on behalf of the CIKM Cup 2016 Organizing Committee and CIKM Chairs, I would like to congratulate the winners and all of you with the impressive results and thank for a highly engaged participation! Cumulatively all of you have made over 1000 submissions in less than 2 months and the top-1 winner has improved the baseline by 14.2%. That's amazing!
The winners are:
1. Si Luo, Yan Ming, Wu Chen (team minerva, Ali-Search)
2. Dmitrii Nikitko
3. Deqiang Kong, Jingyuan Tang (team red fruit yard)
4. Martin Wistuba
5. Joao Rafael de Moura Palotti
During CIKM 2016 happening in Indianapolis, USA on October 24-28th 2016, we will have:
* an award ceremony at the 25th CIKM Anniversary Banquet (October ~26th, exact date TBD soon);
* a dedicated CIKM Cup 2016 Workshop (October 28th), where you could present the approach in person and meet with the fellow participants and other IR/DB/KM/ML researchers.
To share your approach with the community, please follow the guidelines below. Prepare a report (4-8 pages double column ACM tight LaTeX template: https://www.acm.org/publications/proceedings-template) or/and detailed deck (30-50 slides) as follows:
* Title mentioning CIKM Cup 2016, e.g. "Cross-Device Matching for Online Advertising with Ensembling Learning Techniques at CIKM Cup 2016"
* Abstract/Summary (1-2 paragraphs abut the models tried, lessons learned)
* Problem motivation / Application (expand on: https://competitions.codalab.org/competitions/11161#learn_the_details-overview)
* Problem formalization
* Related work (in addition to the one shared: https://competitions.codalab.org/competitions/11161#learn_the_details-related_work)
* Unique domain/data specific insights that you have uncovered (e.g. what did you help boost NDCG dramatically?)
* Methods and techniques tried and lessons learned
* Method formalization
* Experiments/Results/A comparative table of all approaches tried and how it evolved over time
If you plan to present your method during CIKM, notify me at spirin2 illinois edu. To submit the report (DEADLINE: October 26th), also use this email.
Finally, the data provider and partner of the competition (DIGINETICA) allowed us to share the correct labels in case you want to make more experiments or for report prep. You can access it here: https://drive.google.com/open?id=0B7XZSACQf0KdXzZFS21DblRxQ3c