First page > Competitions > Competition details


Teams: 607

Participants: 691

Competition submissions: 705

Competition background

Disparate networks, including social networks, communication networks and biological networks, are playing an increasingly important role on natural and socio-economic systems. A core problem, therein, is to measure the significance of individual nodes. For instance, a super spreader in Hong Kong triggered transmission of SARS to a significantly greater number of other people than 100 normal infected persons;a rumor re-tweeted by a celebrity may spread much broader than that by an obscure person.

Therefore it is necessary to develop a method to identify thevirulence genes in large-scale gene regulatory networks, to find the super-spreaders in large-scale social networks, and to detect the key enterprises with serious systematic financial risk in large-scale financial networks.

Those tasks could be formalized as a generic challenge that is identifying vital nodes in networks that are important for sustaining connectivity .This challenge, aka optimal percolation, is a well-documented issue in network science. With great anticipation of making big progress on this problem, we successfully invited some experts and hope the great participants will create novel and effective solutions。


① Collaboration with the Academic Board Member

   -The solutions of outperforming teams will be commented by the Academic Board Member.

   -The competitive participants can collaborate with the Academic Board Member to wrap up their solutions into patents or publications.

② Academic visit

   -The outperforming participants has the opportunity to be visiting scholars of the Academic Board Member.

 ③ Exclusive secret gifts from Academic Board Member

   -Invitation letters from the Academic Board Member

   -Secret Gifts signed by the Academic Board Member

List of Academic Board Member(AMB):

DataCastle has invited 8 excellent experts in the complex network area to be the Academic Board Member of this competition, who would provide domain-specific advice to participants and judge the competition. Furthermore, the competitive participants could collaborate with the Academic Board Member to wrap up their solutions into patents or publications.

1、Wei Chen

Title:Professor of Tsinghua University

2、Petter Holme

Title:Professor of Tokyo Institute of Technology、Professor of Sungkyunkwan University

3、Yanqing Hu

Title:Professor of Sun Yat-sen University

4、Linyuan Lü

Title:Professor of Hangzhou Normal University

5、Hernan Makse  

Title:APS Fellow,Professor of City College of New York

6、Flaviano Morone

Title: Postdoc presso City College of New York

7、Haijun Zhou

Title:Researcher of Chinese Academy of Sciences

8、Tao Zhou

Title: Professor of  University of Electronic Science and Technology of China

Time schedule

① Stage 1 (2017.6.13 - 2017.10.15)

  Participants can submit their results to the platform of DataCastle to see their positions in the rankings.

② Stage2(2017.10.16-2017.10.20)

  The top-50 teams need to upload their solutions.

③ Stage3(2017.10.21-2017.11.19)

  The solutions will be evaluated by the Academic Board Member. DataCastle would make effort to help participants establish connections to the Academic Board Member. DataCastle does hope the participants can successfully build cooperation with the Academic Board Member and wrap up the solutions into patents or publications.

④ Stage4(2017.11.28 )

The outperforming teams and their cooperative partner (Academic Board Member) will be invited to present their solutions and accept the award in the annual summit of DataCastle.

Participating and team rules

The captain is responsible for the team member, who is agree with the< DC competition's cheating management regulations >and the other pertinent rules.



团队人数上限5people。 在第一阶段的最后3天无法新建队伍,但是可以加入其它队伍。 在最后一个阶段最后3天无法新建队伍,无法加入队伍。 竞赛进入历史阶段后解除一切限制,注意:答辩队伍成员仅限活跃期间加入的成员。

Scoring criteria

Scoring algorithm:
Scoring criteria

The performance of algorithms is measured by the damage to network’s robustness. Parameter p is defined as the proportion of removed nodes, σ is the size of the giant component of the remaining networks (in proportion) after removing p of nodes. The σ-p curve could be derived from plotting p on the x-axis and σ on the y-axis. Robustness is defined as the area under the σ-p curve , mathematically reads as:

whereis the size of the giant component after removingof nodes from the network. The smaller R-value is, the better results the algorithm achieves. 

We would calculate Robustness for each network, such as ,R represents robustness value for network j.The performance of the participants would be measured by the average of the Robustness value of the 8 networks.

Considering the incomplete result submissions would cause damage to participants who submit the result cover all 8 networks completely we could automatically set R-value as 1 to those networks without results. . The final result reads as:

the smaller the R-value is, the higher the rank is.

Note:View the "Master Competition" score algorithm detailed description and implementation, click here.

  • © 2013-2017 DataCastle 蜀ICP备14018015号-2