Intelligent grading method for game elements

By constructing a topological index and interpolation method to reconstruct item attribute data, generating a sensitive node subgraph and performing node mapping calibration, the problem of blurred item classification boundaries in online games is solved, and a more stable item classification structure is achieved.

CN121731770APending Publication Date: 2026-03-27HANGZHOU HURRICANE NETWORK CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202610093225.9
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-01-23
Publication Date
2026-03-27

AI Technical Summary

Technical Problem

In online games, under the conditions of limited computing power and response latency, existing technologies suffer from problems such as missing item attribute data and dynamic changes in player interaction, leading to blurred item classification boundaries and reduced structural stability.

Method used

By acquiring initial item attribute data and player interaction data, a missing topological index is constructed. The target item attribute data is reconstructed using topological constraint interpolation methods, an attribute topological relationship graph is generated, sensitive attribute nodes are identified, node topological mapping and position calibration are performed, and the item classification boundary is output.

Benefits of technology

The ambiguity of item tier boundaries has been reduced, the stability of the item tier structure has been improved, and the item tier boundaries are ensured to be clear and consistent with actual player behavior.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN121731770A_ABST
    Figure CN121731770A_ABST
Patent Text Reader

Abstract

The invention relates to the technical field of graph structure data processing, and discloses an intelligent game element grading method which comprises the steps of obtaining initial prop attribute data and player interaction data, constructing a missing topological index and reconstructing the initial prop attribute data in combination with a topological constraint interpolation method, and generating target prop attribute data. Data incomplete interference is weakened; generating an attribute topological relation graph based on the target prop attribute data, identifying sensitive attribute nodes to form a sensitive node subgraph, and depicting key differences in a centralized manner; determining a node topology mapping area according to the sensitive node sub-graph, executing node topology position mapping, generating a prop grading topological graph, and unifying grading coordinates; and performing node topological position calibration on the sensitive node sub-graph based on the prop grading topological graph, outputting a prop grading boundary, reducing the fuzziness of the prop grading boundary, and improving the stability of a prop grading structure.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of graph structure data processing technology, and more specifically, to a method for intelligent hierarchical classification of game elements. Background Technology

[0002] In online game interactive scenarios, game props are usually used as carriers. Under the conditions of limited computing power and response latency, intelligent hierarchical management of prop elements is achieved. Existing technologies mostly use feature mapping, cluster analysis, rule judgment and threshold screening methods to classify and grade props. The premise for their applicability is usually high data integrity, stable prop attributes and slow changes in the interactive environment.

[0003] In real-world game environments, there are random missing data for item interactions, and item attribute definitions change dynamically with player interactions. These two factors combined reduce the ability of existing methods to express differences in item attributes, leading to blurred item classification boundaries and reduced structural stability. Therefore, the technical problem that needs to be solved is how to ensure the stability of the item classification structure under conditions of incomplete data and while reducing the ambiguity of the intelligent classification boundaries of game items.

[0004] In view of this, the present invention proposes a smart classification method for game elements to solve the above problems. Summary of the Invention

[0005] To overcome the aforementioned shortcomings of the prior art, this invention provides a method for intelligent grading of game elements.

[0006] To achieve the above objectives, the present invention provides the following technical solution: Firstly, a method for intelligently classifying game elements is provided, including: Obtain initial item attribute data and player interaction data, construct a missing topological index based on the player interaction data, and reconstruct the initial item attribute data using topological constraint interpolation and the missing topological index to generate target item attribute data; Generate an attribute topology graph based on the target item attribute data, determine sensitive attribute nodes based on the node connectivity structure of the attribute topology graph, and generate a sensitive node subgraph. The node topology mapping area is determined based on the sensitive node subgraph, and the node topology position is mapped to the attribute topology relationship graph based on the node topology mapping area to generate the item hierarchical topology graph. Based on the item classification topology graph, perform node topology position calibration on the sensitive node subgraph and output the item classification boundary.

[0007] In some embodiments, the initial item attribute data is reconstructed using a topological constraint interpolation method and a missing topological index to generate target item attribute data, including: Read the order in which players choose items during consecutive actions from player interaction data, and record each consecutive item choice as an ordered record to form an item choice sequence; The item selection sequence is combined and split, adjacent item pairs are extracted from the item selection sequence, and records with the same item pair structure are merged to form an item combination set; Based on the order in which each pair of items appears in the item combination set, calculate the sequential connection value of each pair of items, and then aggregate all sequential connection values ​​to form the item connection weight. Construct an item combination network based on the item connection weights, and identify the node positions corresponding to the missing topology indexes in the item combination network to form a set of missing nodes; Based on the attributes of the neighboring nodes of the missing node set in the item combination network, the initial item attribute data is subjected to topological constraint interpolation to generate the target item attribute data.

[0008] In some embodiments, based on the order in which each item pair appears in the item combination set, the sequential connection value of each item pair is calculated, and all sequential connection values ​​are aggregated to form the item connection weight, including: Read the position of each item pair in the item selection sequence one by one from the item combination set, and record the order of each item pair to form a sequence record set; For items with the same sequential marker in the sequential record set, the records are accumulated to form an item sequence count; Calculate the corresponding connection weight value for each item pair in the item combination set according to the item order count, and collect all connection weight values ​​to form the item connection weight.

[0009] In some embodiments, topological constraint interpolation is performed on the initial item attribute data based on the attributes of the neighboring nodes of the missing node set in the item combination network to generate target item attribute data, including: Traverse the set of missing nodes and extract the adjacent item nodes that are directly connected to the positions of each node in the set of missing nodes from the item combination network to form an adjacent node set; Read the known item attributes of each adjacent item node from the adjacent node set, and arrange the read known item attributes according to the node order of the adjacent item nodes to form an adjacent attribute set; Based on the item connection weight, interpolation coefficients are assigned to each known item in the adjacent attribute set, and the interpolation coefficients are combined with the corresponding known items to form an interpolation input set. Perform topological constraint interpolation calculations on the interpolation input set, and write the results of the topological constraint interpolation calculations to the node positions corresponding to the missing topological indexes to generate target item attribute data.

[0010] In some embodiments, sensitive attribute nodes are determined based on the node connectivity structure of the attribute topology graph, and a sensitive node subgraph is generated, including: Read the topological connection information of all nodes from the attribute topology graph and record the connection paths between each node to form a set of node connection paths; The connection paths of nodes in the node connectivity path set are split to divide the attribute topology graph into multiple sub-regions, forming a set of connected sub-regions. Based on the number of nodes contained in each sub-region of the connected sub-region set, sub-regions with a number of nodes exceeding a preset node number threshold are identified and formed as candidate sensitive regions; Extract nodes and their corresponding item attributes from candidate sensitive regions, mark the extracted nodes as sensitive attribute nodes, and generate a sensitive node subgraph based on the connection information of the sensitive attribute nodes in the attribute topology graph.

[0011] In some embodiments, the connection paths of nodes in the node connectivity path set are subjected to connectivity segmentation processing, dividing the attribute topology graph into multiple sub-regions to form a set of connected sub-regions, including: Identify and extract node groups with cyclic connection structures from the node connectivity path set, and perform node cyclic cutting processing on the node groups to form initial sub-regions; Calculate the path intersection nodes between the initial sub-region and other nodes, and record the positions of the path intersection nodes to form an intersection node record; The boundary nodes of the initial sub-region are segmented based on the intersection node records to obtain multiple independent sub-regions without intersection nodes, forming a set of connected sub-regions.

[0012] In some embodiments, a node topology mapping region is determined based on a sensitive node subgraph, and the node topology position is mapped to the attribute topology graph based on the node topology mapping region to generate an item hierarchical topology graph, including: Read the topological position of each sensitive attribute node in the sensitive node subgraph, and extract the positions of non-sensitive nodes connected to the sensitive attribute nodes to form a candidate set of topological mappings; Perform node clustering on the non-sensitive node locations in the candidate set of topology mapping, and merge the regions of non-sensitive nodes with adjacent topological locations to form node mapping regions; Extract the topological region boundary position of the node mapping area, and adjust the node position of the attribute topology relationship graph according to the region boundary position to generate the item hierarchical topology map.

[0013] In some embodiments, node topology position calibration is performed on the sensitive node subgraph according to the item classification topology graph, and the item classification boundary is output, including: Extract the topological location information of all sensitive attribute nodes from the prop grading topology map, and record the positions of the adjacent nodes of each sensitive attribute node to form a node calibration input set; Based on the positional differences between sensitive attribute nodes and adjacent nodes in the node calibration input set, the node position of the sensitive node subgraph is corrected to obtain the calibrated sensitive node subgraph. Generate topological boundary lines based on the node positions of the calibration sensitive node subgraph, extract the position data of the topological boundary lines, and output the prop classification boundary.

[0014] In some embodiments, based on the positional differences between sensitive attribute nodes and adjacent nodes in the node calibration input set, node position correction is performed on the sensitive node subgraph to obtain a calibrated sensitive node subgraph, including: For each sensitive attribute node in the node calibration input set, calculate the coordinate vector difference between its topological position and that of each neighboring node, and record the coordinate vector difference to form a node difference set; The calibration displacement vector of each sensitive attribute node is calculated based on the node difference set, and the displacement direction and displacement distance are determined. The coordinates of the corresponding nodes in the sensitive node subgraph are updated based on the displacement direction and displacement distance to obtain the calibrated sensitive node subgraph.

[0015] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention acquires initial item attribute data and player interaction data, constructs a missing topological index based on the player interaction data, and reconstructs the initial item attribute data using a topological constraint interpolation method in conjunction with the missing topological index to generate target item attribute data. This reduces the disturbance of attribute difference expression caused by data incompleteness. Based on the target item attribute data, an attribute topological relationship graph is generated. Sensitive attribute nodes are determined according to the node connectivity structure of the attribute topological relationship graph, and sensitive node subgraphs are formed, enabling the grading judgment to focus on high-impact attributes. Based on the sensitive node subgraphs, node topological mapping regions are determined, and node topological position mapping is performed on the attribute topological relationship graph to generate an item grading topological graph. Item grading coordinates are unified in a multi-scale attribute space. Based on the item grading topological graph, node topological position calibration is performed on the sensitive node subgraphs, and the item grading boundary is output, thereby reducing the ambiguity of the item grading boundary and improving the stability of the item grading structure. Attached Figure Description

[0016] Figure 1 This is a flowchart illustrating a method for intelligent grading of game elements according to the present invention. Figure 2 This is a schematic diagram of the structure of a game element intelligent grading system according to the present invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments. Based on the described embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] Unless otherwise defined, the technical or scientific terms used in this invention shall have the ordinary meaning as understood by one of ordinary skill in the art to which this invention pertains. The terms “first,” “second,” and similar terms used in this invention 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 term covers the element or object listed after the term and its equivalents, without excluding other elements or objects. Terms such as “connection” or “linked” are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect.

[0019] In the technical solution of this invention, the user information (including but not limited to user personal information, user image information, user device information, such as location information) and data (including but not limited to data used for analysis, stored data, and displayed data) involved are all information and data authorized by the user or fully authorized by all parties. Furthermore, the collection, storage, use, processing, transmission, provision, disclosure, and application of related data are all carried out in accordance with relevant laws, regulations, and standards, and necessary confidentiality measures have been taken. They do not violate public order and good morals, and corresponding operation entry points are provided for users to choose to authorize or refuse.

[0020] Example 1 Figure 1 This disclosure illustrates at least one embodiment of a smart rating method for game elements, comprising: S10: Obtain initial item attribute data and player interaction data, construct a missing topological index based on the player interaction data, and reconstruct the initial item attribute data using the topological constraint interpolation method and the missing topological index to generate target item attribute data; In this embodiment, "initial item attribute data" refers to the original attribute values ​​of game items before player interaction, such as the item's initial attack power, defense power, or special effects. "Player interaction data" refers to the specific behavior records of players actually using items during the game, such as the order of item use, frequency of use, and combination method. "Missing topological index" is used to identify the location index where attribute data is incomplete due to random player operations or data loss. By establishing topological constraint relationships between items and using topological constraint interpolation methods, it is possible to reasonably interpolate and reconstruct the missing data based on the attributes of adjacent items within the topological structure to accurately supplement the missing parts. This ensures that the reconstructed target item attribute data can reflect the attribute changes after actual item interaction, effectively solving the problem of blurred and unstable item classification boundaries caused by data loss.

[0021] The initial item attribute data is reconstructed using topological constraint interpolation and missing topological indices to generate the target item attribute data, including: Read the order in which players choose items during consecutive actions from player interaction data, and record each consecutive item choice as an ordered record to form an item choice sequence; The item selection sequence is combined and split, adjacent item pairs are extracted from the item selection sequence, and records with the same item pair structure are merged to form an item combination set; Based on the order in which each pair of items appears in the item combination set, calculate the sequential connection value of each pair of items, and then aggregate all sequential connection values ​​to form the item connection weight. Construct an item combination network based on the item connection weights, and identify the node positions corresponding to the missing topology indexes in the item combination network to form a set of missing nodes; Based on the attributes of the neighboring nodes of the missing node set in the item combination network, the initial item attribute data is subjected to topological constraint interpolation to generate the target item attribute data.

[0022] It should be noted that the above method uses real player interaction data to meticulously uncover the structured relationships between items during actual use, thereby enabling the precise determination of interpolation benchmarks for data gaps.

[0023] For example, in a player's consecutive actions, if they select item A (attack power 50), item B (defense power 30), item C (speed 15), and then item A again, the corresponding item selection sequence is [A,B,C,A]. After combining and splitting, the resulting item combination set is {(A,B), (B,C), (C,A)}. Further analysis of the actual order in which these combinations appear reveals that if the combination (A,B) has a higher connection value, it indicates that the player prefers to use A before B. Using this connection value as the basis for item connection weights, an item combination network is constructed. This allows subsequent interpolation and supplementation processes to reasonably reference the player's actual item combination preferences, ultimately resulting in target item attribute data that more accurately reflects the actual player behavior characteristics in the game environment.

[0024] Based on the order in which each item pair appears in the item combination set, calculate the sequence connection value for each item pair, and then aggregate all sequence connection values ​​to form the item connection weight, including: Read the position of each item pair in the item selection sequence one by one from the item combination set, and record the order of each item pair to form a sequence record set; For items with the same sequential marker in the sequential record set, the records are accumulated to form an item sequence count; Calculate the corresponding connection weight value for each item pair in the item combination set according to the item order count, and collect all connection weight values ​​to form the item connection weight.

[0025] Furthermore, specifically, the purpose of calculating the item connection weight is to accurately measure the degree of tendency of players to combine various items in actual gameplay, so as to have a clear and reliable reference basis when reconstructing missing data later.

[0026] For example, in the item combination set {(A,B), (B,C), (C,A)} mentioned earlier, if the (A,B) combination appears 3 times in the player's selection sequence, and each time it appears in the order of A first, then B, while the (B,C) combination appears 2 times and the (C,A) combination appears only once, then the order of the item combinations in the sequence record set is used as follows: for example, it can be recorded as {(A→B: 3 times), (B→C: 2 times), (C→A: 1 time)}. Then, the order is accumulated based on the number of occurrences. The item sequence count is obtained, and then the connection weight value of each item pair is calculated. For example, the connection weight value of item pair (A,B) is 0.5, (B,C) is 0.3, and (C,A) is 0.2. Thus, the item connection weight is {(A,B):0.5,(B,C):0.3,(C,A):0.2}. Through this connection weight, the actual interaction strength between items can be clearly identified in the interpolation reconstruction, effectively improving the matching degree between the interpolation result and the real interaction situation.

[0027] Based on the attributes of the neighboring nodes of the missing node set in the item combination network, the initial item attribute data is subjected to topological constraint interpolation to generate the target item attribute data, including: Traverse the set of missing nodes and extract the adjacent item nodes that are directly connected to the positions of each node in the set of missing nodes from the item combination network to form an adjacent node set; Read the known item attributes of each adjacent item node from the adjacent node set, and arrange the read known item attributes according to the node order of the adjacent item nodes to form an adjacent attribute set; Based on the item connection weight, interpolation coefficients are assigned to each known item in the adjacent attribute set, and the interpolation coefficients are combined with the corresponding known items to form an interpolation input set. Perform topological constraint interpolation calculations on the interpolation input set, and write the results of the topological constraint interpolation calculations to the node positions corresponding to the missing topological indexes to generate target item attribute data.

[0028] It should be further explained that the core of the topological constraint interpolation method for missing item attribute data lies in accurately determining the interpolation weight based on the adjacent node attributes and item connection weights in order to obtain target data that conforms to the actual situation in the game.

[0029] For a specific example, in the constructed item combination network, if item node X has a missing attribute, its adjacent node set is {item A, item B}, and the item connection weights are defined as 0.6 from item A to item X and 0.4 from item B to item X, if item A's attributes are [Attack 60, Defense 20] and item B's attributes are [Attack 40, Defense 30], then the adjacent attribute set is [(Attack 60, Defense 20), (Attack 40, Defense 30)]. After allocating interpolation coefficients according to the connection weights, an interpolation input set [(60×0.6,20×0.6),(40×0.4,30×0.4)] is formed. After topological constraint interpolation calculation, the target item attribute data reconstructed at the node X position corresponding to the missing topological index is [Attack Power 52, Defense Power 24]. This method effectively utilizes real player operation behavior, provides reliable data basis for interpolation reconstruction, and makes the reconstructed attribute data closer to the actual player behavior habits.

[0030] S20: Generate an attribute topology graph based on the target item attribute data, determine sensitive attribute nodes based on the node connectivity structure of the attribute topology graph, and generate a sensitive node subgraph; In this embodiment, the "attribute topology graph" is a network structure based on the target item attribute data, representing the degree of attribute correlation between items. For example, if items have similar or complementary attributes, a topological connection will be established between the corresponding nodes. "Sensitive attribute nodes" represent nodes that play a key role in item classification and have close attribute correlation in the actual game environment. By analyzing the connectivity structure of the nodes, sensitive nodes can be clearly extracted to improve the accuracy of subsequent item classification processing. For example, based on the aforementioned items A, B, and C, if the attribute of item A is strongly correlated with the attributes of items B and C, it constitutes a sensitive attribute node and forms a corresponding sensitive node subgraph, thereby highlighting the node group that has a significant impact on the classification boundary and effectively avoiding the phenomenon of ambiguous boundary division.

[0031] Based on the node connectivity structure of the attribute topology graph, sensitive attribute nodes are identified, and a sensitive node subgraph is generated, including: Read the topological connection information of all nodes from the attribute topology graph and record the connection paths between each node to form a set of node connection paths; The connection paths of nodes in the node connectivity path set are split to divide the attribute topology graph into multiple sub-regions, forming a set of connected sub-regions. Based on the number of nodes contained in each sub-region of the connected sub-region set, sub-regions with a number of nodes exceeding a preset node number threshold are identified and formed as candidate sensitive regions; Extract nodes and their corresponding item attributes from candidate sensitive regions, mark the extracted nodes as sensitive attribute nodes, and generate a sensitive node subgraph based on the connection information of the sensitive attribute nodes in the attribute topology graph.

[0032] To further clarify, the above steps aim to accurately pinpoint the nodes that play a core role in item grading based on the actual topological characteristics between nodes. Specifically, assuming that nodes {A, B, C, D, E, F} in the attribute topology graph are connected to form a topological structure, where nodes A have direct connections with B and C, nodes B, C, D, and E form a densely connected sub-region, and node F is relatively independent, then by analyzing the topological structure, for example, by defining a preset threshold of 3 for the number of nodes, the region {B, C, D, E} with 4 nodes can be clearly identified as a candidate sensitive region. After extracting the node attributes within this region, they are marked as sensitive attribute nodes, and a corresponding sensitive node sub-graph is generated, thereby significantly improving the accuracy of identifying key nodes and providing a clear and robust structured basis for subsequent item grading.

[0033] The connection paths of nodes in the node connectivity path set are processed by connectivity segmentation, dividing the attribute topology graph into multiple sub-regions, forming a set of connected sub-regions, including: Identify and extract node groups with cyclic connection structures from the node connectivity path set, and perform node cyclic cutting processing on the node groups to form initial sub-regions; Calculate the path intersection nodes between the initial sub-region and other nodes, and record the positions of the path intersection nodes to form an intersection node record; The boundary nodes of the initial sub-region are segmented based on the intersection node records to obtain multiple independent sub-regions without intersection nodes, forming a set of connected sub-regions.

[0034] It is important to further emphasize that the key to the above-mentioned process of connecting and splitting the node connection paths lies in handling the complex node connection structure in the attribute topology graph. By finely cutting, the topology structure becomes clearer and easier to distinguish. For example, continuing with the above node structure, nodes {B, C, D, E} constitute the initial sub-region of the cyclic connection structure. When analyzing the connection path, nodes B and C are identified as intersecting nodes with node A. Using the intersecting node record {B, C} as the dividing basis, the initial sub-region is cut from the position of the intersecting node to form two independent connected sub-regions {B, D, E} and {C} without intersecting nodes. This significantly reduces the structural complexity between nodes in the attribute topology graph, making the identification of sensitive nodes more accurate and the expression of the prop attribute structure clearer and more stable.

[0035] S30: Determine the node topology mapping area based on the sensitive node subgraph, and map the node topology position of the attribute topology relationship graph based on the node topology mapping area to generate the item hierarchical topology graph; In this embodiment, the core objective of the above processing steps is to accurately delineate the topology mapping region based on the sensitive node subgraph and perform node position mapping in the attribute topology relationship graph to enhance the boundary structure stability of the item grading topology graph. Specifically, taking the previously generated sensitive node subgraph as an example, assuming that the subgraph includes sensitive attribute nodes {B, D, E}, firstly, based on the position of the sensitive attribute nodes, extract the directly connected non-sensitive nodes {A, C, F}, then merge the regions based on these node positions, divide the topology mapping region, and clarify the region boundary position. Using this boundary position, remap the nodes in the overall attribute topology relationship graph to form an item grading topology graph with clear boundaries and obvious grading effect.

[0036] Based on the sensitive node subgraph, the node topology mapping region is determined, and based on the node topology mapping region, the node topology position is mapped to the attribute topology relationship graph to generate an item hierarchical topology graph, including: Read the topological position of each sensitive attribute node in the sensitive node subgraph, and extract the positions of non-sensitive nodes connected to the sensitive attribute nodes to form a candidate set of topological mappings; Perform node clustering on the non-sensitive node locations in the candidate set of topology mapping, and merge the regions of non-sensitive nodes with adjacent topological locations to form node mapping regions; Extract the topological region boundary position of the node mapping area, and adjust the node position of the attribute topology relationship graph according to the region boundary position to generate the item hierarchical topology map.

[0037] To further explain, the above steps aim to clearly delineate regions for sensitive and non-sensitive nodes based on their topological locations, and adjust node positions through region boundary mapping. This improves the stability and structural clarity of the hierarchical topology graph. For example, based on the sensitive nodes {B, D, E} in the sensitive node subgraph, the non-sensitive node positions {A, C, F} directly connected to the sensitive nodes are first located and extracted as a candidate set for topology mapping. Then, clustering is performed on the nodes {A, C, F} in the candidate set to form adjacent node cluster regions. For example, if the nodes {A, C} are close, they are merged into node mapping region 1, and node {F} is treated as node mapping region 2. After that, the region boundary positions are extracted, and the positions of each node in the attribute topology graph are readjusted based on the boundaries of region 1 and region 2. This constructs a prop hierarchical topology graph with clear node position structure, well-defined boundaries, and ease of subsequent management and operation, making the hierarchical structure more stable and reliable.

[0038] S40: Perform node topology position calibration on sensitive node subgraphs based on the item classification topology graph, and output item classification boundaries.

[0039] Based on the item grading topology map, perform node topology position calibration on the sensitive node subgraph and output the item grading boundary, including: Extract the topological location information of all sensitive attribute nodes from the prop grading topology map, and record the positions of the adjacent nodes of each sensitive attribute node to form a node calibration input set; Based on the positional differences between sensitive attribute nodes and adjacent nodes in the node calibration input set, the node position of the sensitive node subgraph is corrected to obtain the calibrated sensitive node subgraph. Generate topological boundary lines based on the node positions of the calibration sensitive node subgraph, extract the position data of the topological boundary lines, and output the prop classification boundary.

[0040] Furthermore, the above node position calibration process determines the calibration displacement by calculating the position difference between the sensitive node and its adjacent nodes. Taking node B as an example, assuming that the position of node B is (2,3) and the adjacent nodes are A (1,2) and C (3,2), the coordinate difference vectors between node B and nodes A and C are calculated first. The difference between node B and node A is (1,1), and the difference between node B and node C is (-1,1). Then, based on these two difference vectors, the calibration displacement vector of node B is calculated as (0,1), indicating that node B should move 1 unit in the positive direction. Based on this calculation result, the position of node B is updated to (2,4). Similarly, the positions of nodes D and E are calibrated. Finally, the topological boundary line is generated based on the calibrated sensitive node positions to clearly indicate the item classification boundary, making the item classification boundary more accurate and clear.

[0041] Based on the positional differences between sensitive attribute nodes and their neighboring nodes in the node calibration input set, the sensitive node subgraph is corrected to obtain a calibrated sensitive node subgraph, including: For each sensitive attribute node in the node calibration input set, calculate the coordinate vector difference between its topological position and that of each neighboring node, and record the coordinate vector difference to form a node difference set; The calibration displacement vector of each sensitive attribute node is calculated based on the node difference set, and the displacement direction and displacement distance are determined. The coordinates of the corresponding nodes in the sensitive node subgraph are updated based on the displacement direction and displacement distance to obtain the calibrated sensitive node subgraph.

[0042] In this embodiment, the main purpose of this step is to further correct the node positions in the sensitive node subgraph by calculating the specific positional differences between the sensitive attribute node and its neighboring nodes, so as to ensure that the node position layout is reasonable and the item classification boundary is clear. Specifically, taking the position correction of node B in the above calibration example as an example, firstly, according to the node calibration input set, the coordinate difference between node B (2,3) and its neighboring nodes A (1,2) and C (3,2) is calculated to obtain the node difference set {(1,1), (-1,1)}. Then, the calibration displacement vector of node B is determined by combining the node difference set. The displacement direction is calculated as the positive direction of the vertical coordinate and the displacement distance is 1 unit. Then, the coordinates of node B are updated to the new position (2,4). Similarly, the calibration displacement vectors of nodes D and E are calculated and the coordinate positions are updated. Through the above processing, the calibration sensitive node subgraph is obtained, which makes the node distribution structure more reasonable and further ensures that the item classification boundary is clear and stable, overcoming the shortcomings of the existing technology of fuzzy and unstable item classification boundary.

[0043] Example 2 Please see Figure 2As shown, based on the same inventive concept, this embodiment discloses a game element intelligent grading system. For details not covered in this embodiment, please refer to the relevant parts of Embodiment 1. The system includes: Attribute Reconstruction Module: Used to obtain initial item attribute data and player interaction data, construct missing topological indexes based on player interaction data, and reconstruct the initial item attribute data using topological constraint interpolation methods and missing topological indexes to generate target item attribute data; Sensitive Identification Module: Used to generate an attribute topology graph based on the target item attribute data, determine sensitive attribute nodes based on the node connectivity structure of the attribute topology graph, and generate a sensitive node subgraph; Location Mapping Module: Used to determine the node topology mapping area based on the sensitive node subgraph, and to perform node topology location mapping on the attribute topology relationship graph based on the node topology mapping area to generate a prop hierarchical topology graph; Position calibration module: used to perform node topology position calibration on sensitive node subgraphs according to the item classification topology map, and output item classification boundaries.

[0044] The accompanying drawings of the embodiments of this invention only involve the structures involved in the embodiments of this invention. Other structures can refer to the general design. In the absence of conflict, the features of the same embodiment and different embodiments of this invention can be combined with each other. The above are only specific implementations of this invention, but the protection scope of this invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in this invention should be included within the protection scope of this invention. Therefore, the protection scope of this invention should be determined by the protection scope of the claims.

Claims

1. A method for intelligently classifying game elements, characterized in that, include: Obtain initial item attribute data and player interaction data, construct a missing topological index based on the player interaction data, and reconstruct the initial item attribute data using topological constraint interpolation and the missing topological index to generate target item attribute data; Generate an attribute topology graph based on the target item attribute data, determine sensitive attribute nodes based on the node connectivity structure of the attribute topology graph, and generate a sensitive node subgraph. The node topology mapping area is determined based on the sensitive node subgraph, and the node topology position is mapped to the attribute topology relationship graph based on the node topology mapping area to generate the item hierarchical topology graph. Based on the item classification topology graph, perform node topology position calibration on the sensitive node subgraph and output the item classification boundary.

2. The intelligent classification method for game elements according to claim 1, characterized in that, The initial item attribute data is reconstructed using topological constraint interpolation and missing topological indices to generate the target item attribute data, including: Read the order in which players choose items during consecutive actions from player interaction data, and record each consecutive item choice as an ordered record to form an item choice sequence; The item selection sequence is combined and split, adjacent item pairs are extracted from the item selection sequence, and records with the same item pair structure are merged to form an item combination set; Based on the order in which each pair of items appears in the item combination set, calculate the sequential connection value of each pair of items, and then aggregate all sequential connection values ​​to form the item connection weight. Construct an item combination network based on the item connection weights, and identify the node positions corresponding to the missing topology indexes in the item combination network to form a set of missing nodes; Based on the attributes of the neighboring nodes of the missing node set in the item combination network, the initial item attribute data is subjected to topological constraint interpolation to generate the target item attribute data.

3. The intelligent classification method for game elements according to claim 2, characterized in that, Based on the order in which each item pair appears in the item combination set, calculate the sequence connection value for each item pair, and then aggregate all sequence connection values ​​to form the item connection weight, including: Read the position of each item pair in the item selection sequence one by one from the item combination set, and record the order of each item pair to form a sequence record set; For items with the same sequential marker in the sequential record set, the records are accumulated to form an item sequence count; Calculate the corresponding connection weight value for each item pair in the item combination set according to the item order count, and collect all connection weight values ​​to form the item connection weight.

4. The intelligent classification method for game elements according to claim 3, characterized in that, Based on the attributes of the neighboring nodes of the missing node set in the item combination network, the initial item attribute data is subjected to topological constraint interpolation to generate the target item attribute data, including: Traverse the set of missing nodes and extract the adjacent item nodes that are directly connected to the positions of each node in the set of missing nodes from the item combination network to form an adjacent node set; Read the known item attributes of each adjacent item node from the adjacent node set, and arrange the read known item attributes according to the node order of the adjacent item nodes to form an adjacent attribute set; Based on the item connection weight, interpolation coefficients are assigned to each known item in the adjacent attribute set, and the interpolation coefficients are combined with the corresponding known items to form an interpolation input set. Perform topological constraint interpolation calculations on the interpolation input set, and write the results of the topological constraint interpolation calculations to the node positions corresponding to the missing topological indexes to generate target item attribute data.

5. The intelligent classification method for game elements according to claim 1, characterized in that, Based on the node connectivity structure of the attribute topology graph, sensitive attribute nodes are identified, and a sensitive node subgraph is generated, including: Read the topological connection information of all nodes from the attribute topology graph and record the connection paths between each node to form a set of node connection paths; The connection paths of nodes in the node connectivity path set are split to divide the attribute topology graph into multiple sub-regions, forming a set of connected sub-regions. Based on the number of nodes contained in each sub-region of the connected sub-region set, sub-regions with a number of nodes exceeding a preset node number threshold are identified and formed as candidate sensitive regions; Extract nodes and their corresponding item attributes from candidate sensitive regions, mark the extracted nodes as sensitive attribute nodes, and generate a sensitive node subgraph based on the connection information of the sensitive attribute nodes in the attribute topology graph.

6. The intelligent classification method for game elements according to claim 5, characterized in that, The connection paths of nodes in the node connectivity path set are processed by connectivity segmentation, dividing the attribute topology graph into multiple sub-regions, forming a set of connected sub-regions, including: Identify and extract node groups with cyclic connection structures from the node connectivity path set, and perform node cyclic cutting processing on the node groups to form initial sub-regions; Calculate the path intersection nodes between the initial sub-region and other nodes, and record the positions of the path intersection nodes to form an intersection node record; The boundary nodes of the initial sub-region are segmented based on the intersection node records to obtain multiple independent sub-regions without intersection nodes, forming a set of connected sub-regions.

7. The intelligent classification method for game elements according to claim 1, characterized in that, Based on the sensitive node subgraph, the node topology mapping region is determined, and based on the node topology mapping region, the node topology position is mapped to the attribute topology relationship graph to generate an item hierarchical topology graph, including: Read the topological position of each sensitive attribute node in the sensitive node subgraph, and extract the positions of non-sensitive nodes connected to the sensitive attribute nodes to form a candidate set of topological mappings; Perform node clustering on the non-sensitive node locations in the candidate set of topology mapping, and merge the regions of non-sensitive nodes with adjacent topological locations to form node mapping regions; Extract the topological region boundary position of the node mapping area, and adjust the node position of the attribute topology relationship graph according to the region boundary position to generate the item hierarchical topology map.

8. The intelligent classification method for game elements according to claim 1, characterized in that, Based on the item grading topology map, perform node topology position calibration on the sensitive node subgraph and output the item grading boundary, including: Extract the topological location information of all sensitive attribute nodes from the prop grading topology map, and record the positions of the adjacent nodes of each sensitive attribute node to form a node calibration input set; Based on the positional differences between sensitive attribute nodes and adjacent nodes in the node calibration input set, the node position of the sensitive node subgraph is corrected to obtain the calibrated sensitive node subgraph. Generate topological boundary lines based on the node positions of the calibration sensitive node subgraph, extract the position data of the topological boundary lines, and output the prop classification boundary.

9. The intelligent classification method for game elements according to claim 8, characterized in that, Based on the positional differences between sensitive attribute nodes and their neighboring nodes in the node calibration input set, the sensitive node subgraph is corrected to obtain a calibrated sensitive node subgraph, including: For each sensitive attribute node in the node calibration input set, calculate the coordinate vector difference between its topological position and that of each neighboring node, and record the coordinate vector difference to form a node difference set; The calibration displacement vector of each sensitive attribute node is calculated based on the node difference set, and the displacement direction and displacement distance are determined. The coordinates of the corresponding nodes in the sensitive node subgraph are updated based on the displacement direction and displacement distance to obtain the calibrated sensitive node subgraph.