Eastwind’s Employees take Second Prize in SNA Hackathon
18 april 2016
Eastwind’s employees took the second prize in SNA Hackathon event which was held in Ekaterinburg from February 8 through April 8. Eastwind finished second in the second IRL stage of the competition which took place in Ekaterinburg on April 6-8, as part of the V Conference AIST 2016 dedicated to analysis of images, social networks and texts.
This time, the marathon offered a task which is very important for any social network: prediction of possible social relations.
The competition included two stages. The first online stage challenged the participants to predict social connections by a fragment of a social graph and demographic characteristics of a 1 million users. Those who were the most successful and exceeded the accuracy threshold of the prediction algorithm made it to the top 50 and moved on to the next stage.
The second stage put the contestants in teams: they had to improve the pre-developed prediction algorithms by supplementing them with the data on the users’ membership in societies and with counters of interactions between different types of users. Two days were given to solving the task, and then each team had to show their results in the form of presentations. A competent jury consisting of IT specialists and scientists assessed the presentations. The verdict was based on the formal metric of the prediction accuracy, as well as on the applicability and novelty of the suggested models.
Ivan Kargapoltsev, the mastermind and the main driver of Eastwind team says: “Eastwind team had already possessed the expertise in social relations analysis, but only with regard to mobile operators. Our participation in Hackathon has proved that our experience is unique and based on the fundamental insight into the ways people interact, irrespective of the environment. This is the main conclusion we have made in the course of the competition, and the most important reward for the team.”
More News
This time, the marathon offered a task which is very important for any social network: prediction of possible social relations.
The competition included two stages. The first online stage challenged the participants to predict social connections by a fragment of a social graph and demographic characteristics of a 1 million users. Those who were the most successful and exceeded the accuracy threshold of the prediction algorithm made it to the top 50 and moved on to the next stage.
The second stage put the contestants in teams: they had to improve the pre-developed prediction algorithms by supplementing them with the data on the users’ membership in societies and with counters of interactions between different types of users. Two days were given to solving the task, and then each team had to show their results in the form of presentations. A competent jury consisting of IT specialists and scientists assessed the presentations. The verdict was based on the formal metric of the prediction accuracy, as well as on the applicability and novelty of the suggested models.
Ivan Kargapoltsev, the mastermind and the main driver of Eastwind team says: “Eastwind team had already possessed the expertise in social relations analysis, but only with regard to mobile operators. Our participation in Hackathon has proved that our experience is unique and based on the fundamental insight into the ways people interact, irrespective of the environment. This is the main conclusion we have made in the course of the competition, and the most important reward for the team.”