet similarity function from previously annotated data. • Predict whether 2 given tweets are about the same topic or not. • Use: terms, concepts, hashtags, urls, author, namedusers, timestamp • Use clustering algorithm (based on similarity function) to detect related topics. • Found: Twitter signals can be used to improve topic detection (compared to content-only).

  14. PHOTOS!!!

  15. Thank You

More Related