METHOD OF CREATING RANKING, A SYSTEM FOR CREATING RANKING AND A RANKING

Abstract

A method of creating ranking, especially a ranking of players or teams in esport gameplay, on the basis of historical data, with the use of neural networks, including: neural network learning on the basis of historical data, and particularly data on the parameters of esport gameplays and the results of these gameplays; creating ranking of players or teams through a neural network on the basis of current data, and particularly data on the parameters of esport gameplays and the results of these gameplays; and possible forecasting by a neural network the result of the future game between given players or teams. The embodiments also include the system for implementation of this method and a ranking created by the use of this method.

Claims

1. A method of creating a ranking of players or teams in esport gameplay, on the basis of historical data, with the use of neural networks, including the following steps: a) neural network learning on the basis of historical data, and data on the parameters of esport gameplay and the results of this gameplay; b) creating a ranking of players or teams through a neural network on the basis of current data, and data on the parameters of esport gameplay and the results of this gameplay; and c) possible forecasting, by a neural network, the result of the future game between given players or teams.

2. The method according to claim 1, wherein the mentioned data on the parameters of esport gameplay and the results of the gameplay include at least the rankings of players participating in the game, and one or more of the following information: game result—as: winning, losing, tie; numeric game result—as a number of points of one team and a number of points of the other team; ID of a player and a character they are using in the game; information on the course of the game—including: the number of killed opponents, the amount of acquired goods, data related to entire teams, information on the course of the game performed by individual players—including: the number of killed opponents, the amount of acquired gold.

3. The method according to claim 1, wherein the neural network is in the form of a Transformer algorithm, deep neural networks, or a convolutional neural network shall be used.

4. The method according to claim 1, wherein the data is collected and/or the ranking is made available via the internet.

5. The system for creating a ranking of players or teams in esport gameplay, on the basis of historical data, with the use of neural networks, including one or more computers having access to historical data, and data on the parameters of esport gameplay and the results of this gameplay, wherein said computers are configured and programmed to perform the method of claim 1.

6. A ranking of players or teams in esport gameplay, derived from implementation claim 5.

Description

DETAILED DESCRIPTION OF THE EMBODIMENTS

[0025] The invention will be now closely discussed showing its preferred embodiments.

[0026] As mentioned above, the proposed method of creating a ranking is based on a construction of a neural network model using these data, where the said neural network—after the stage of learning from historical data and the results—forecasts results of the games between the teams and also assigns individual rankings to players. For the purpose of creating a ranking it is not excluded to use any of the historical data derived from historical games. It is recommended to use data in the form of at least the rankings of players participating in the game.

[0027] For example, rankings can assume the use of the following data: [0028] a) Rankings of individual players and a game result—as: winning, losing, tie, and preferably solely these data. [0029] b) Rankings of individual players and a numerical game result—a number of points of one team and a number of points of the other team, and preferably solely these data. [0030] c) ID of a player and the character they are using in the game (a player can use different characters in different games). [0031] d) Additionally (in relation to points a)-c)—information on the course of the game—including the number of killed opponents, amount of acquired goods, data related to entire teams. [0032] e) Additionally (in relation to points a)-d)—information on the course of the game of the individual players—including: the number of killed opponents, the amount of acquired gold etc.

[0033] According to the invention, a neural network first learns on the basis of data on games (i.e. such as listed above) and the results of these games. Next, on the basis of current data—a neural network is used for forecasting results of future games and scheduling team rankings, competitors rankings etc. The term “neural network”, used in the herein specification, shall be understood according to the subclass G06N3/02 of the class G06N3/00 of the International Patent Classification. In particular, neural networks known i.e. from the patent publications US20200057889A1—“convolutional neural network”, US20200077282A1—“deep recurrent neural network” or US20200065479A1—“deep neural network” can be used. Using the algorithm Transformer—known from the USA patent publication no. US20190354567A1 “Universal transformer” also will be possible.

[0034] The invention can be applicable, among others, to the game League of Legends, DotA 2 or StarCraft.

[0035] It is planned to set up a webservice which would create rankings for players of the games listed above, i.e. League of Legends, in the way specified above. The system will be available for any internet user from any country. The proposed system requires an enormous computing power, which is the reason why it has been available to practical use only recently—the newest computer systems have sufficient computing power to calculate neural network models as big as those anticipated in the present invention.