Data Anonymization Competition

iPWS Cup 2024

(held in conjunction with IWSEC 2024)

What’s new

Useful links for competitions

About iPWS Cup 2024

Story

Company A wants to develop a movie recommendation system using customer data. The company decides to anonymize customer data for a competition to develop a recommendation system and make it available to the participants of the competition. However, even with the intention of anonymizing the data, there have been cases of personal identification and privacy breaches due to matching with external data. More recently, there has also been the problem of “database reconstruction attacks”, where even anonymized data can be combined with supposedly secure statistical data to reconstruct the original data. Can Company A create highly useful anonymized data while preventing personal identification attacks and database reconstruction attacks?

Competition Overview

Each participating team takes on the roles of both anonymizer and attacker, competing in data anonymizing techniques and attacks on the anonymized data. In the data anonymization phase, each team takes the role of a company that wants to publish customer data and aims to protect the privacy of the people in the data by anonymizing the given data. In the attack phase, each team becomes an attacker who wants to discover the contents of the data and aims to discover more secret information about the individual contained in the data anonymized by other teams.

Procedure

The competition consists of two phases.

Anonymization Phase: Each team will produce a set of anonymized data from fictitious data on movie ratings (each anonymized data will be processed from original multi-attribute data, from which only the attributes necessary for analysis will be extracted). The anonymized data must be processed in a way that makes it difficult for others to identify the original information, with as little loss of utility as possible.

Attack Phase: Each team restores the values of the original data, some of which have been redacted by the other team, using anonymized data (database reconstruction attack). They also try to identify individuals by linking the original data, from which names etc. have been removed, to their names etc. using anonymized data.

After these have been performed in turn, the results of each team’s anonymization and attacks will be evaluated by the organizer. The anonymization is evaluated in terms of the closeness of the analysis results obtained from the anonymized and original data (the closer, the better) and the difficulty of correctly guessing the secret data of the other team (the more difficult, the better), while the attack is evaluated in terms of the accuracy of the guesses made by the other team on the anonymized data (the more accurate, the better).

iPWS Cup 2024 schedule

Date Event
July 12th (Fri) - 24th (Wed), 2024 Team registration
July 26th (Fri) - August 16th (Fri) August 19th (Mon), 2024 Anonymization phase
August 20th (Tue) - September 10th (Tue), 2024 Attack phase
September 20th (Fri), 2024 Final presentation in Kyoto, Japan (on-site event)

Final presentation

Final Presentation Program

  1. 9:30- Reception
  2. 10:00-10:10 Opening Ceremony
  3. 10:10-11:00 Presentation of Team 01-05 (10 minutes each)
  4. 11:00-11:20 Break
  5. 11:20-12:10 Presentation of Team 06-10 (10 minutes each)
  6. 12:10-12:30 Awards and Closing Ceremony
    1. Group photo
  7. 12:30-14:00 Lunch & Open discussion

NOTICE
There will be a reception for iPWS Cup participants on 19 September at 18:00(JST).
If you would like to attend, please contact ipwscup2024-inquiry(at)csec.ipsj.or.jp (replace (at) with @) by 11 September. We will respond with more information.

Prizes

Documents/Resources

How to participate in the iPWS Cup 2024

Team requirement

How to apply for the competition

How to play the anonymization phase

How to play the attack phase

Participating teams

Team ID Team Name Message Representative Affiliation
01 Strive-Legends - - -
02 KUνττ Both the horse and Botchi escaped. Masaya Kobayashi Kanagawa University
03 Takenoko Movie Guardians Victory!! - -
04 Kamo - Koki Hamada NTT
05 brian Yay! Kabuto Okajima Virginia Tech / Gunma University
06 Thieves Sttk No more movie thief Taisei Tashiro -
07 Diamond - - -
08 TokumeiAnonymous ^^ - -
09 AnoNICSmous Let’s preserve privacy! Pablo Sanchez-Serrano NICS Lab, University of Malaga
10 re:Botchi All your reviews are belong to us Makoto Iguchi Kii Corporation
Team ID Account Disribution Data Selected Data
01 rq1543179 00, 04, 05 12, 16, 17 16
02 mkoba 35, 42, 47 38, 44, 49 38
03 takenoko 51, 52, 53 54, 56, 57 57
04 user1234 06, 07, 08 18, 19, 20 18
05 brian 25, 60, 89 26, 62, 90 90
06 tashiro 37, 43, 59 39, 45, 61 61
07 diamond2024 01, 02, 03 13, 14, 15 15
08 tokumei 32, 48, 87 34, 50, 88 88
09 pablosanserr 09, 10, 11 21, 22, 23 23
10 igucci 33, 55, 77 36, 58, 78 36

Final result

Overall

Best Attack

Best Presentation

Best Data Scientist

Committee

Chair
Koji Chida, Gunma University, Japan
Members
Masahiro Fujita, Mitsubishi Electric, Japan
Makoto Iguchi, Kii, Japan
Hiroaki Kikuchi, Meiji University, Japan
Ruiqiang Ma, Inner Mongolia University of Technology, China
Takayuki Miura, NTT, Japan
Yuichi Nakamura, SoftBank Corp., Japan
Takuma Hatano, NS Solutions Corporation, Japan

Contact

ipwscup2024-inquiry(at)csec.ipsj.or.jp (replace (at) with @)