Game prop processing method and device, electronic equipment and storage medium

Abstract

The game prop processing method and device, the electronic equipment and the storage medium provided by the application, in response to a recommendation request for a first game prop, determine user information and first game prop information associated with the recommendation request, and input the above information into a trained prediction model to predict the predicted number of second game props corresponding to each first game prop after the first game prop is put in. It is shown that the type of first game prop affects the number of second game props after the first game prop is put in. Through the number of second game props, the virtual value and the virtual value of the corresponding first game prop, the global resource consumption value of the game platform after the first game prop is put in can be calculated. The first game prop that maximizes the global resource consumption value is the target first game prop, thereby improving the overall resource consumption value of the game platform.

Description

Technical Field

[0001] This application relates to the field of data processing technology, and in particular to a method, apparatus, electronic device and storage medium for processing game props. Background Technology

[0002] In multiplayer games, to enhance the circulation of virtual resources and increase gameplay enjoyment, game operators often distribute gift packs at appropriate times. These gift packs typically include one or more in-game store items with relatively low resource consumption, encouraging players to spend resources in the game. Currently, when a player triggers the gift pack distribution mechanism, game operators usually distribute the gift pack with the highest exchange quantity or the highest resource consumption. However, gift pack distribution affects the exchange quantity of in-game store items. The above distribution method only considers maximizing the resource consumption value of the gift pack, without considering the decrease in the resource consumption value of the in-game store caused by the gift pack distribution, resulting in the overall resource consumption value of the game platform not reaching its maximum. Summary of the Invention

[0003] In view of this, the purpose of this application is to propose a method for handling game items in order to maximize the global resource consumption of the game platform.

[0004] The first aspect of this application provides a method for handling game items, including:

[0005] In response to a recommendation request for a first game item, user information and information of the first game item corresponding to the recommendation request are determined, wherein the first game item is a virtual game item to be deployed, and there are several types of the first game item;

[0006] Several types of information about the first game item and the user information are input into a trained prediction model. The prediction model outputs the predicted quantity of the second game item that affects the quantity of the second game item after each type of the first game item is deployed, where the second game item is a virtual game item that has already been deployed.

[0007] Based on the predicted placement quantity, calculate the resource consumption value of the second game item after each type of the first game item is placed;

[0008] Based on the resource consumption value of the second game item and the virtual value of the first game item associated with the second game item, a target first game item is determined, and the target first game item is recommended to the user.

[0009] A second aspect of this application also provides a game item processing device, comprising:

[0010] The determination module is configured to, in response to a recommendation request for a first game item, determine user information and first game item information corresponding to the recommendation request, wherein the first game item is a virtual game item to be deployed, and the first game item is of several types;

[0011] The prediction module is configured to input several types of first game item information and the user information into a trained prediction model, and output the predicted quantity of the second game item that affects the quantity of the second game item after each type of first game item is deployed, wherein the second game item is a deployed virtual game item.

[0012] The calculation module is configured to calculate the resource consumption value of the second game item after each type of the first game item is placed, based on the predicted placement quantity.

[0013] The recommendation module is configured to determine a target first game item based on the resource consumption value of the second game item and the virtual value of the first game item associated with the second game item, and recommend the target first game item to the user.

[0014] A third aspect of this application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable by the processor, wherein the processor implements the method described above when executing the computer program.

[0015] A fourth aspect of this application also provides a computer-readable storage medium storing computer instructions for causing a computer to perform the methods described above.

[0016] As can be seen from the above, the game item processing method, apparatus, electronic device, and storage medium provided in this application, in response to a recommendation request for a first game item, determine the user information and first game item information associated with the recommendation request, and input the above information into a trained prediction model to predict the predicted quantity of a second game item corresponding to each type of first game item after its deployment, reflecting the impact of the deployment of a first game item on the quantity of the second game item. By using the quantity of the second game item and the virtual value of the corresponding first game item, the global resource consumption value of the game platform after the deployment of the first game item can be calculated, and the first game item that maximizes the global resource consumption value is selected as the target first game item. The game item processing method of this application considers the impact of the deployment of the first game item on the quantity of the second game item, recommending the target first game item to the user with the goal of maximizing the global resource consumption value, thereby improving the overall resource consumption value of the game platform. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in this application or related technologies, the drawings used in the description of the embodiments or related technologies will be briefly introduced below. Obviously, the drawings described below are only embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is a flowchart illustrating the game item processing method according to an embodiment of this application;

[0019] Figure 2 This is a flowchart illustrating the method for obtaining the predicted placement quantity according to an embodiment of this application;

[0020] Figure 3 This is a flowchart illustrating the training method of the prediction model according to an embodiment of this application;

[0021] Figure 4 This is a schematic diagram of the structure of the game item processing device according to an embodiment of this application;

[0022] Figure 5 This is a schematic diagram of the hardware structure of an electronic device according to an embodiment of this application. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with specific embodiments and the accompanying drawings.

[0024] It should be noted that, unless otherwise defined, the technical or scientific terms used in the embodiments of this application should have the ordinary meaning understood by one of ordinary skill in the art to which this application pertains. The terms "first," "second," and similar terms used in the embodiments of this application do not indicate any order, quantity, or importance, but are merely used to distinguish different components. Terms such as "comprising" or "including" mean that the element or object preceding the word encompasses the elements or objects listed after the word and their equivalents, without excluding other elements or objects. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect. Terms such as "upper," "lower," "left," and "right" are only used to indicate relative positional relationships; when the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0025] As described in the background section, to increase players' virtual resource consumption, operators regularly launch a gift pack distribution function in the game, offering different gift pack types to different players. Typically, operators choose gift packs with the highest click-through rate or the highest resource consumption. Since these gift packs contain virtual game items with relatively low resource consumption, players who need them will prioritize purchasing them, thus affecting the availability of virtual game items in the game store and causing a decrease in their availability. The above gift pack distribution method only considers maximizing global gift pack resource consumption and does not account for the decrease in the availability of virtual items in the store. This could lead to a decrease in overall resource consumption due to increased gift pack resource consumption and decreased virtual item resource consumption in the store. To address this issue, this application proposes a game item processing method that considers both gift pack and store resource consumption, determining the gift pack type with the goal of maximizing global resource consumption.

[0026] The embodiments of this application will be described in detail below with reference to the accompanying drawings.

[0027] This application provides a method for handling game items, see reference. Figure 1 This includes the following steps:

[0028] Step 102: In response to the recommendation request for the first game item, determine the user information and the first game item information corresponding to the recommendation request, wherein the first game item is a virtual game item to be deployed, and there are several types of the first game item.

[0029] Specifically, when a player logs into the game, a first-item drop mechanism is triggered based on the player's virtual character level, login time, etc. At this time, a recommended first-item is to be dropped on the player. The player's client sends a first-item recommendation request to the server. After receiving the recommendation request, the server drops the recommended first-item. First, after receiving the first-item recommendation request, the server needs to determine the user information and first-item information corresponding to the request. User information may include the virtual character's average online time, the user's gender and age, and the user's in-game acquisition information, etc. Acquisition information can be the user's acquisition records over a period of time, such as 3 days or a month, etc., with no specific limitation. The first-item can be a virtual gift pack to be dropped in the game. Typically, the gift pack contains at least one virtual game item with relatively low resource consumption. The first-item information may include a non-repeating drop sequence of gift packs over a period of time, as well as the specific type and virtual value of the first-item. There are several types of game items. Different game items may contain different types and quantities of virtual game items, and their release timestamps differ, forming a release time sequence. The server needs to select the target game item from these types and recommend it to the player.

[0030] Step 104: Input several types of information on the first game item and the user information into the trained prediction model, and output the predicted quantity of the second game item that affects the quantity of the second game item after the first game item is deployed, wherein the second game item is a virtual game item that has been deployed.

[0031] To determine the target first game item, the server needs to use a prediction model. This model can predict the impact of the release of each first game item on the quantity of second game items already released in the game. Specifically, the second game item can be a virtual game item in the game store. Users can exchange resources for virtual game items in the game store, such as experience points, virtual coins, points, or magic points accumulated during gameplay, to improve the attribute values ​​of the user's virtual character. Taking a gift pack as an example, the release of a gift pack will inevitably reduce the quantity of similar virtual game items in the game store. The prediction model can predict the predicted quantity of virtual game items in the game store after the gift pack is released, thus providing basic data for the calculation of maximizing global resource consumption. During prediction, the first game item information obtained in step 102 and the user information are input into the trained prediction model, and the prediction model outputs the predicted quantity of the second game item. This prediction model supports simultaneous prediction of multiple gift packs, and the output data includes the predicted quantity of the second in-game item corresponding to each gift pack. For example, when the gift pack is a virtual skin gift pack, after the gift pack is distributed, the predicted quantity of virtual skins in the game store is obtained.

[0032] Step 106: Calculate the resource consumption value of placing the second game item after each type of the first game item is placed, based on the predicted placement quantity.

[0033] Step 108: Based on the resource consumption value of the second game item and the virtual value of the first game item associated with the second game item, determine the target first game item and recommend the target first game item to the user.

[0034] The resource consumption value of the second game item is calculated based on the quantity acquired. From the perspective of maximizing global resource consumption, a target first game item is determined based on the resource consumption value of the second game item and the virtual value of the associated first game item. Recommending the target first game item to the user can maximize the resource consumption value of the game platform, thereby improving the overall resource consumption value of the game platform.

[0035] Following steps 102 to 108, in response to a recommendation request for a first game item, user information and first game item information associated with the recommendation request are determined. This information is then input into a trained prediction model to predict the predicted quantity of a second game item that will be acquired after each type of first game item is deployed. This reflects the impact of deploying a first game item of a certain type on the quantity of the second game item acquired. By considering the quantity of the second game item acquired, its virtual value, and the corresponding virtual value of the first game item, the global resource consumption of the game platform after deploying the first game item can be calculated. The first game item that maximizes the global resource consumption is selected as the target first game item. This application's game item processing method considers the impact of deploying a first game item on the quantity of the second game item acquired, recommending the target first game item to the user with the goal of maximizing the global resource consumption, thereby improving the overall resource consumption of the game platform.

[0036] In some embodiments, the first game item information includes historical drop information of the first game item and attribute information of the first game item, with reference to... Figure 2 Step 104 includes the following steps:

[0037] Step 202: Input the historical delivery information and the user information into the trained prediction model, and output the first quantity of the second game item corresponding to each first game item through the prediction model.

[0038] Step 204: Input the attribute information of several types of the first game props, the historical deployment information, and the user information into the trained prediction model, and output the second quantity of the second game props corresponding to each type of the first game props through the prediction model.

[0039] Step 206: Take the difference between the first placement quantity and the second placement quantity as the predicted placement quantity.

[0040] Specifically, if the first game item is not distributed (i.e., no game item is recommended to players), players may acquire a certain number of second game items in the game store over a period of time. If game items are recommended to players, the number of second game items acquired by players in the game store will decrease or cease to be acquired in the future. In prediction, it is necessary to know the first acquisition quantity of the second game item under the condition of no intervention (i.e., no distribution of the first game item), and the second acquisition quantity of the second game item under the condition of intervention (i.e., distribution of the first game item). The difference between the first and second acquisition quantities represents the impact of distributing the first game item on the acquisition quantity of the second game item, and this difference is used as the predicted acquisition quantity. When predicting the first acquisition quantity, the historical distribution information and the user information are input into a trained prediction model, and the output data of the prediction model is the first acquisition quantity. When predicting the second placement quantity, the attribute information of several types of the first game props, the historical placement information, and the user information are input into the trained prediction model, and the output data of the prediction model is the second placement quantity.

[0041] In some embodiments, the first game item information includes historical drop information and attribute information of the first game item, and the user information includes user attribute information and user historical placement information. Before step 104, the following is included:

[0042] The historical delivery information and the user's historical placement information are sorted according to timestamps, and the user attribute information is encoded using an encoding algorithm to obtain an embedding vector.

[0043] Specifically, before inputting historical delivery information, user historical placement information, and user attribute information into the prediction model, the above information needs to be preprocessed. Both historical delivery information and user historical placement information are time-related; each historical delivery or placement is recorded and accompanied by a timestamp, which identifies the time node information of that delivery or placement. Sorting the historical delivery information and user historical placement information according to the timestamps to generate corresponding historical delivery sequences and user historical placement sequences allows the prediction model to fully learn the impact of time information on user placement behavior during prediction, thereby improving the prediction accuracy of the prediction model. User attribute information, also known as user profile features, is used to identify multi-dimensional data of user attributes, behaviors, and expectations. User attribute information includes the average online time of the virtual character controlled by the user, the placement level of the virtual character, the user's gender information, and age information, etc. User attribute information is unstructured information and needs to be encoded to obtain embedding vectors. Embedding vectors are multi-dimensional vectors that represent user attribute information.

[0044] In this embodiment, the encoding algorithm can be either one-hot encoding or a normalization algorithm, for example. One-hot encoding, also known as one-bit normalization, uses an N-bit state register to encode N states, each state having its own independent register bit, and at any given time, only one bit is valid. Normalization is a method to simplify computation by transforming dimensional expressions into dimensionless expressions, i.e., into scalar expressions.

[0045] In some embodiments, prior to step 104, the method further includes:

[0046] The first game item information and the user information are used as request parameters, and the prediction model is called according to the request parameters.

[0047] In some embodiments, it also includes:

[0048] The trained prediction model is then converted to a new format; the converted prediction model is then deployed online.

[0049] Specifically, the trained prediction model undergoes a format conversion. For example, the trained prediction model can be permanently stored as a .pb (protocol buffer) file, which is a binary file representing the model structure. The format-converted prediction model is then deployed as an online inference service, capable of responding promptly to client prediction requests. When using the prediction model for prediction, it needs to be invoked; the request parameters include several types of information about the first game item and the user information.

[0050] In some embodiments, step 106 includes: calculating the product of the predicted quantity and the virtual value of the second game item to obtain the resource consumption value of the second game item.

[0051] After obtaining the predicted quantity, the predicted quantity is multiplied by the virtual value of the second game item, and the product is used as the resource consumption value of the second game item.

[0052] In some embodiments, the quantity of the second game item is at least one, and step 108 includes:

[0053] The first game item corresponding to the maximum sum of the resource consumption values ​​of all second game items and the virtual value of the associated first game item is selected as the recommended game item.

[0054] When multiple second game items exist, it indicates that the placement of the first game item affected the quantity of multiple second game items obtained. The resource consumption values ​​of all second game items are calculated. The quantity of each second game item is multiplied by its virtual value, and the sum is then added to the virtual value of the associated first game item. The first game item corresponding to the maximum sum is selected as the target first game item.

[0055] Alternatively, calculate the resource consumption value of all second game items, multiply the quantity of each second game item by its virtual value, sum the results, weight the sum with the virtual value of the associated first game item, and select the first game item corresponding to the maximum sum as the target first game item. The weights can be set according to actual needs, which will not be elaborated on here.

[0056] In some embodiments, step 102 includes:

[0057] In response to a recommendation request for a first game item, a user identifier associated with the recommendation request is determined; and the user information associated with the user identifier and the information of the first game item are queried in the database based on the user identifier.

[0058] In some embodiments, the user information and several types of the first game item information within a preset time interval are obtained at preset time intervals, stored in the database, and associated timestamps are generated.

[0059] Specifically, the recommendation request includes a user identifier. For example, the user identifier can be the character ID of the virtual character operated by the user. The user information and the first game item information are queried in the database using the character ID as an index. Since user information and the first game item information change over time—for example, the average online time of the virtual character in the user information, the types and quantities of items the user acquires in the game will change—the database needs to periodically retrieve and store user information and the first game item information. When a query is made in the database based on the user identifier, the database provides the client with the most recently updated user information and the first game item information so that the prediction model can predict the quantity of the second game item acquired based on the latest updated data, thereby improving the real-time accuracy of the model's predictions.

[0060] It should be noted that the preset time interval can be several hours or several days, and the preset time range can be a countdown of several days, including the current day. For example, the database can retrieve data from the game platform every few hours or days, and the time range for retrieval is all historical data from the previous 31 days, including the current day. After successfully retrieving historical data, a corresponding timestamp is generated to identify the time information of the currently retrieved historical data.

[0061] In some embodiments, the first game item information includes historical drop information and attribute information of the first game item, and the user information includes user attribute information and user historical placement information, with reference to... Figure 3 The training method for the prediction model includes the following steps:

[0062] Step 302: Select the historical delivery information corresponding to the current timestamp, the attribute information of several types of the first game item, the user attribute information, and the user historical placement information from the database, wherein the current timestamp is updated in real time.

[0063] Specifically, because users' purchasing preferences change over time, and the information on the distribution of first-time game items is constantly updated, as game operators continuously develop new first-time game items, the types of first-time game items are also constantly updated with the development of the game. Therefore, using a static prediction model to predict users' purchasing behavior is inaccurate. The prediction model in this embodiment is trained in real time. Each training session uses historical data corresponding to the current timestamp to train the prediction model, and the current timestamp is updated in real time.

[0064] It should be noted that when retrieving historical data used to train the predictive model, the database needs to filter players based on certain criteria. For example, it selects historical data of players who have placed items in the game within the 31 days prior to the current date, and whose overall ranking on the server is within the top 6000 on that day. This ensures that the player is in a normal gaming state, while excluding players with abnormal gaming status or no recent placed items, thus eliminating the impact on the predictive model training.

[0065] Step 304: Divide the historical delivery information into historical delivery received information and historical delivery undelivered information. Each time the first game item is delivered, if the user has a need to receive it, they will receive the first game item; if the user does not have a need to receive it, they will not receive the first game item. Therefore, the historical delivery information can be divided into historical delivery received information and historical delivery undelivered information. The historical delivery received information contains delivery sequences for the first game item corresponding to the player's delivery action, while the historical delivery undelivered information contains delivery sequences for the first game item corresponding to the player's non-delivered action.

[0066] Step 306: Use the user's historical placement information corresponding to the historical placement information and the historical non-placement information as the first tag and the second tag, respectively.

[0067] User history acquisition information includes the quantity of the second game item acquired by the user after the first game item was acquired. If the user acquired the first game item after its acquisition, the quantity of the second game item acquired within a certain period of time will be recorded as the first tag. If the user did not acquire the first game item after its acquisition, the quantity of the second game item acquired within a certain period of time will be recorded as the second tag.

[0068] Step 308: Construct a training sample set based on the user attribute information corresponding to the current timestamp, the historical delivery information, the user historical placement information, the attribute information of several first game items, the first tag, and the second tag.

[0069] When there is no first game item being placed, the input data of the prediction model in the training samples are the user attribute information, the historical placement information, and the user historical placement information, and the output data is the first label.

[0070] When the first game item is deployed, the input data of the corresponding prediction model in the training samples are the user attribute information, the historical deployment information, the user historical placement information, and several attribute information of the first game item, and the output data is the second label.

[0071] Step 310: Iteratively train the prediction model based on the training sample set. The prediction model is iteratively trained using the training samples obtained in step 308.

[0072] It should be noted that multiple prediction models can be used in this embodiment. Each prediction model corresponds to predicting the quantity of second game items acquired within a certain time range. These prediction models can be categorized into short-term, medium-term, and long-term prediction models. For example, a short-term prediction model can predict the quantity of second game items acquired within two hours, a medium-term prediction model can predict the quantity within three days, and a long-term prediction model can predict the quantity within one week. Game operators can select and train the appropriate prediction model based on actual prediction needs to complete the prediction of the quantity of second game items acquired. Correspondingly, training samples within different time ranges can be constructed using a sliding window approach to train the prediction model. For instance, the first and second labels of the training samples for the short-term prediction model are the historical acquisition quantities of users within two hours. The construction methods for the training samples of the medium-term and long-term prediction models are similar and will not be elaborated here.

[0073] Step 312: In response to determining that the training cutoff condition has been met, complete the training of the prediction model.

[0074] In this embodiment, the training cutoff condition is to minimize the loss between the model output data and the true label. The loss function is the square of the difference between the actual number of second game items obtained and the predicted number of items obtained by the model. Minimizing the loss function adjusts the parameters of the prediction model, thus completing the training of the prediction model. It should be noted that other loss functions can be selected as the objective optimization function according to actual training needs to optimize and adjust the parameters of the prediction model. The loss function in this embodiment is only illustrative and has no limiting effect.

[0075] In some embodiments, the prediction model is a deep learning model based on the S-learner algorithm in meta-learning. The principle of this algorithm is to train the prediction model based on variables and intervention items, and then estimate the scores under the intervention and non-intervention conditions respectively, with the difference being the increment. The S-learner algorithm can be applied to existing models, and when making predictions, only one model is needed to complete the prediction, avoiding the error accumulation problem caused by multiple model predictions.

[0076] The deep learning model in this embodiment includes two ReLU nonlinear transformation layers and one linear transformation layer. The model outputs a multi-dimensional vector, where the dimension of the vector represents the number of the second game item predicted by the user. For example, in this embodiment, the vector dimension is 500, so the number of the second game item is 500. The activation function of the ReLU nonlinear transformation layer is y = max(wx + b, 0), and the activation function of the linear transformation layer is y = wx + b, where w is the slope and b is the intercept.

[0077] To verify the prediction results of the prediction model and the global resource consumption, the following specific examples are used for comparison and explanation.

[0078] The game data predicted for the next 48 hours was used for verification. Two sets of experiments were set up: a baseline group and an experimental group. The baseline group used the traditional method to distribute game items, while the experimental group used a prediction model to predict and then distribute game items. All other experimental configurations were the same for the baseline group and the experimental group. The experimental results are shown in the table below.

[0079] Table 1. Experimental results of the baseline group and the experimental group.

[0080] benchmark group 100％ 100％ 100％ experimental group 78％ 79％ 105％

[0081] As can be seen from Table 1, the baseline group has higher resource consumption per person and total resource consumption per person than the experimental group. However, the experimental group has higher resource consumption than the baseline group when considering the sum of resource consumption of gift packs and the store, with an increase of about 5%. This shows that compared with the traditional distribution method, the recommended game items obtained by prediction model can maximize the overall resource consumption of the game platform.

[0082] It should be noted that the method in this embodiment can be executed by a single device, such as a computer or server. The method can also be applied in a distributed scenario, where multiple devices cooperate to complete the task. In such a distributed scenario, one of these devices may execute only one or more steps of the method in this embodiment, and the multiple devices will interact with each other to complete the method described.

[0083] It should be noted that the above description describes some embodiments of this application. Other embodiments are within the scope of the appended claims. In some cases, the actions or steps recorded in the claims can be performed in a different order than that shown in the above embodiments and still achieve the desired result. Furthermore, the processes depicted in the drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0084] This application also provides a device for processing game props.

[0085] refer to Figure 4 The game item processing device includes:

[0086] The determining module 402 is configured to, in response to a recommendation request for a first game item, determine user information and first game item information corresponding to the recommendation request, wherein the first game item is a virtual game item to be deployed, and the first game item is of several types;

[0087] The prediction module 404 is configured to input several types of first game item information and the user information into a trained prediction model, and output the predicted quantity of the second game item that affects the quantity of the second game item after each type of first game item is deployed, wherein the second game item is a deployed virtual game item.

[0088] The calculation module 406 is configured to calculate the resource consumption value of the second game item after each type of the first game item is placed, based on the predicted placement quantity.

[0089] The recommendation module 408 is configured to determine a target first game item based on the resource consumption value of the second game item and the virtual value of the first game item associated with the second game item, and recommend the target first game item to the user.

[0090] In some embodiments, the first game item information includes historical distribution information of the first game item and attribute information of the first game item. The prediction module 404 is further configured to input the historical distribution information and the user information into a trained prediction model, and output the first quantity of the second game item corresponding to each first game item through the prediction model.

[0091] The attribute information of several types of the first game item, the historical deployment information, and the user information are input into a trained prediction model, and the prediction model outputs the second quantity of the second game item corresponding to each type of the first game item.

[0092] The difference between the first set quantity and the second set quantity is taken as the predicted set quantity.

[0093] In some embodiments, the first game item information includes historical delivery information and attribute information of the first game item, the user information includes user attribute information and user historical placement information, and further includes a preprocessing module 410 configured to sort the historical delivery information and the user historical placement information according to timestamps, and to encode the user attribute information using an encoding algorithm to obtain an embedding vector.

[0094] In some embodiments, a calling module 412 is further included, configured to use several types of the first game item information and the user information as request parameters, and to call the prediction model according to the request parameters.

[0095] In some embodiments, the calculation module 406 is further configured to calculate the product of the predicted quantity and the virtual value of the second game item to obtain the resource consumption value of the second game item.

[0096] In some embodiments, the number of the second game items is at least one, and the recommendation module 408 is further configured to use the first game item corresponding to the maximum value of the sum of the resource consumption values ​​of all the second game items and the virtual value of the associated first game items as the target first game item.

[0097] In some embodiments, the determining module 402 is further configured to, in response to a recommendation request for a first game item, determine a user identifier associated with the recommendation request; and query the database for the user information associated with the user identifier and the information of the first game item based on the user identifier.

[0098] In some embodiments, a storage module 414 is further included to acquire the user information and several types of the first game item information within a preset time interval at preset time intervals, store them in the database, and generate associated timestamps.

[0099] In some embodiments, the first game item information includes historical deployment information and attribute information of the first game item, the user information includes user attribute information and user historical placement information, and further includes a training module 416 configured to select from the database the historical deployment information corresponding to the current timestamp, several types of attribute information of the first game item, the user attribute information and the user historical placement information, wherein the current timestamp is updated in real time;

[0100] The historical delivery information is divided into historical delivery information that has been set and historical delivery information that has not been set.

[0101] The user's historical placement information corresponding to the historical placement information and the historical non-placement information are respectively used as the first tag and the second tag;

[0102] A training sample set is constructed based on the user attribute information corresponding to the current timestamp, the historical delivery information, the user historical placement information, the attribute information of several types of the first game items, the first tag, and the second tag;

[0103] The prediction model is iteratively trained based on the training sample set;

[0104] In response to determining that the training cutoff condition has been met, the training of the prediction model is completed.

[0105] In some embodiments, a deployment module 416 is also included, configured to perform format conversion on the trained prediction model and deploy the format-converted prediction model online.

[0106] For ease of description, the above devices are described in terms of function, divided into various modules. Of course, in implementing this application, the functions of each module can be implemented in one or more software and / or hardware.

[0107] The apparatus described above is used to implement the corresponding game item processing method in any of the foregoing embodiments, and has the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0108] This application also provides an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the game item processing method described in any of the above embodiments.

[0109] Figure 5 This embodiment illustrates a more specific hardware structure of an electronic device, which may include a processor 1010, a memory 1020, an input / output interface 1030, a communication interface 1040, and a bus 1050. The processor 1010, memory 1020, input / output interface 1030, and communication interface 1040 are interconnected internally via the bus 1050.

[0110] The processor 1010 can be implemented using a general-purpose CPU (Central Processing Unit), microprocessor, application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of this specification.

[0111] The memory 1020 can be implemented in the form of ROM (Read Only Memory), RAM (Random Access Memory), static storage device, dynamic storage device, etc. The memory 1020 can store the operating system and other applications. When the technical solutions provided in the embodiments of this specification are implemented by software or firmware, the relevant program code is stored in the memory 1020 and is called and executed by the processor 1010.

[0112] The input / output interface 1030 is used to connect input / output modules to realize information input and output. Input / output modules can be configured as components within the device (not shown in the figure) or externally connected to the device to provide corresponding functions. Input devices may include keyboards, mice, touchscreens, microphones, various sensors, etc., while output devices may include displays, speakers, vibrators, indicator lights, etc.

[0113] The communication interface 1040 is used to connect a communication module (not shown in the figure) to enable communication between this device and other devices. The communication module can communicate via wired means (such as USB, Ethernet cable, etc.) or wireless means (such as mobile network, WIFI, Bluetooth, etc.).

[0114] Bus 1050 includes a pathway for transmitting information between various components of the device, such as processor 1010, memory 1020, input / output interface 1030, and communication interface 1040.

[0115] It should be noted that although the above-described device only shows the processor 1010, memory 1020, input / output interface 1030, communication interface 1040, and bus 1050, in specific implementations, the device may also include other components necessary for normal operation. Furthermore, those skilled in the art will understand that the above-described device may only include the components necessary for implementing the embodiments of this specification, and not necessarily all the components shown in the figures.

[0116] The electronic devices described above are used to implement the corresponding game item processing methods in any of the foregoing embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0117] This application also provides a computer-readable storage medium storing computer instructions for causing the computer to perform the game item processing method as described in any of the above embodiments.

[0118] The computer-readable medium of this embodiment includes permanent and non-permanent, removable and non-removable media, and information storage can be implemented by any method or technology. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, magnetic magnetic disk storage or other magnetic storage devices, or any other non-transfer medium that can be used to store information accessible by a computing device.

[0119] The computer instructions stored in the storage medium of the above embodiments are used to cause the computer to execute the game item processing method as described in any of the above embodiments, and have the beneficial effects of the corresponding method embodiments, which will not be repeated here.

[0120] Those skilled in the art should understand that the discussion of any of the above embodiments is merely exemplary and is not intended to imply that the scope of this application (including the claims) is limited to these examples; within the framework of this application, the technical features of the above embodiments or different embodiments can also be combined, the steps can be implemented in any order, and there are many other variations of different aspects of the embodiments of this application as described above, which are not provided in the details for the sake of brevity.

[0121] Additionally, to simplify the description and discussion, and to avoid obscuring the embodiments of this application, the well-known power / ground connections to integrated circuit (IC) chips and other components may or may not be shown in the provided drawings. Furthermore, the apparatus may be shown in block diagram form to avoid obscuring the embodiments of this application, and this also takes into account the fact that the details of the implementation of these block diagram apparatuses are highly dependent on the platform on which the embodiments of this application will be implemented (i.e., these details should be fully understood by those skilled in the art). While specific details (e.g., circuits) have been set forth to describe exemplary embodiments of this application, it will be apparent to those skilled in the art that the embodiments of this application can be implemented without these specific details or with variations thereof. Therefore, these descriptions should be considered illustrative rather than restrictive.

[0122] Although this application has been described in conjunction with specific embodiments thereof, many substitutions, modifications, and variations of these embodiments will be apparent to those skilled in the art from the foregoing description. For example, other memory architectures (e.g., dynamic RAM (DRAM)) may be used with the embodiments discussed.

[0123] The embodiments of this application are intended to cover all such substitutions, modifications, and variations that fall within the broad scope of the appended claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the embodiments of this application should be included within the protection scope of this application.

Claims

1. A game prop processing method, characterized in that, The method comprises: determining user information corresponding to a recommendation request and first game prop information in response to the recommendation request, wherein the first game prop is a virtual game prop to be put in, and the first game prop is a plurality of types; inputting the plurality of first game prop information and the user information into a trained prediction model, and outputting, by the prediction model, a predicted number of draws of a second game prop that is affected by the number of draws of the second game prop after each of the first game props is put in, wherein the second game prop is a virtual game prop that has been put in; the predicted number of draws is the difference between a first number of draws and a second number of draws, the first number of draws is the number of draws of the second game prop when the first game prop is not put in, and the second number of draws is the number of draws of the second game prop when the first game prop is put in; calculating a resource consumption value for drawing the second game prop after each of the first game props is put in according to the predicted number of draws; calculating a global resource consumption value of a game platform after the first game prop is put in according to the resource consumption value of the second game prop and the virtual value of the first game prop associated with the second game prop, taking the first game prop that maximizes the global resource consumption value as a target first game prop, and recommending the target first game prop to the user.

2. The method of claim 1, wherein, The first game prop information includes historical putting information of the first game prop and attribute information of the first game prop, The inputting the plurality of first game prop information and the user information into the trained prediction model and outputting, by the prediction model, the predicted number of draws of the second game prop that is affected by the number of draws of the second game prop after each of the first game props is put in comprises: inputting the historical putting information and the user information into the trained prediction model, and outputting, by the prediction model, a first number of draws of the second game prop corresponding to each of the first game props; inputting the attribute information of the plurality of first game props, the historical putting information, and the user information into the trained prediction model, and outputting, by the prediction model, a second number of draws of the second game prop corresponding to each of the first game props; taking the difference between the first number of draws and the second number of draws as the predicted number of draws.

3. The method of claim 1, wherein, The first game prop information includes historical putting information of the first game prop and attribute information of the first game prop, and the user information includes user attribute information and user historical draw information, Before inputting the plurality of first game prop information and the user information into the trained prediction model, the method comprises: sorting the historical putting information and the user historical draw information according to timestamps respectively, and encoding the user attribute information to obtain an embedding vector by using an encoding algorithm.

4. The method of claim 1, wherein, Before inputting the plurality of first game prop information and the user information into the trained prediction model, the method further comprises: The first game prop information and the user information are used as request parameters to call the prediction model according to the request parameters.

5. The method of claim 1, wherein, The resource consumption value of the second game prop after the first game prop is put in is calculated according to the predicted number of bets, and the resource consumption value of the second game prop is calculated. The product of the predicted number of bets and the virtual value of the second game prop is calculated to obtain the resource consumption value of the second game prop.

6. The method of claim 1, wherein, The number of the second game prop is at least one, The target first game prop is determined according to the resource consumption value of the second game prop and the virtual value of the first game prop associated with the second game prop, and the target first game prop is determined. The first game prop corresponding to the maximum value of the sum of the resource consumption value of all second game props and the virtual value of the associated first game prop is used as the target first game prop.

7. The method of claim 1, wherein, The user information and the first game prop information corresponding to the recommendation request for the first game prop are determined in response to the recommendation request for the first game prop, and the first game prop is a virtual game prop to be put in. The user information and the first game prop information corresponding to the recommendation request for the first game prop are determined in response to the recommendation request for the first game prop. The user information and the first game prop information corresponding to the recommendation request for the first game prop are determined in response to the recommendation request for the first game prop.

8. The method of claim 7, wherein, Further comprising: The user information and the first game prop information in a preset time interval are obtained at a preset time interval, stored in the database and a corresponding time stamp is generated.

9. The method of claim 8, wherein, The first game prop information includes historical putting information of the first game prop and attribute information of the first game prop, and the user information includes user attribute information and user historical bet information, The training method of the prediction model comprises: The historical putting information corresponding to the current time stamp, the attribute information of the first game prop, the user attribute information and the user historical bet information are selected from the database, wherein the current time stamp is updated in real time; The historical putting information is divided into historical putting bet information and historical putting non-bet information; The user historical bet information corresponding to the historical putting bet information and the historical putting non-bet information is used as a first label and a second label, respectively; A training sample set is constructed according to the user attribute information corresponding to the current time stamp, the historical putting information, the user historical bet information, the attribute information of the first game prop, the first label and the second label; The prediction model is iteratively trained based on the training sample set; The training of the prediction model is completed in response to the determination that the training cutoff condition is reached.

10. The method of claim 9, wherein, Further comprising: The trained prediction model is format-converted; The format-converted prediction model is deployed online.

11. A game prop processing device, characterized by, The determination module is configured to determine the user information and the first game prop information corresponding to the recommendation request for the first game prop in response to the recommendation request for the first game prop, wherein the first game prop is a virtual game prop to be put in, and the first game prop is of several types. ​ The prediction module is configured to input the first game prop information and the user information into a trained prediction model, and output, by the prediction model, a predicted number of placements of the second game prop that is affected by each of the first game props after the first game props are placed, wherein the second game prop is a virtual game prop that has been placed; the predicted number of placements is a difference between a first number of placements and a second number of placements, the first number of placements is a number of placements of the second game prop when the first game prop is not placed, and the second number of placements is a number of placements of the second game prop when the first game prop is placed; The calculation module is configured to calculate, according to the predicted number of placements, a resource consumption value of placing the second game prop after each of the first game props is placed; The recommendation module is configured to calculate, according to the resource consumption value of the second game prop and a virtual value of the first game prop associated with the second game prop, a global resource consumption value of a game platform after the first game prop is placed, take the first game prop that maximizes the global resource consumption value as a target first game prop, and recommend the target first game prop to the user.

12. An electronic device comprising a memory, a processor, and a computer program stored on the memory and executable on the processor, characterized in that, The processor implements the method in any one of claims 1 to 10 when executing the program.

13. A non-transitory computer-readable storage medium storing computer instructions, wherein, The computer instructions are used to make the computer execute the method in any one of claims 1 to 10.

Images