Game character equipment recommendation method based on AI and storage medium

Abstract

The invention discloses a game character equipment recommendation method based on AI, and the method comprises the steps: obtaining client data, and carrying out the leveling and classification labeling of the client data; the client data comprises game scene information and skin and equipment information of roles and characters; utilizing an AI model to generate index values of the skin and the equipment in the current game scene according to the client data and the classification label information; recommending the skin and the equipment according to the index values of the skin and the equipment; according to the method, the client information including the game scene information and the skin and equipment information is acquired, the client information is leveled and classified, the pre-trained AI model is utilized to output the index values of the skin and the equipment in the current game scene according to the classification label information, and the equipment is recommended according to the index values. The adaptive equipment and skin recommendation for the player according to different game scenes is realized, and the convenience of selecting the skin equipment of the player is improved.

Description

Technical Field

[0001] The present invention relates to the technical field, and in particular to an AI-based game character equipment recommendation method and storage medium. Background Art

[0002] Nowadays, all the main characters in games have a lot of appearance resources, which are divided into equipment and fashion (also known as skin). When making appearance resources, artists often split them into multiple wholes and parts to facilitate subsequent modification and adjustment of the same style.

[0003] With the increase in equipment and skins, and the changes in game scenes, the selection of equipment and skins in different scenes has become more complicated for players and requires more extra effort.

[0004] Therefore, how to recommend suitable equipment and skins to players according to the changes in scenes and reduce unnecessary waste of players' energy is a technical problem that needs to be solved at present. Summary of the invention

[0005] The technical problem to be solved by the present invention is to provide an AI-based game character equipment recommendation method and storage medium, which can recommend suitable equipment and skins for players according to different game scenes.

[0006] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0007] A method for recommending equipment for game characters based on AI, comprising the steps of:

[0008] S1. Obtain client data, flatten and classify the client data;

[0009] The client data includes game scene information and character skin and equipment information;

[0010] S2. Generate index values ​​of skins and equipment in the current game scenario using an AI model based on the client data and classification label information;

[0011] S3. Recommend skins and equipment based on the indicator values ​​of the skins and equipment.

[0012] In order to solve the above technical problems, another technical solution adopted by the present invention is:

[0013] A storage medium having a computer program stored thereon, wherein when the computer program is executed, the following steps are implemented:

[0014] S1. Obtain client data, flatten and classify the client data;

[0015] The client data includes game scene information and character skin and equipment information;

[0016] S2. Generate index values ​​of skins and equipment in the current game scenario using an AI model based on the client data and classification label information;

[0017] S3. Recommend skins and equipment based on the indicator values ​​of the skins and equipment.

[0018] The beneficial effects of the present invention are: an AI-based game character equipment recommendation method and storage medium of the present invention obtain client information including game scene information and skin and equipment information, and flatten and classify the client information, use a pre-trained AI model to output the index value of the skin and equipment in the current game scene according to the classification label information, and recommend equipment based on the index value, so as to achieve the recommendation of equipment and skin adapted for players according to different game scenes, and improve the convenience of players' skin and equipment selection. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] Figure 1 A flowchart of a method for recommending equipment for game characters based on AI according to an embodiment of the present invention;

[0020] Figure 2 This is a flowchart of an AI-based game character equipment recommendation method according to the first embodiment of the present invention;

[0021] Figure 3 This is an example flowchart of an AI-based game character equipment recommendation method according to Embodiment 2 of the present invention. DETAILED DESCRIPTION

[0022] In order to explain the technical content, achieved objectives and effects of the present invention in detail, the following is an explanation in conjunction with the implementation modes and the accompanying drawings.

[0023] Please refer to Figure 1 and Figure 3 , a method for recommending game character equipment based on AI, comprising the steps of:

[0024] S1. Obtain client data, flatten and classify the client data;

[0025] The client data includes game scene information and character skin and equipment information;

[0026] S2. Generate index values ​​of skins and equipment in the current game scenario using an AI model based on the client data and classification label information;

[0027] S3. Recommend skins and equipment based on the indicator values ​​of the skins and equipment.

[0028] From the above description, it can be seen that the beneficial effects of the present invention are: the present invention provides an AI-based game character equipment recommendation method and storage medium, which obtains client information including game scene information and skin and equipment information, and flattens and classifies the client information, and uses a pre-trained AI model to output the index values ​​of skins and equipment in the current game scene according to the classification label information, and recommends equipment based on the index values, thereby realizing the recommendation of equipment and skins adapted to players according to different game scenes, and improving the convenience of players' skin and equipment selection.

[0029] Furthermore, the client data includes game scene information, historical behavior information of characters, and skin and equipment information;

[0030] The game scene information includes big world type information, battlefield type information, opponent information, interface configuration information, and skill configuration information;

[0031] The skin and equipment information includes item configuration information, clothing configuration information, weapon configuration information, light effect configuration information, fantasy beast configuration information and mount configuration information.

[0032] From the above description, it can be seen that the recommendation of skins and equipment needs to take into account the type information of the big world. For example, volcanoes need to consider fire resistance, ice fields consider cold resistance, etc. According to the different type information of the big world, equipment with different attributes should be considered; the battlefield type needs to be considered. For example, the battlefield types of PVP and PVE often require the selection of different equipment; the opponent information needs to be considered, such as the "professional" characteristics of the opponent in PVP, the attributes of the "monster" in PVE, etc.; the configuration information needs to be considered, such as the attribute information emphasized by different skill combinations, etc.

[0033] Furthermore, the client data also includes version activity information, holiday activity information, task status information and role attribute status information.

[0034] From the above description, we can see that the recommendation of equipment also needs to consider information such as version activities and holiday activities. For example, for Halloween activities, fashion information with the classification label of ghost type is given priority. At the same time, it is necessary to consider whether the character attributes meet the equipment requirements, task status, etc.

[0035] Furthermore, step S1 further includes the steps of:

[0036] S11, preprocessing the acquired client data:

[0037] Data cleaning: remove invalid and abnormal data, and fill in missing values;

[0038] Data transformation: discretize and normalize numerical features according to data categories and encode classification information.

[0039] From the above description, we can see that before inputting into the AI ​​model, the data needs to be cleaned and converted to remove invalid and abnormal data and fill in missing values ​​to improve the effectiveness of the recommendation.

[0040] Furthermore, step S3 is specifically as follows:

[0041] According to the index values ​​of the skin and equipment, combined with the similarity information of the item equipment description, weight calculation is performed to obtain a recommended weight value;

[0042] Based on the recommendation weight value, the equipment is sorted according to its position, and skins and equipment are recommended according to the sorting result.

[0043] From the above description, it can be seen that the weight calculation is performed based on the indicator value output by the AI ​​model and the similarity information of the equipment description, and the ranking recommendation is performed based on the recommended weight value finally calculated.

[0044] Furthermore, the method further comprises the steps of:

[0045] S4, determining whether the player accepts to use the recommended skins and equipment, if yes, storing the recommended data, otherwise changing the recommended content, recommending again, and recording the number of recommendations;

[0046] If the number of recommendations reaches the preset threshold, the next time the player does not accept the recommended skins and equipment, the ranking results will be displayed for the player to choose.

[0047] From the above description, it can be seen that the data of whether the player accepts the recommendation plan is recorded and feedback is given, and the recommendation record is recorded. The recommendation record includes the client data at the time of recommendation, the recommended fashion or skin information, and whether the user accepts it, etc. It is used to optimize the AI ​​model and can be used as the historical behavior information of the character for subsequent recommendations.

[0048] A storage medium having a computer program stored thereon, wherein when the computer program is executed, the following steps are implemented:

[0049] S1. Obtain client data, flatten and classify the client data;

[0050] The client data includes game scene information and character skin and equipment information;

[0051] S2. Generate index values ​​of skins and equipment in the current game scenario using an AI model based on the client data and classification label information;

[0052] S3. Recommend skins and equipment based on the indicator values ​​of the skins and equipment.

[0053] From the above description, it can be seen that the beneficial effects of the present invention are: the present invention provides an AI-based game character equipment recommendation method and storage medium, which obtains client information including game scene information and skin and equipment information, and flattens and classifies the client information, and uses a pre-trained AI model to output the index values ​​of skins and equipment in the current game scene according to the classification label information, and recommends equipment based on the index values, thereby realizing the recommendation of equipment and skins adapted to players according to different game scenes, and improving the convenience of players' skin and equipment selection.

[0054] Furthermore, the client data includes game scene information, historical behavior information of characters, and skin and equipment information;

[0055] The game scene information includes big world type information, battlefield type information, opponent information, interface configuration information, and skill configuration information;

[0056] The skin and equipment information includes item configuration information, clothing configuration information, weapon configuration information, light effect configuration information, fantasy beast configuration information and mount configuration information.

[0057] From the above description, it can be seen that the recommendation of skins and equipment needs to take into account the type information of the big world. For example, volcanoes need to consider fire resistance, ice fields consider cold resistance, etc. According to the different type information of the big world, equipment with different attributes should be considered; the battlefield type needs to be considered. For example, the battlefield types of PVP and PVE often require the selection of different equipment; the opponent information needs to be considered, such as the "professional" characteristics of the opponent in PVP, the attributes of the "monster" in PVE, etc.; the configuration information needs to be considered, such as the attribute information emphasized by different skill combinations, etc.

[0058] Furthermore, the client data also includes version activity information, holiday activity information, task status information and role attribute status information.

[0059] From the above description, it can be seen that the recommendation of equipment also needs to consider information such as version activities and holiday activities. For example, for Halloween activities, fashion information with the classification label of ghost type is given priority. At the same time, it is necessary to consider whether the character attributes meet the equipment requirements, task status, etc.

[0060] Furthermore, step S1 further includes the steps of:

[0061] S11, preprocessing the acquired client data:

[0062] Data cleaning: remove invalid and abnormal data, and fill in missing values;

[0063] Data transformation: discretize and normalize numerical features according to data categories and encode classification information.

[0064] From the above description, we can see that before inputting into the AI ​​model, the data needs to be cleaned and converted to remove invalid and abnormal data and fill in missing values ​​to improve the effectiveness of the recommendation.

[0065] Furthermore, step S3 is specifically as follows:

[0066] According to the index values ​​of the skin and equipment, combined with the similarity information of the item equipment description, weight calculation is performed to obtain a recommended weight value;

[0067] Based on the recommendation weight value, the equipment is sorted according to its position, and skins and equipment are recommended according to the sorting result.

[0068] From the above description, it can be seen that the weight calculation is performed based on the indicator value output by the AI ​​model and the similarity information of the equipment description, and the ranking recommendation is performed based on the recommended weight value finally calculated.

[0069] Furthermore, the method further comprises the steps of:

[0070] S4, determining whether the player accepts to use the recommended skins and equipment, if yes, storing the recommended data, otherwise changing the recommended content, recommending again, and recording the number of recommendations;

[0071] If the number of recommendations reaches the preset threshold, the next time the player does not accept the recommended skins and equipment, the ranking results will be displayed for the player to choose.

[0072] From the above description, it can be seen that the data of whether the player accepts the recommendation plan is recorded and feedback is given, and the recommendation record is recorded. The recommendation record includes the client data at the time of recommendation, the recommended fashion or skin information, and whether the user accepts it, etc. It is used to optimize the AI ​​model and can be used as the historical behavior information of the character for subsequent recommendations.

[0073] The present invention discloses an AI-based game character equipment recommendation method and storage medium, which are suitable for providing a convenient equipment recommendation solution for game players.

[0074] Please refer to Figure 1 to Figure 2 , Embodiment 1 of the present invention is:

[0075] A method for recommending equipment for game characters based on AI, comprising the steps of:

[0076] S1. Obtain client data, flatten and classify the client data;

[0077] The client data at least includes game scene information, skin and equipment information, and version activity information, festival activity information, task status information, character attribute status information, and character character historical behavior information;

[0078] The game scene information includes at least one of big world type information, battlefield type information, opponent information, interface configuration information and skill configuration information;

[0079] The skin and equipment information includes at least one of item configuration information, clothing configuration information, weapon configuration information, light effect configuration information, fantasy beast configuration information and mount configuration information.

[0080] In this embodiment, it is first necessary to collect the client parameters and user data as described above, so as to construct the feature vector of equipment recommendation and indicators such as personalized scoring preference.

[0081] Specific parameters include map type (big world type) (plains, swamps, forests, etc.), player preferences (pursuing high attack, pursuing balance, pursuing defense, etc.), and characteristics of monsters faced (high defense, high attack, various resistances).

[0082] Feature vectors can be understood as label vectors, which are obtained through presetting and cleaning statistics. These text vectors will be put into the specified model for training, and some will also be based on indicators such as cosine similarity to perform operations such as similarity sorting to enrich indicators. Personalized preferences are related indicators built based on player data analysis.

[0083] Client data is mainly based on client configuration files. Depending on the different version packages, there are generally hundreds to thousands of small files. AI training will first flatten the data of each dimension and then classify and label them.

[0084] S11, preprocessing the acquired client data:

[0085] Data cleaning: remove invalid and abnormal data, and fill in missing values;

[0086] Data transformation: discretize and normalize numerical features according to data categories and encode classification information.

[0087] In this embodiment, it is necessary to clean and transform the collected and extracted data, remove invalid or abnormal data, fill missing values, normalize or discretize numerical features, encode type features and other operations, depending on the data type. Filter abnormal and non-compliant data according to preset rules, such as data that is obviously abnormal (such as the time to kill monsters is much longer than the average time). Since the parameters are all known meaning data, the main method is to delete missing values ​​and fill them with fixed values ​​(mostly default values). If other probabilities and similarities are used to fill them, data distortion may occur.

[0088] Discretization is used for continuous data, while normalization is used for most other collected data to further avoid data bias and outliers affecting the expected results. For categorical text, such as gender, etc., in addition to basic label encoding (hard coding), model encoding and one-hot encoding may also be used.

[0089] S2. Generate index values ​​of skins and equipment in the current game scenario using an AI model based on the client data and classification label information.

[0090] In this embodiment, the indicators are fine-tuned according to the game characteristics, and model selection and training, as well as model optimization and evaluation are performed based on the fine-tuned indicators.

[0091] In this embodiment, the index value of the AI ​​model is generated based on data such as game scene information, version activity information, festival activity information, task status information, character attribute status information, and historical behavior information of the character. For example, pk equipment recommendations will be dynamically recommended based on the battlefield and opponent type; when it comes to PVE fighting wild monsters, corresponding recommendations will be made based on the attributes of wild monsters; functional equipment recommendations will be recommended based on the matching of functions and scenes, and different types of situations, such as open terrain, will give priority to long-range and long-handled weapons and equipment; such as complex terrain, will give priority to general distance and conventional functional weapons and equipment; if the scene involves fire damage, poisonous fog damage, etc., corresponding auxiliary equipment will also be recommended (such as lightning protection in rainy days, heatstroke prevention in deserts, fire prevention in volcanoes, cold prevention, etc.), which will be specifically recommended based on the client configuration and the actual game process status of the player. At the same time, considering holiday activities and other situations, such as Halloween, equipment or fashion with ghost characteristics will be recommended first.

[0092] S3. Recommend skins and equipment based on the indicator values ​​of the skins and equipment.

[0093] Step S3 is specifically as follows:

[0094] According to the index values ​​of the skin and equipment, combined with the similarity information of the item equipment description, weight calculation is performed to obtain a recommended weight value;

[0095] Based on the recommendation weight value, the equipment is sorted according to its position, and skins and equipment are recommended according to the sorting result.

[0096] In this embodiment, the AI ​​model is called, and mixed recommendations are performed based on auxiliary filtering functions, item equipment description similarity recommendation and other functions. Finally, weight values ​​are calculated based on the recommendation results, and the ranking recommendations of item equipment are performed based on the weight values.

[0097] Among them, the auxiliary filtering function mainly plays the role of filtering invalid and abnormal recommendation results, which is the filtering of recommendation results; it adopts the filtering of mandatory indicator range and type to increase the robustness of the system and improve the rationality of the recommendation results.

[0098] The item description similarity recommendation function compares and recommends based on the similarity of the specific text descriptions of items. It supplements the recommendation results at the text level. The previous article mainly recommends based on attribute parameters and user data dimensions.

[0099] Please refer to Figure 3 , Embodiment 2 of the present invention is:

[0100] A method for recommending equipment for game characters based on AI, which differs from the first embodiment in that it further comprises the steps of:

[0101] S4, judging whether the player accepts to use the recommended skins and equipment, if yes, storing the recommendation record, otherwise storing the recommendation record, changing the recommendation content, recommending again, and recording the number of recommendations;

[0102] If the number of recommendations reaches the preset threshold, the next time the player does not accept the recommended skins and equipment, the ranking results will be displayed for the player to choose.

[0103] In this embodiment, data is recorded and fed back on whether the player accepts the recommendation scheme, and the recommendation record is recorded. The recommendation record includes the client data at the time of recommendation, the recommended fashion or skin information, and whether the user accepts it, etc. It is used to optimize the AI ​​model and can be used as the historical behavior information of the character for subsequent recommendations.

[0104] Embodiment 3 of the present invention is:

[0105] A storage medium having a computer program stored thereon, wherein the computer program, when executed, implements the steps in the above-mentioned method for recommending equipment for game characters based on AI.

[0106] In summary, the present invention provides an AI-based game character equipment recommendation method and storage medium, which obtain client information including game scene information and skin and equipment information, and flatten and classify the client information, and use a pre-trained AI model to output the index value of the skin and equipment in the current game scene according to the classification label information, and recommend equipment based on the index value, so as to achieve the recommendation of equipment and skins adapted to players according to different game scenes, thereby improving the convenience of players' skin and equipment selection.

[0107] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent transformations made using the contents of the present invention's specification and drawings, or directly or indirectly applied in related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for recommending game character equipment based on AI, characterized in that: Includes steps: S1. Obtain client data, flatten and classify the client data; The client data includes game scene information and character skin and equipment information; S2. Generate index values ​​of skins and equipment in the current game scenario using an AI model based on the client data and classification label information; S3. Recommend skins and equipment based on the indicator values ​​of the skins and equipment.

2. The AI-based game character equipment recommendation method according to claim 1, characterized in that: The client data at least includes game scene information, skin and equipment information, and version activity information, festival activity information, task status information, character attribute status information, and character character historical behavior information; The game scene information includes at least one of big world type information, battlefield type information, opponent information, interface configuration information and skill configuration information; The skin and equipment information includes at least one of item configuration information, clothing configuration information, weapon configuration information, light effect configuration information, fantasy beast configuration information and mount configuration information.

3. The AI-based game character equipment recommendation method according to claim 1, characterized in that: Step S1 also includes the steps of: S11, preprocessing the acquired client data: Data cleaning: remove invalid and abnormal data, and fill in missing values; Data transformation: discretize and normalize numerical features according to data categories and encode classification information.

4. The AI-based game character equipment recommendation method according to claim 1, characterized in that: Step S3 is specifically as follows: According to the index values ​​of the skin and equipment, combined with the similarity information of the item equipment description, weight calculation is performed to obtain a recommended weight value; Based on the recommendation weight value, the equipment is sorted according to its position, and skins and equipment are recommended according to the sorting result.

5. The AI-based game character equipment recommendation method according to claim 4, characterized in that: Also includes the steps: S4, determining whether the player accepts to use the recommended skins and equipment, if yes, storing the recommended data, otherwise changing the recommended content, recommending again, and recording the number of recommendations; If the number of recommendations reaches the preset threshold, the next time the player does not accept the recommended skins and equipment, the ranking results will be displayed for the player to choose.

6. A storage medium having a computer program stored thereon, characterized in that: When the computer program is executed, the following steps are implemented: S1. Obtain client data, flatten and classify the client data; The client data includes game scene information and character skin and equipment information; S2. Generate index values ​​of skins and equipment in the current game scenario using an AI model according to the client data and classification label information; S3. Recommend skins and equipment based on the indicator values ​​of the skins and equipment.

7. A storage medium according to claim 6, characterized in that: The client data at least includes game scene information, skin and equipment information, and version activity information, festival activity information, task status information, character attribute status information, and character character historical behavior information; The game scene information includes at least one of big world type information, battlefield type information, opponent information, interface configuration information and skill configuration information; The skin and equipment information includes at least one of item configuration information, clothing configuration information, weapon configuration information, light effect configuration information, fantasy beast configuration information and mount configuration information.

8. A storage medium according to claim 6, characterized in that: Step S1 also includes the steps of: S11, preprocessing the acquired client data: Data cleaning: remove invalid and abnormal data, and fill in missing values; Data transformation: discretize and normalize numerical features according to data categories and encode classification information.

9. A storage medium according to claim 6, characterized in that: Step S3 is specifically as follows: According to the index values ​​of the skin and equipment, combined with the similarity information of the item equipment description, weight calculation is performed to obtain a recommended weight value; Based on the recommendation weight value, the equipment is sorted according to its position, and skins and equipment are recommended according to the sorting result.

10. A storage medium according to claim 6, characterized in that: Also includes the steps: S4, determining whether the player accepts to use the recommended skins and equipment, if yes, storing the recommended data, otherwise changing the recommended content, recommending again, and recording the number of recommendations; If the number of recommendations reaches the preset threshold, the next time the player does not accept the recommended skins and equipment, the ranking results will be displayed for the player to choose.

Images