KR20180063912A - Method and apparatus for estimating operational results of online game service - Google Patents

Abstract

Provided is a method of predicting an operational result of an online game service provided by an online game providing system. The method includes: a step of extracting behavioral pattern data of a user character from game log data; a step of learning a first correlation between behavioral pattern data corresponding to a cause and behavioral pattern data corresponding to a result in the extracted behavioral pattern data, thereby generating a first behavioral model which has learned the first correlation; a step of learning a second correlation between the extracted behavioral pattern data and operational factors of the online game service, thereby generating a second behavioral model which has learned the second correlation; and a step of predicting an operational result of the online game service in regard to a newly determined operational factor by using the first and second behavioral models.

Description

METHOD AND APPARATUS FOR ESTIMATING OPERATIONAL RESULTS OF ONLINE GAME SERVICE [0002]

The present invention relates to a method and apparatus for predicting the operation result of an online game service, and a method and apparatus for predicting an operation result of an online game service for predicting profitability of an online game service.

In recent years, there has been a growing interest in various predictive modeling methods for predicting the operation result of game service through prediction of in-game behavior of gamers.

Despite this trend, studies on predictive modeling related to conventional game services are limited to modeling studies for observing behavioral data of gamers and then predicting outcome behavior.

As a concrete prior art patent, there is a system for learning the game style of the gamer and transplanting it into the NPC (Non-Player Character). This is an example limited to predictive modeling for analyzing game style of gamer for improvement of game service. It is general purpose such as disposition of monsters decided for game operation such as game departure, purchase, and interest, method of arranging features, It can not be applied to gamer behavior prediction for analyzing profitability of game service based on operational factors. Therefore, a specific profitability forecasting method for the operation of the game service is needed.

Accordingly, it is an object of the present invention to provide a method and apparatus for predicting changing gamer behavior, linking predicted gamer behavior and an operating factor for managing game service, to predict the operation result of an online game service such as profitability of an online game service And to provide that device.

The problems to be solved by the present invention are not limited to those mentioned above, and other solutions not mentioned can be clearly understood by those skilled in the art from the following description.

According to an aspect of the present invention, there is provided a method of predicting an operation result of an online game service, the method including extracting behavior pattern data of a user character from game log data; Learning a first correlation between behavior pattern data corresponding to a cause and behavior pattern data corresponding to a result from the extracted behavior pattern data to generate a first behavior model in which the first correlation is learned; Learning a second correlation between the extracted behavior pattern data and an operating element of the online game service to generate a second behavior model in which the second correlation is learned; And predicting an operation result of the online game service for the newly determined operating element using the first and second behavior models.

An apparatus for predicting an operation result of an online game service according to another aspect of the present invention includes a game operation server for extracting behavior pattern data of a user character from game log data; Learning a first correlation between behavior pattern data corresponding to a cause and behavior pattern data corresponding to a result from among the extracted behavior pattern data input from the game operating server and comparing the first behavior model with the first behavior model A first behavior model generation unit for generating a first behavior model; Generating a second behavior model in which the second correlation is learned by learning a second correlation between the extracted behavior pattern data input from the game operating server and an operating element of the online game service, part; And an operation result predicting module for predicting an operation result of an online game service for a newly determined operating element by using the first and second behavior models.

According to the present invention, a specific profitability prediction method for commercial service operation is required. To this end, a new prediction method for analyzing the correlation between the existing behavior prediction modeling method based on the machine learning methodology and the game service operation element is added , It can be used as an important index for determining the game operation policy by not only estimating the game style of the user but also estimating the game style of the predicted user and the profitability generated thereby.

The effects of the present invention are not limited to those mentioned above, and other effects not mentioned can be clearly understood by those skilled in the art from the following description.

1 is an overall system configuration diagram including an online game providing system according to an embodiment of the present invention.
2 is a functional block diagram illustrating an internal configuration of the game operating server shown in FIG.
3 is a diagram schematically illustrating a process of predicting operation results in the operation result prediction module shown in FIG.
4 is a flowchart illustrating a method of predicting an operation result of an online game service according to an embodiment of the present invention.

The various embodiments of the present invention are capable of various changes and may have various embodiments, and specific embodiments are illustrated in the drawings and the detailed description is described with reference to the drawings. It should be understood, however, that it is not intended to limit the various embodiments of the invention to the specific embodiments, but includes all changes and / or equivalents and alternatives falling within the spirit and scope of the various embodiments of the invention. In the description of the drawings, like reference numerals are used for similar elements.

The present invention adds a new prediction method that can analyze the correlation between the behavior prediction modeling method based on the machine learning methodology and the operating element of the online game service, and adds an operating element (operating there is no attempt to predict the changing behavior of gamer behavior.

Hereinafter, embodiments of the present invention will be described in detail with reference to the accompanying drawings.

FIG. 1 is an overall system configuration diagram including an online game providing system according to an embodiment of the present invention, and FIG. 2 is a functional block diagram illustrating an internal configuration of the game operating server shown in FIG.

Referring to FIG. 1, the overall system may include an online game providing system 100 that provides online game services to the user client 10 via a user client 10 and a communication network 20.

The user client 10 may be any type of computing device that has a communication function to connect to the communication network 20 and executes the online game service provided by the online game providing system 100.

Although not shown, the user client 10 may include a controller configured by a microcomputer or a central processing unit (CPU), a nonvolatile storage medium storing digital data such as a hard disk drive (HDD) An input unit such as a keyboard, a keypad, a mouse, a joystick, and a microphone, and an input unit such as a CD-ROM or DVD-ROM driver, a display unit, Wireless communication unit for connecting to the providing system 30, and the like.

Such a user client 10 may be, for example, a notebook PC, a desktop PC, a cellular phone, a Personal Communications Services phone (PCS phone), a synchronous / asynchronous International Mobile Telecommunication 2000 A personal digital assistant (PDA), a smart phone, a wireless application protocol phone (WAP phone), a game machine (Playstation), and the like.

The communication network 20 may be configured without discriminating the communication modes such as wired and wireless, and may be a personal area network (PAN), a local area network (LAN), a metropolitan area network ), A wide area network (WAN), an Internet network, a mobile communication network, or a heterogeneous communication network in which these are mixed.

The online game providing system 100 provides an online game service to the user client 10 through the communication network 20 and in particular the online game providing system 100 may include an operating element the operating pattern of the online game service is predicted based on the prediction result of the gamer's behavior pattern.

The online game providing system 100 includes a game operation server 110, first and second behavior model generation units 120 and 130, first and second storage units 140 and 150, Module 160. The < / RTI >

Game operating server 110,

The game operating server 110 is a broad concept representing a computer in which a program for performing a server function is installed so that game users can play the game.

2, the game operation server 110 includes a graphics / sound operation engine 110A for managing overall output of game related graphics / sounds, A game event management module 110C for collectively managing game events of users connected to the system 100, a game related event management module 110C for managing the game related scenarios, A behavior pattern extraction module 110E for extracting behavior pattern data of a user character from game log data for a user's game use record and a game scenario management module 110D for collectively managing the game progress of users .

In the present embodiment, the game log is information on the use history of the game as the user proceeds the game, and the game log includes information on the subject and action of the action and additional information when the action occurs in the game use of the user Lt; / RTI >

More specifically, the system log, the character log, and the NPC log (all actions that are not the subject but are not the system log) can be included according to the subject of the action.

The game log may include information such as account name, character name, character class, character level, login information, game asset related information, game cost related information, personal information, login history information, membership withdrawal, . ≪ / RTI >

In addition, the game log may include information on a reason why a user performs an action, information on time, location coordinate information on a game where an action occurs, and information on an object of an action.

The action pattern extraction module 110E extracts action pattern data from the game log data described above and stores the extracted action pattern data in the game operation related database 110F.

In addition to the behavior pattern data, the game operation related database 110F may store various operating element sets related to the profitability of the game service among the information provided from the configurations 110A to 110E as separate items.

The profitability of the game service can be determined based on membership, withdrawal of membership, history of item purchase, etc., and is an operating element closely related to the profitability of the game service. For example, the monster arrangement on the map, A game commodity compensation rule according to the number of missions or attendance, and a distribution rule of the user character's ability.

Hereinafter, a method of predicting profitability of a game service including withdrawal of a member and purchase of an item using a learned behavior model based on behavior pattern data provided by the game operation server 110 Let's take a look at an example.

The first behavioral model Generating unit (120)

1, the first behavior model generation unit 120 generates behavior pattern data corresponding to the cause and behavior pattern data corresponding to the result using the behavior pattern data provided from the game operation server 110 And the correlation between them. Here, the behavior pattern data corresponding to the result includes data related to membership withdrawal and data related to item purchase.

Specifically, the first behavior model generation unit 120 uses a machine learning method based on a principle of back propagation to calculate a correlation between behavior pattern data corresponding to the cause and behavior pattern data corresponding to the result And calculating the weighted value obtained by digitizing the learned correlation, the first behavioral model can be generated.

That is, the first behavior model generation unit 120 learns the correlation of whether the user's character evokes a desire to withdraw a member or an item when the user shows a behavior pattern pattern in the game, and the user's character If you see a behavior pattern that hunts a monster placed in a specific layout, if the membership withdrawal increases, you will see a behavior pattern that is the cause of hunting the monsters placed in a specific layout and a behavior pattern Have a close correlation with each other and can learn a behavioral model in a way that gives a high weight to such a correlation.

Examples of the machine learning method based on the principle of back propagation include decision tree, random forest, SVM, and deep learning.

In this manner, the first behavior model generated by the first behavior model generation unit 120 is stored in the first storage unit 140.

The second behavioral model Generating unit (130)

The second behavior model generation unit 130 learns the correlation between the behavior pattern data corresponding to the cause and the game operation element using the behavior pattern data provided from the game operation server 110, The second behavioral model can be generated by calculating the weighted value. Here, the game operating element may be a monster arrangement on a map, a topical object, a game commodity compensation rule according to a mission or a number of times of attendance in a game, a rules for allocating a capability value of a user character, and the like.

Specifically, the second behavior model generation unit 120 learns the correlation between the behavior pattern data corresponding to the cause and the game operation element using the machine learning method based on the principle of back propagation, The second behavior model can be generated by calculating a weighted value obtained by digitizing the correlation.

That is, when the user's character is intensively monitored in a certain pattern of behavior pattern in the game, the second behavior model generation unit 130 identifies the operating element that is operating in the game at the time when the behavior pattern pattern is displayed , And learns the correlation between the elements of the behavior pattern and the operating elements that are intensively monitored.

For example, if a pattern of behavior patterns in which the number of users accessing a game suddenly increases within an event period compensating for game goods according to the number of attendances, a game pattern corresponding to a sudden increase in the number of users accessing the game, The compensating operating elements have a close correlation with each other and the second behavioral model can be learned in a way that gives a high weight to these correlations.

Examples of the machine learning method based on the back propagation principle performed by the second behavior model generation unit 130 include a decision tree, a random forest, an SVM, and a deep learning.

In this manner, the second behavior model generated by the second behavior model generation unit 130 is stored in the second storage unit 150.

Operation result prediction module 160

When the construction of the first and second behavioral models is completed, the operation result prediction module 160 uses the constructed first and second behavioral models to perform an operational result (profitability) through a two- Predict. This will be described in detail with reference to FIG.

3 is a diagram schematically illustrating a process of predicting operation results in the operation result prediction module shown in FIG.

Referring to FIG. 3, the operation result prediction module 160 includes a first step S31 of predicting a behavior pattern using a first behavior model in which a correlation between a behavior pattern and an operation element is learned, And a second step (S33) of predicting the behavior pattern using the second behavior model in which the correlation between the pattern and the behavior pattern corresponding to the result is learned.

Specifically, in the first step S31, in order to predict a game operation result (profitability) for the determined operating element before reflecting the newly determined operating element (or the updated operating element) in the game, And an operational element is input to the operational result predicting module 160.

Then, the operation result predicting module 160 predicts a behavior pattern for the new operating element using the second behavior model in which the correlation between the operating element and the behavior pattern is learned. That is, in the first step S31, not the profitability through the newly determined association of the operating elements is determined but the behavior pattern according to the newly determined operating elements is predicted. Direct profitability judgment is carried out in the second stage below.

In the second step S33, the operation result prediction module 160 sets a behavior pattern corresponding to the cause of the predicted behavior pattern, and determines a correlation between a behavior pattern corresponding to the cause and a behavior pattern corresponding to the result Predicts a behavior pattern corresponding to the result of the predicted behavior pattern, that is, a member withdrawal and an item purchase occurrence, using the learned first behavioral model.

Based on the predicted withdrawal of membership and the occurrence of item purchase, the administrator of the system 100 determines whether to reflect or supplement the newly determined operating element in the game.

As described above, the present invention requires a specific profitability prediction method for commercial service operation. For this purpose, a new prediction method that can analyze the correlation between the existing behavior prediction modeling method based on the machine learning methodology and the game service operation element It can be used as an important indicator in determining the game operation policy by not only predicting the game style of the user but also estimating the game style of the predicted user and the profitability generated thereby.

FIG. 4 is a flowchart illustrating a method for predicting an operation result of an online game service according to an embodiment of the present invention. For facilitating the understanding of the description, FIGS. 1 to 3 may be referred to. The contents and the duplicated description will be briefly described or omitted.

Referring to FIG. 4, in step S410, a process of extracting behavior pattern data from game log data is performed. Herein, the game log data includes information on the subject of the action, action information and additional information when the action occurs in the game use by the user as the use log information of the game as the user proceeds the game And may be recorded information. Specifically, the system log, the character log, and the NPC log (all actions that are not the subject but are not the system log) can be included according to the subject of the action. The behavior pattern data that can be extracted from the game log data includes user account name, character name, character class, character level, login information, game asset related information, game cost related information, personal information, login history information, Information on the reason why the user performs the action, information on the time, location coordinate information on the game where the action occurs, and information on the object of the action.

Then, in step S420, the first behavior model generation unit 120 performs a process of generating the first behavior model. Specifically, by using the machine learning method based on the back propagation principle, the correlation between the behavior pattern data corresponding to the cause and the behavior pattern data corresponding to the result is learned, and a weighted value obtained by quantifying the learned correlation To thereby generate the first behavioral model.

In step S430, the second behavior model generation unit 130 generates a second behavior model. Specifically, by using a machine learning method based on a back propagation principle, a correlation between behavior pattern data corresponding to a cause and a game operation element is learned, and a weighted value obtained by quantifying the learned correlation is calculated, A second behavioral model can be generated.

Examples of the machine learning method used in steps S420 and S430 include a decision tree, a random forest, an SVM, and a deep learning. The machine learning method for generating the first behavior model and the second behavior model The machine learning methods for generating may be identical to each other or may be different from each other.

Next, in step S440, a process of predicting the behavior pattern is performed. Specifically, a newly determined operational element is input, and a behavior pattern for the newly determined operational element is predicted using the second behavioral model in which the correlation between the behavioral pattern and the operational element is learned.

Next, in step S450, a process of predicting the operating result of the online game service, that is, the profitability of the online game service, is performed. Specifically, in step S440, the predicted behavior pattern is input, and the correlation between the behavior pattern corresponding to the cause and the behavior pattern corresponding to the result is calculated using the learned first behavior model A series of all processes for predicting the operation result of the online game service is terminated by performing a behavior pattern corresponding to the result of the on-line game service, that is, a membership withdrawal or an item purchase occurrence closely related to the profitability of the online game service .

While the present invention has been particularly shown and described with reference to exemplary embodiments thereof, it is to be understood that the invention is not limited to the disclosed exemplary embodiments, but, on the contrary, It will be understood that various modifications and applications not illustrated in the drawings are possible. For example, each component specifically shown in the embodiments of the present invention can be modified and implemented. It is to be understood that all changes and modifications that come within the meaning and range of equivalency of the claims are therefore intended to be embraced therein.

Claims (12)

In a method for predicting an operation result of an online game service provided by an online game providing system,
Extracting behavior pattern data of a user character from game log data;
Learning a first correlation between behavior pattern data corresponding to a cause and behavior pattern data corresponding to a result from the extracted behavior pattern data to generate a first behavior model in which the first correlation is learned;
Learning a second correlation between the extracted behavior pattern data and an operating element of the online game service to generate a second behavior model in which the second correlation is learned; And
Predicting an operation result of an online game service for a newly determined operating element using the first and second behavior models,
A method for predicting an operation result of an online game service including an online game service.
2. The method of claim 1, wherein the step of generating the first behavior model and the step of generating the second behavior model, respectively,
And learning the first and second correlations using a machine learning method, respectively.
The machine learning method according to claim 2,
Wherein the method is a learning method based on a principle of back propagation.
2. The method according to claim 1, wherein the behavior pattern data corresponding to the result includes:
And data related to the profitability of the online game service.
5. The method of claim 4, wherein the profitability-
And the data related to the withdrawal of the member and the purchase of the item.
2. The system according to claim 1, wherein the operating element of the online game service comprises:
A monster arrangement on the map, a feature value compensation rule according to the feature in the game, a mission in the game or a number of times of attendance, and a rules for allocating the ability value of the user character.
2. The method of claim 1, wherein predicting the operational result comprises:
Predicting a behavior pattern of a user character for the newly determined operational element using the second behavioral model; And
And estimating a behavior pattern corresponding to a result of the predicted behavior pattern using the first behavior model to predict an operation result related to the profitability of the online game service.
In an apparatus for predicting the operation result of an online game service,
A game operation server for extracting behavior pattern data of a user character from game log data;
Learning a first correlation between behavior pattern data corresponding to a cause and behavior pattern data corresponding to a result from among the extracted behavior pattern data input from the game operating server and comparing the first behavior model with the first behavior model A first behavior model generation unit for generating a first behavior model;
Generating a second behavior model in which the second correlation is learned by learning a second correlation between the extracted behavior pattern data input from the game operating server and an operating element of the online game service, part; And
An operation result predicting module for predicting an operation result of an online game service for a newly determined operating element by using the first and second behavior models,
Wherein the online game service comprises:
9. The information processing apparatus according to claim 8, wherein each of the first and second behavior-
Wherein the first and second correlations are respectively learned by a machine learning method according to a principle of back propagation.
9. The method according to claim 8, wherein the behavior pattern data corresponding to the result includes:
Data related to the profitability of the online game service,
The data relating to the profitability,
And data related to the withdrawal of the member and the purchase of the item.
The system according to claim 8,
Predicts a behavior pattern of a user character for the newly determined operating element by using the second behavior model and then predicts a behavior pattern corresponding to a result of the predicted behavior pattern using the first behavior model And predicts an operation result related to profitability of the online game service.
12. The method according to claim 11, wherein the behavior pattern corresponding to the predicted result includes:
And data related to the profitability related to the withdrawal of the member and the purchase of the item.

Images