International Workshop on Computational Surprise in Information Retrieval (CompS’18)
July 12, 2018, Ann Arbor, Michigan
HELD IN CONJUNCTION WITH
Introduction
Today we are exposed to vast amounts of information online and we often cope by paying attention to what well-known search engines or recommender systems can provide. Current information retrieval systems have been criticized as reinforcement of the same, relatively limited set of information rather than promoting unexpected exploration and discovery. Meanwhile, Facebook has admitted that its algorithms form "echo chambers" . The notion of "filter bubbles" are one of the prime motivations behind this workshop on computational surprise - bursting the bubble through thoughtful design of digital systems.
Natural human information seeking processes are full of surprises, from finding a movie at a friend's house to scientific breakthroughs like the discovery of X-rays. Miksa described human acquisition of information as a relatively unfocused sense of inquiry, and said that information seeking is ``better conceived as an exploratory and game-like mechanism rather than a precise response mechanism". However, even today, very few information retrieval systems support systematically discovering surprise. Computational models of surprise are in their infancy, due to the vague and elusive nature of the concept. In addition, surprise by nature contradicts intention and control. The result is that the word surprise remains imprecisely defined, relying on its use in everyday vocabulary. How to scientifically quantify surprise and how to incorporate it into IR systems are issues of ongoing research.
This workshop brings together researchers and practitioners to identify a set of core research questions in computational surprise. This will include discussion of a broad, multi-purpose research agenda. First, research collaboration: it is a networking opportunity intending to promote rigorous research on computational surprise in IR. Second, support for researchers: it will help attract research funding and resources in this area. Finally, it will help broaden the definition of information retrieval at the SIGIR conference.
This workshop will also discuss the challenge of data availability. Any machine learning or computational approach highly relies on high quality corpus that requires human intelligence. So far such data that contains human annotations or evaluations on surprise is not easily available. We expect that the discussion on data will result in a collaborative effort for producing such a dataset.
Schedule
On the afternoon of July 12 (Thursday), the workshop is in Room Michigan (Floor 2).
- 1:30pm - 2:30pm: Keynote Talk from Professor Laurent Itti (http://ilab.usc.edu/itti/)
- 2:40pm - 3:00pm: Paper Presentation 1: Detect Clickbait: From the Aspect of User Behavior
- 3:00pm -3:30pm: Break, per the main conference schedule
- 3:30pm -3:50pm: Paper Presentation 2: ARCHEAN: Facilitating Discovery in Social Science
- 3:50pm – 4:10pm: Paper Presentation 3: Text that Arouses the Feeling of Surprise: A Human Labeled Corpus through Crowdsourcing
- 4:10pm – 4:30pm: Paper Presentation 4: Human Perception of Surprise: A User Study
- 4:30pm – 4:50pm: Paper Presentation 5: SWDE: A Sub-Word And Document Embedding Based Engine for Clickbait Detection
- 4:50pm – 5:00pm Wrap-up
Accepted papers will have 15 minutes oral presentation and 5 additional minutes for questions & answers.
Important Dates
May 14, 2018: workshop papers dueMay 25, 2018: workshop paper notificationsJune 8, 2018: camera-ready deadline for workshop papers- July 12, 2018: Workshop Day at SIGIR 2018
Accepted Papers
We have 5 accepted papers:
- Detect Clickbait: From the Aspect of User Behavior. Yunqiu Shao (Tsinghua University), Yiqun Liu ( Tsinghua University), Jiaxin Mao (Tsinghua University), Cheng Luo (Tsinghua University), Huanbo Luan (Tsinghua University), Min Zhang (Tsinghua University), Shaoping Ma (Tsinghua University).
- ARCHEAN: Facilitating Discovery in Social Science. Scott Appling (Georgia Tech Research Institute), Alexandra Trani (Georgia Tech Research Institute), Erica Briscoe (Georgia Tech Research Institute).
- Text that Arouses the Feeling of Surprise: A Human Labeled Corpus through Crowdsourcing. Fakhri Abbas (University of North Carolina at Charlotte), Xi Niu (University of North Carolina at Charlotte).
- Human Perception of Surprise: A User Study. Nalin Chhibber (University of Waterloo), Rohail Syed (University of Michigan), Mengqiu Teng (University of Michigan), Joslin Goh (University of Waterloo), Kevyn Collins-Thompson (University of Michigan), Edith Law (University of Waterloo).
- SWDE : A Sub-Word And Document Embedding Based Engine for Clickbait Detection. Vaibhav Kumar (International Institute of Information Technology) , Mrinal Dhar (International Institute of Information Technology), Dhruv Khattar (International Institute of Information Technology), Yash Kumar Lal (Manipal Institute of Technology), Abhimanshu Mishra (Birla Institute of Technology), Manish Shrivastava (International Institute of Information Technology), Vasudeva Varma (International Institute of Information Technology).
Invited Speaker
Program Committee
- Xi Niu (University of North Carolina at Charlotte), PC Chair
- Wlodek Zadrozny (University of North Carolina at Charlotte), PC Member
- Kazjon Grace (University of Sydney), PC Member
- Weimao Ke (Drexel University), PC Member