A shoe style intelligent matching method and system based on multi-level weight constraint

Abstract

The present application belongs to the technical field of data processing, and particularly relates to a shoe style intelligent matching method and system based on multi-level weight constraint. The method comprises: obtaining user demand data and shoe attribute data; converting the user demand data into a demand weight set containing demand level, action object identifier and weight share; converting the shoe attribute data into a style bearing set containing action object identifier and bearing capacity, and forming a shared capacity group; establishing a bearing channel according to the action object identifier, distributing the weight share according to the level order, distributing the amount according to the bearing capacity and the upper limit of the capacity constraint, locking the occupied bearing capacity and the cumulative bearing capacity layer by layer, and determining the demand residual weight; screening shoes and boots according to the demand residual weight, and generating a shoe matching result. The present application can limit the repeated occupation of attribute bearing capacity, avoid the compensation of low-level matching results for unmet high-level demands, and make the shoe screening result have a clear level basis and residual weight record.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, specifically to a method and system for intelligent matching of footwear styles based on multi-level weight constraints. Background Technology

[0002] Footwear style matching is used to identify footwear from multiple candidate footwear that matches a user's foot condition, usage conditions, appearance preferences, and transaction conditions. Existing technologies typically convert user input into keywords, tags, or feature vectors, and then complete candidate selection based on the similarity between footwear attributes and user features, weighted scores, or ranking results.

[0003] This type of processing often places different types of needs in the same scoring space. There is a lack of clear hierarchical constraints between shoe structure needs, scenario-specific functional needs, style aesthetic needs, and transaction preference needs. This leads to low-level matching results, such as appearance or price, potentially offsetting high-level mismatches such as forefoot space, shoe cavity height, slip resistance, or cushioning. Existing technologies also frequently treat various shoe attributes as independent matching criteria, failing to set unified capacity boundaries for multiple satisfaction capabilities formed by the same shoe structure attribute. This allows the same structural capability to be repeatedly counted by multiple needs, resulting in an overestimation of the satisfiability of candidate shoes. Furthermore, existing weighted scoring typically outputs a total score or ranking position directly, without recording the weights of each need after attribute assignment. Therefore, it is difficult to distinguish between slight overall deviations and severe mismatches in individual items, and it is impossible to reliably determine at which need level a candidate shoe should be excluded. These processing mechanisms make candidate selection results susceptible to changes in weight allocation and the candidate set, and the matching results lack traceable needs-satisfaction relationships. Therefore, it is necessary to address the issues of lack of hierarchical separation among various demands, repeated calculation of footwear attribute capacity, and the inability to form clear screening criteria for unmet demands. Summary of the Invention

[0004] This invention provides a footwear style intelligent matching method and system based on multi-level weight constraints, which is used to at least solve the problem of how to allocate demand weights in a restricted manner and filter footwear based on unassigned weights under the condition that multiple types of user needs have different levels and footwear attribute acceptance capabilities have shared constraints.

[0005] In a first aspect, the present invention provides a footwear style intelligent matching method based on multi-level weight constraints, the method comprising: Obtain user demand data and footwear attribute data; User demand data is converted into a set of demand weights. The demand weight items include the demand level, the target object identifier, and the weight share constrained by the level weight budget. The footwear attribute data is converted into a style carrying set. The style carrying item includes the target identifier and carrying capacity. Some style carrying items form a shared capacity group. Establish a transfer channel from the demand weight item to the style carrying item according to the target object identifier. Allocate weight shares through the transfer channel in hierarchical order. The allocation amount is constrained by the upper limit of the carrying capacity and the shared capacity group. Lock the occupied carrying capacity and the cumulative carrying capacity of the shared capacity group of the style carrying item layer by layer, and determine the unallocated weight share as the demand remaining weight. Footwear is filtered based on the remaining weight of demand, and footwear matching results are generated.

[0006] Secondly, the present invention provides a footwear style intelligent matching system based on multi-level weight constraints, for implementing a footwear style intelligent matching method based on multi-level weight constraints. The system includes: The data acquisition module is used to acquire user demand data and footwear attribute data; The weight construction module is used to convert user demand data into a set of demand weights. The demand weight items include the demand level, the target object identifier, and the weight share constrained by the level weight budget. The carrier construction module is used to convert footwear attribute data into style carrier sets. The style carrier item includes the target object identifier and the carrying capacity. Some style carrier items form a shared capacity group. The weight allocation module is used to establish a transfer channel from the demand weight item to the style carrying item according to the target object identifier. The weight share is allocated through the transfer channel in hierarchical order. The allocation amount is constrained by the upper limit of the carrying capacity and the shared capacity group. The occupied carrying capacity and the cumulative carrying capacity of the shared capacity group of the style carrying item are locked layer by layer, and the unallocated weight share is determined as the demand remaining weight. The matching and filtering module is used to filter footwear according to the remaining weight of the requirements and generate footwear matching results.

[0007] Compared with the prior art, the advantages and beneficial effects of the present invention are as follows: By converting user demand data into a set of demand weights and organizing different demands according to demand hierarchy, target identifier, and weight shares constrained by hierarchical weight budget, a hierarchical expression of multiple demands is achieved in a unified data structure. By converting footwear attribute data into a set of style carrying capacity and setting capacity limits for style carrying items and shared capacity groups, the duplicate carrying capacity of a single footwear attribute and the same structural capacity for multiple demands can be limited. A carrying channel from demand weight items to style carrying items is established through target identifiers, ensuring that weight shares are allocated only between demands and attributes with corresponding relationships. By allocating weight shares hierarchically and locking occupied capacity layer by layer, the matching results of lower-level demands can be prevented from changing the allocation status of higher-level demands. By determining unallocated weight shares as remaining demand weights and filtering footwear accordingly, quantitative retention of unmet demands and layer-by-layer judgment of candidate footwear are achieved. Therefore, footwear matching results can simultaneously reflect demand hierarchy, attribute carrying boundaries, and degree of unmet needs, ensuring that the filtering criteria remain clear, consistent, and traceable. Attached Figure Description

[0008] Figure 1 This is a schematic diagram of the intelligent footwear style matching method based on multi-level weight constraints in an embodiment of the present invention; Figure 2 This is a hierarchical mapping diagram of the demand weight set in an embodiment of the present invention; Figure 3 This is a comparison diagram of the bearing capacity of candidate footwear in the embodiments of the present invention; Figure 4 This is a diagram showing the redistribution relationship of candidate footwear A within the same layer in an embodiment of the present invention; Figure 5 This is a candidate footwear hierarchy screening path diagram in an embodiment of the present invention; Figure 6 This is a block diagram of the module composition of the intelligent footwear style matching system based on multi-level weight constraints in an embodiment of the present invention. Detailed Implementation

[0009] The embodiments of the present invention will be described below with reference to the accompanying drawings. These embodiments are used to illustrate the technical solutions of the present invention and are not intended to limit the scope of protection of the present invention. Those skilled in the art can make adaptive adjustments to the relevant processing order, data structure, or parameter values ​​without changing the technical concept of the present invention.

[0010] In this embodiment, multi-level weight constraints correspond to the organization of weight shares of user demand data at different demand levels, and the hierarchical allocation relationship of weight shares under the capacity upper limit constraints of the carrying capacity and shared capacity group; footwear styles correspond to candidate footwear objects jointly characterized by footwear structural attributes, footwear functional attributes, footwear appearance attributes, and transaction attributes; intelligent matching corresponds to the establishment of the carrying channel between demand weight items and style carrying items, the allocation of weight shares, and the footwear screening process. User demand data comes from the foot conditions, usage conditions, style conditions, and transaction conditions input by the user, and footwear attribute data comes from the structured attribute records of candidate footwear. In this embodiment, the footwear style intelligent matching system based on multi-level weight constraints is referred to as the footwear style intelligent matching system, and the matching processing server is the processing device that performs data processing in the footwear style intelligent matching system. After receiving the above data, the matching processing server forms a demand weight set and a style carrying set, and uses the demand level, target identifier, weight share, carrying capacity, and shared capacity group as the processing basis for subsequent carrying channel establishment, hierarchical capacity locking, determination of remaining demand weights, and generation of footwear matching results.

[0011] like Figure 1 As shown, a footwear style intelligent matching method based on multi-level weight constraints includes the following processing steps.

[0012] In this embodiment, user demand data and footwear attribute data are acquired, wherein the user demand data is used to form a demand weight set, and the footwear attribute data is retained and transferred to the style carrying set generation stage.

[0013] A consistency check is performed on the content, source, and current usage scenario of user requirement data. Data whose target object cannot be determined is excluded from the generation of requirement weight items. Valid user requirement data is aggregated according to requirement hierarchy, a target object identifier is written for each requirement, and the weight share is determined within the hierarchical weight budget of the corresponding requirement level, forming a requirement weight set.

[0014] The demand weight set is passed to the channel establishment stage, and the footwear attribute data is passed to the style carrying set generation stage.

[0015] In one embodiment, user demand data is categorized according to the actual wearer and usage conditions to which the demand pertains. Footwear structural requirements record needs related to foot length, forefoot width, instep capacity, toe box space, heel stability, and sole support; scenario-specific functional requirements record needs related to slip resistance, waterproofing, breathability, warmth, suitability for prolonged standing, suitability for long-distance walking, and sports support; style and aesthetic requirements record needs related to toe box contour, color, material texture, decorative complexity, and matching style; and transaction preference requirements record needs related to price range, brand preference, delivery time, and acceptance of new products.

[0016] When categorizing, the data is based on content that clearly points to the corresponding need within the user demand data. When the same content points to different needs, separate demand weight items are generated for each need to avoid using a single demand weight item for multiple demand objects simultaneously.

[0017] Footwear structural requirements, scenario-based functional requirements, style and aesthetic requirements, and transaction preference requirements are sequentially divided into different requirement levels. These requirement levels determine the order of subsequent weight allocation and the order of level selection. Footwear structural requirements precede scenario-based functional requirements, which in turn precede style and aesthetic requirements, which in turn precede transaction preference requirements.

[0018] When user demand data cannot be clearly categorized into any demand level, no demand weight item is generated, and the uncategorized user demand data is retained for subsequent supplementation and confirmation.

[0019] The target identifier is used to indicate the actual footwear structural attributes, functional attributes, appearance attributes, or transaction attributes affected by each requirement. Forefoot width requirements are written in the forefoot space target identifier; instep accommodation requirements are written in the shoe cavity height target identifier; anti-slip requirements are written in the sole anti-slip target identifier; formality requirements are written in the formal appearance target identifier; and color requirements are written in the color appearance target identifier.

[0020] When generating a demand weight item, the demand level, the target object identifier, and the user demand data associated with the demand content are written into the demand weight item, so that the demand weight item can establish a connection with the style carrier item with the same target object identifier or satisfy the preset function relationship during the channel establishment stage.

[0021] In one embodiment, after generating the demand weights, the basic importance of the demand, the confidence level of the demand evidence, and the scenario reinforcement coefficient are determined for each demand weight. The basic importance of the demand is determined based on the degree to which demand mismatch affects actual wear; higher values ​​are used for footwear structural demands, while lower values ​​are used for style aesthetic demands and transaction preference demands.

[0022] The confidence level of demand evidence is determined based on the directness, consistency, and repetition of user demand data. Higher values ​​are used for direct foot measurement data, consecutive and consistent size change records, and clear reasons for returns, while lower values ​​are used for demands formed by a single browsing or short-term click. When different user demand data conflict with each other, the confidence level of demand evidence is reduced.

[0023] The scenario enhancement coefficient is determined based on the current usage scenario. For commuting in the rain, the scenario enhancement coefficient increases the requirements for anti-slip and waterproofing. For scenarios involving prolonged standing, the scenario enhancement coefficient increases the requirements for cushioning and heel stability.

[0024] The initial weights of each requirement are determined according to the following formula: ; in, For the first The initial weight of each demand weight item Based on the importance of the demand base, To determine the confidence level of the required evidence, This represents the scenario reinforcement coefficient. The confidence level of the requirement evidence is normalized to the range of zero to one, and positive values ​​are used for the basic importance of the requirement and the scenario reinforcement coefficient.

[0025] The weighting percentage within the same demand level is determined according to the following formula: ; in, For the first The weight share of each demand weight item For the first Budget for the hierarchical weights of each demand level. For the first A set of demand weight items within each demand level. This is the index of the demand weight item within the set.

[0026] The hierarchical weight budget is set according to the priority of the demand hierarchy in terms of the constraints on footwear matching results. The sum of all weight shares within the same demand hierarchy equals the hierarchical weight budget. When the sum of the initial weights of all demands within the same demand hierarchy is zero, the allocation of weight shares for the current demand hierarchy is stopped.

[0027] After the allocation is completed, the weight share is written into the demand weight item to form a demand weight set and then transferred to the channel establishment stage.

[0028] In this implementation, the matching processing server receives footwear attribute data and performs field validation and unified dimensional processing on footwear structural attributes, functional attributes, appearance attributes, and transaction attributes. After validation, an object identifier and attribute value are assigned to each attribute, and the carrying capacity is determined based on the data reliability of the attribute value, forming style carrying items. Style carrying items formed by the same footwear structural attribute and jointly consuming the same structural capacity are grouped into a shared capacity group, and the capacity limit is recorded along with the group information. The resulting style carrying set is then transmitted to the receiving channel establishment stage.

[0029] In one embodiment, footwear attribute data is derived from product structured fields, size charts, material parameters, inventory records, and verified reviews. The matching server categorizes footwear attribute data into structural attributes, functional attributes, appearance attributes, and transaction attributes based on how these attributes affect footwear matching. Last width, shoe cavity height, and closure adjustment range are classified as structural attributes; outsole anti-slip properties, outsole cushioning, waterproofing, and breathability are classified as functional attributes; color, material texture, and decorative complexity are classified as appearance attributes; and price range, inventory status, and delivery time are classified as transaction attributes.

[0030] Assign an object identifier to each attribute. The object identifier indicates the target object that the attribute can serve. For example, the shoe last width is assigned an object identifier for the forefoot space, and the sole slip resistance is assigned an object identifier for the sole slip resistance. Attribute values ​​are expressed as numerical values, ranges, or categories. Data from different brand size tables is uniformly converted into comparable physical quantities.

[0031] The data reliability of attribute values ​​is determined based on the reliability of the data source and the consistency among multiple sources. Manufacturer testing data is assigned a higher data reliability, while attribute values ​​inferred from a single evaluation content are assigned a lower data reliability. When multiple sources conflict, the attribute value with the higher reliability is retained, and the corresponding data reliability is reduced. The capacity is determined according to the following formula: ; in, For the first The load-bearing capacity of each style of item. The data credibility of attribute values. For attribute values, This is the normalized carrying capacity function corresponding to the attribute type. For numeric attributes whose carrying capacity monotonically increases with increasing attribute value, the normalized carrying capacity function is determined according to the following formula: ; in, This forms the boundary value for the effective continuation of the attribute. The boundary value where the tolerance capacity no longer increases as the attribute continues to change. For numerical attributes where the tolerance capacity increases as the attribute value decreases, the attribute direction is first reversed before applying the same normalization method; For interval-type attributes, let the effective interval be... When the attribute value is within the valid range, the normalized bearer function is set to one; when the attribute value exceeds the valid range, the normalized bearer function is determined according to the following formula: ; in, The distance between the attribute value and the nearest interval boundary. The maximum distance allowed beyond the effective range while the corresponding attribute still has carrying capacity is determined by historical effective wear records. For categorical attributes, the normalized carrying capacity function is determined as one, a value between one and zero, and zero, respectively, based on completely consistent, compensable, and inconsistent category mapping relationships. The minimum effective value, saturation value, effective range, and category mapping relationship are determined based on historical effective wear records.

[0032] Historical valid wear records refer to records in which user demand data, footwear attribute data, footwear matching results, and wear feedback are completely correlated, and the wear feedback or reasons for returns and exchanges can be verified; historical valid attribute records refer to records in which the attribute data source, original attribute value, and verified attribute value have a complete correspondence.

[0033] In one embodiment, a shared capacity group is used to restrict the same footwear structural attribute from being repeatedly occupied by multiple style carrier items. The matching processing server reads the generation source of each style carrier item. When two or more style carrier items are generated from the same footwear structural attribute, and the corresponding style carrier items consume the same structural capacity when accepting demand weights, these style carrier items are assigned to the same shared capacity group. When the adjustable upper structural attribute simultaneously generates an upper extension style carrier item and a closed adjustment range style carrier item, both of them consume the adjustable upper structural capacity, so they enter the same shared capacity group; color attributes and sole cushioning attributes do not have a common structural source and do not enter the same shared capacity group.

[0034] The upper limit of the shared capacity group is determined based on the available capacity of the corresponding footwear structural attribute, the data reliability of the attribute value, and the proportion that can be shared by multiple needs. Available capacity is determined by the result of processing the corresponding footwear structural attribute according to the normalized load-bearing function. The allowed proportion of shared capacity is determined based on the physical range within which the footwear structural attribute can simultaneously serve multiple needs. The adjustable structural capacity of the upper can only be shared by the upper extension style load-bearing item and the closed adjustment range style load-bearing item within a limited range; therefore, the upper limit of the shared capacity group of the upper is lower than the sum of the capacities of the two style load-bearing items.

[0035] The capacity limit is determined by the following formula: ; in, For the first The maximum capacity of a shared capacity group To allow for the proportion of shared use, To assess the data reliability of footwear structural attributes corresponding to shared capacity groups. This refers to the available capacity of the footwear structural attributes corresponding to the shared capacity group. The allowed proportion of shared occupancy is determined based on the sample proportion of multiple demands being met simultaneously in historical valid wear records, and is limited to a range of zero to one.

[0036] When a style carrying item receives a weight share, the matching processing server synchronously updates the occupied capacity of the style carrying item and the cumulative capacity of the shared capacity group. In this embodiment, the occupied capacity of a style carrying item and the cumulative capacity of the shared capacity group are collectively referred to as the occupied capacity. The occupied capacity of any style carrying item may not exceed its own capacity, and the cumulative capacity of all style carrying items within the same shared capacity group may not exceed the capacity limit. After reaching the capacity limit, the style carrying items within the group stop receiving new weight shares, and the remaining weight shares are retained until the stage of calculating the remaining weight required.

[0037] The matching processing server receives the demand weight set and the style carrying set, reads the target identifier of the demand weight item and the style carrying item, and establishes a receiving channel that can be used to transfer the weight share.

[0038] Once the receiving channel is established, the matching processing server allocates weight shares according to the demand hierarchy, and simultaneously verifies the capacity limits of the receiving capacity and the shared capacity group. When the current demand hierarchy is processed, the occupied capacity is locked, and the unallocated weight shares form the remaining demand weights, which are then passed to the footwear screening stage.

[0039] In one embodiment, the matching processing server reads the requirement content, value range, and scenario associated with the requirement weight item from the user requirement data. Quantifiable requirements are recorded as numerical values ​​or numerical ranges, and non-quantifiable requirements are recorded as category values, forming requirement targets. The attribute values ​​in the footwear attribute data adopt the unified dimensional results from the style carrying item generation stage. If an attribute value is missing, or the data credibility of an attribute value is lower than the minimum data credibility determined by historical valid attribute records, no acceptance channel is established for the corresponding style carrying item. The minimum data credibility is determined according to the lower limit of data credibility in historical valid attribute records that can maintain consistency between the attribute value and the verification result, and is used to exclude attribute values ​​that cannot support the calculation of carrying capacity.

[0040] The pre-defined interaction relationship is determined by the technical connection between the object of demand and the actual object of footwear attribute interaction. When the object identifiers are the same, it indicates that the style-supporting item can directly satisfy the demand represented by the demand weight item. When the object identifiers are different but a clear compensation relationship exists, the pre-defined interaction relationship record can provide the compensated attributes and compensation boundaries. For example, the upper extension attribute can compensate for forefoot space requirements or instep accommodation requirements within a limited range, but it cannot compensate for sole anti-slip requirements. The compensation boundary is determined based on the stable correspondence between attribute changes and demand satisfaction states in historical effective wear records; no connection channel is established when the compensation boundary is exceeded.

[0041] When the demand objective and attribute value use the same unit of measurement, the standardized deviation between the two is calculated; when the demand objective is a category value, a standardized deviation in the range of zero to one is generated based on the degree of category consistency.

[0042] For requirements in numerical or range-based form, the standardized deviation is determined according to the following formula: ; in, This is the distance from the attribute value to the target value or the nearest boundary of the target range. The attribute value must be within the target range. Take zero; For the first The allowable deviation range for each demand weight item is determined by the maximum deviation range in historical valid wear records where the corresponding demand is still satisfied. For demand targets in the form of category values, the standardized deviation is zero when the categories are completely consistent, a value between zero and one is determined according to the category mapping value in the preset action relationship when the compensation relationship is satisfied, and one is taken when the relationship is not satisfied.

[0043] The validity of acceptance is determined according to the following formula: ; in, For the first The demand weight item and the first The effectiveness of the connection between the various styles of load-bearing items The value used to determine the relationship is one if the target identifier of the demand weight item and the style carrying item are the same or if the preset relationship is met; otherwise, it is zero. To determine the confidence level of the required evidence, The data credibility of attribute values. This is the standardized deviation between the target requirement and the attribute value.

[0044] If the target identifiers of the demand weight item and the style carrying item are the same or meet the preset action relationship, the attribute value exists, the data credibility of the attribute value is not lower than the minimum data credibility, and the acceptance validity is greater than zero, the demand weight item, style carrying item, and acceptance validity are recorded to form an acceptance channel and be transferred to the weight share allocation stage; if any condition is not met, no acceptance channel is established.

[0045] In one embodiment, the matching processing server determines the allowable relaxation level and the confidence level of the evidence for each demand weight item. The allowable relaxation level is set according to the degree of impact of the adjusted demand objective on the actual use of the footwear. Demands related to shoe length, forefoot space, and heel stability in footwear structure requirements adopt a lower allowable relaxation level, while aesthetic requirements and transaction preference requirements adopt a higher allowable relaxation level. The lower the allowable relaxation level, the less likely the demand objective is to be taken over by substitute attributes. The confidence level of the evidence for the demand follows the result determined in the demand weight set generation stage. When new user demand data is added, it is updated according to the directness, consistency, and repetition of the data.

[0046] The matching processing server counts the number of channels that can accept each demand weight item. The fewer the number of channels, the fewer styles can be used to accept the corresponding weight share. Within the same demand level, the allocation order is formed in the following order: from low to high allowable relaxation level, from few to many channels, from large to small weight share, and from high to low demand evidence confidence; if the previous item is the same, the next item is used to continue the sorting.

[0047] For the demand weight item, the matching processing server reads the acceptance channels from high to low according to the acceptance validity, and determines the allocation amount according to the following formula: ; in, For the first The demand weight item is directed to the first The weighting of each style's carrying item. The unallocated weight share of the current demand weight item. This represents the remaining load-bearing capacity of the current style. This represents the remaining capacity of the shared capacity group to which the current style's carrying item belongs. If the style's carrying item does not belong to a shared capacity group, the remaining capacity of the shared capacity group is not included in the value calculation.

[0048] After allocation is completed, the unallocated weight share, remaining capacity, and remaining capacity of the shared capacity group are deducted simultaneously. When the current demand level completes allocation, the occupied capacity is locked, and the unallocated weight share forms the remaining demand weight.

[0049] In one embodiment, after the matching processing server completes the initial allocation at the same demand level, it checks whether there are still unallocated weight shares for each demand weight item. If any demand weight item still has unallocated weight shares, and all style carrying items connected to all receiving channels have no remaining carrying capacity, the server reads the occupied carrying capacity record of the corresponding style carrying item to determine another demand weight item occupying the carrying capacity. The occupied capacity already locked at higher demand levels does not participate in the reallocation, preventing the current demand level from changing the allocation result of higher demand levels.

[0050] The matching server searches for alternative styles with remaining capacity from the remaining channels of another demand weight item, and verifies the remaining capacity of the corresponding shared capacity group. Preset alternative conditions are used to limit the decrease in capacity effectiveness caused by reallocation, expressed as follows: ; in, The effectiveness of the connection between another demand weight item and the alternative style carrying item. The effectiveness of the connection between another demand weight item and the current style carrying item. This is the allowable decrease in acceptance effectiveness. The allowable decrease in acceptance effectiveness is determined based on the maximum decrease in acceptance effectiveness in historical valid wear records, where the redistributed tier surplus rate does not exceed the corresponding tier surplus rate threshold and the maximum single item surplus rate does not exceed the corresponding single item surplus rate threshold.

[0051] When multiple alternative style carrying items exist, they are selected in descending order of carrying effectiveness. The reallocation amount is determined based on the weight share already allocated to another demand weight item, the remaining carrying capacity of the alternative style carrying items, and the remaining capacity of the shared capacity group. After the reallocation is completed, an equal amount of carrying capacity for the current style carrying item is released. The released carrying capacity is used to allocate the unallocated weight share of any demand weight item, and the occupied capacity is updated synchronously.

[0052] If no alternative style meets the preset replacement conditions, or if sufficient capacity cannot be released after reallocation, the unallocated weight share is retained. The current demand level performs only one complete check. After the check, the updated occupied capacity is locked, and the unallocated weight share is passed to the footwear screening stage as the remaining demand weight.

[0053] The intelligent footwear matching system receives the remaining weights of each footwear item at different demand levels and performs filtering in tiered order. The matching and filtering module aggregates the remaining weights within the current demand level, determines the tier remaining rate based on the tier weight budget, and determines the maximum single-item remaining rate based on the ratio of each remaining weight to its respective weight share. Only footwear items whose tier remaining rate and maximum single-item remaining rate both meet the corresponding thresholds proceed to the next demand level; after filtering across all demand levels, the results are retained to form the footwear matching results.

[0054] In one embodiment, the matching and filtering module takes footwear as the processing object and reads the remaining weight of demand within the demand hierarchy, the weight share of each demand weight item, and the hierarchy weight budget. The hierarchy remaining rate is determined according to the following formula: ; in, For the level of residual rate, For the currently selected footwear, For the requirement level sequence number, For the first A set of demand weight items within each demand level. This is the index of the demand weight item. For footwear Corresponding to the The remaining weight of each demand weight item For the first The hierarchical weight budget for each demand level. The hierarchical surplus rate represents the proportion of unallocated weight shares within a demand level relative to the hierarchical weight budget.

[0055] The maximum single-item residual rate is determined by the following formula: ; in, The maximum single-item residual rate, For the first The weight share of each demand weight item. The maximum single item surplus rate is used to identify the unallocated proportion of a single demand weight item, preventing single item mismatch from being masked by other demand weight items within the same demand level.

[0056] The tiered surplus rate threshold and the individual surplus rate threshold are determined based on the demand surplus weight distribution of acceptable and unacceptable footwear in historical valid wear records, respectively limiting the overall unallocated proportion and the individual unallocated proportion of the current demand tier. Footwear is retained to the next demand tier if the tiered surplus rate is not greater than the tiered surplus rate threshold and the maximum individual surplus rate is not greater than the individual surplus rate threshold; if either condition is not met, the footwear does not enter the next demand tier.

[0057] The screening process is performed in the order of footwear structural requirements, scenario functional requirements, style aesthetic requirements, and transaction preference requirements. Footwear with missing or zero weight budgets at each level will not be included in the footwear matching results; requirements with a weight share of zero will not participate in the calculation of the maximum single-item surplus rate. After completing the screening of all requirement levels, the retained footwear will form the footwear matching results.

[0058] In one embodiment, the footwear style intelligent matching system performs footwear style matching for user sample U-20260425-017. The data acquisition module receives user demand data: forefoot width 98 mm, instep height 58 mm, usage scenario: commuting in the rain, continuous standing time 6 hours, style requirements: dark color and low embellishment, price range 450-650 yuan, and delivery time limit 48 hours.

[0059] The data acquisition module reads footwear attribute data from the footwear product database for four candidate footwear items. These four candidate footwear items are designated as Candidate A, Candidate B, Candidate C, and Candidate D, with corresponding sample numbers S-A01, S-B02, S-C03, and S-D04, respectively. The footwear attribute data originates from footwear size charts, material parameters, structural parameters, functional test records, and transaction records. Different data sources are aggregated using a unified object identifier, and numerical attributes are converted to a unified dimension. Attribute values ​​that are missing or whose source cannot be confirmed are not used to determine the carrying capacity.

[0060] The weighting module determines footwear structural requirements, scenario-specific functional requirements, style and aesthetic requirements, and transaction preference requirements based on user demand data. Footwear structural requirements correspond to forefoot space, shoe cavity height, and heel stability; scenario-specific functional requirements correspond to outsole anti-slip and cushioning for prolonged standing; style and aesthetic requirements correspond to dark-colored appearance and minimal embellishments; and transaction preference requirements correspond to price range and delivery time.

[0061] The hierarchical weight budget for footwear structural requirements is set at 0.40, the hierarchical weight budget for scene function requirements is set at 0.30, the hierarchical weight budget for style aesthetic requirements is set at 0.20, and the hierarchical weight budget for transaction preference requirements is set at 0.10. The hierarchical weight budget is determined based on the degree of impact of mismatch between each requirement on actual wear, so that footwear structural requirements and scene function requirements participate in footwear selection before style aesthetic requirements and transaction preference requirements.

[0062] The basic importance of the demand, the confidence level of the demand evidence, and the scenario reinforcement coefficient for the forefoot space are 0.95, 0.90, and 1.10, respectively, with an initial weight of 0.95 × 0.90 × 1.10, or 0.9405. The three parameters for the shoe cavity height are 0.85, 0.80, and 1.00, with an initial weight of 0.6800. The three parameters for heel stability are 0.80, 0.75, and 1.05, with an initial weight of 0.6300.

[0063] The sum of the initial weights for the requirements within the footwear structure is 2.2505. The weight share for forefoot space is 0.9405 ÷ 2.2505 × 0.40, recorded as 0.1672 to four decimal places; the weight share for shoe cavity height is 0.6800 ÷ 2.2505 × 0.40, recorded as 0.1209; and the weight share for heel stability is 0.6300 ÷ 2.2505 × 0.40, recorded as 0.1120.

[0064] The scenario enhancement coefficient for rainy commuting increases the anti-slip properties of shoe soles, while the scenario enhancement coefficient for prolonged standing (6 hours) increases the cushioning performance. The initial weights for anti-slip properties and prolonged standing cushioning are 0.9180 and 0.7820, respectively. After allocation within a tiered weight budget of 0.30, their respective weight shares are 0.1620 and 0.1380.

[0065] The initial weights for demand for dark exteriors and low trim levels are 0.4125 and 0.2925, respectively. After allocation within a tiered weight budget of 0.20, their respective weight shares are 0.1170 and 0.0830. The initial weights for demand for price ranges and delivery time limits are 0.4500 and 0.3200, respectively. After allocation within a tiered weight budget of 0.10, their respective weight shares are 0.0584 and 0.0416.

[0066] In this embodiment, the weight share, capacity, allocation amount, demand remaining weight, hierarchical remaining rate, and maximum single item remaining rate used for text and figure display are all retained to four decimal places; the matching processing server uses unrounded values ​​in hierarchical weight budget constraints, capacity constraints, shared capacity group capacity upper limit constraints, and ratio calculations, and the difference in the last digit of the displayed value due to rounding is not used to determine whether the hierarchical weight budget or capacity upper limit is exceeded.

[0067] like Figure 2 As shown, the user demand data is processed by the weight construction module to form nine demand weight items. Figure 2 D1 to D9 in the table are the demand weight item numbers, used only to distinguish different demand weight items. Each demand level corresponds to a level weight budget, and each demand weight item records the target identifier and weight share. Figure 2 The left side shows the user demand data for user sample U-20260425-017, and the right side shows the set of demand weights organized according to demand hierarchy.

[0068] The load-bearing construction module converts the footwear attribute data of the four candidate footwear into a style load-bearing set. Footwear structural attributes are used to determine the load-bearing capacity corresponding to forefoot space, shoe cavity height, and heel stability, while footwear functional attributes are used to determine the load-bearing capacity corresponding to sole anti-slip and long-term standing cushioning.

[0069] Candidate A has the following load capacities: forefoot space, shoe cavity height, heel stability, outsole slip resistance, and cushioning for prolonged standing: 0.1672, 0.1209, 0.1050, 0.1620, and 0.1080, respectively. Candidate B has the following load capacities: 0.0772, 0.0709, 0.1120, 0.1620, and 0.1200, respectively. Candidate C has the following load capacities: 0.1672, 0.1129, 0.0900, 0.0800, and 0.1200, respectively. Candidate D has the following load capacities: 0.1372, 0.1009, 0.0920, 0.1000, and 0.1380, respectively.

[0070] In this embodiment, the shared capacity group formed by the structural attributes of the upper is denoted as the upper shared capacity group, and the shared capacity group formed by the structural attributes of the sole is denoted as the sole shared capacity group. Both the upper shared capacity group and the sole shared capacity group belong to the shared capacity group.

[0071] Candidate A's upper extension style load-bearing item C3 and closure adjustment range style load-bearing item C5 are both generated from the adjustable structural properties of the upper and jointly consume the adjustable structural capacity of the upper. Candidate A's sole anti-slip style load-bearing item and long-duration cushioning style load-bearing item both use the composite structural properties of the sole as their common source of generation. The load-bearing capacity is determined by combining the sole anti-slip properties and the sole cushioning properties, respectively, and they jointly consume the composite structural capacity of the sole. Both the adjustable structural properties of the upper and the composite structural properties of the sole belong to footwear structural properties.

[0072] The upper capacity limits for the shared capacity groups of the uppers for candidates A, B, C, and D are 0.0750, 0.1480, 0.2800, and 0.2380, respectively, while the upper capacity limits for the shared capacity groups of the soles are 0.2700, 0.2820, 0.2000, and 0.2200, respectively. These capacity limits are determined based on the footwear structural attributes forming the shared capacity groups, the data reliability of the attribute values, and the range within which multiple style-related elements can share the same structural capacity.

[0073] like Figure 3 As shown, the vertical axis lists the forefoot space, shoe cavity height, heel stability, outsole anti-slip, long-duration cushioning, upper shared capacity group, and outsole shared capacity group in sequence. The horizontal axis represents the carrying capacity or upper limit of the capacity. The four horizontal bars correspond to candidates A to D respectively. Figure 3 This is used to compare the ability of different candidate footwear to meet the structural and functional requirements of footwear in different scenarios, as well as the limitations of the shared capacity group on the cumulative allocation. The results of meeting aesthetic and transaction preference requirements participate in subsequent levels of screening, but are not included. Figure 3 The image is drawn repeatedly in the middle.

[0074] The weight allocation module establishes a connection channel for candidate A. The forefoot space requirement weight item D1 establishes a connection channel with the basic forefoot width style support item C1, the upper extension style support item C3, and the toe shape compensation style support item C4. The shoe cavity height requirement weight item D2 establishes a connection channel with the shoe cavity height style support item C2, the upper extension style support item C3, and the closure adjustment range style support item C5.

[0075] C1 to C5 are style-specific item numbers, used only for distinguishing objects in the attached drawings. The validity of the acceptance channel is determined based on the deviation between the demand target and the attribute value, the confidence level of the demand evidence, and the data credibility of the attribute value. Figure 4 The values ​​in C1, C2, C4 and C5, and the two values ​​in C3, respectively represent the acceptance validity and actual allocation of the corresponding receiving channels.

[0076] The succession validity between D1 and C1 is 0.94, with 0.1300 allocated to C1. The succession validity between D1 and C3 is 0.74, with 0.0372 initially allocated to C3. The succession validity between D2 and C2 is 0.88, with 0.0800 allocated to C2; the succession validity between D2 and C3 is 0.84, with 0.0228 initially allocated to C3; the succession validity between D2 and C5 is 0.79, with 0.0150 allocated to C5.

[0077] Upon completion of the initial allocation, C3 had accumulated 0.0600, C5 had accumulated 0.0150, and the total accumulated capacity of the shared capacity group for shoe uppers reached the capacity limit of 0.0750. D2 still had 0.0031 unallocated weight share.

[0078] The weight allocation module checks the remaining connection channels of D1 and determines that C4 has remaining connection capacity. The connection effectiveness between D1 and C4 is 0.71, which differs from the connection effectiveness between D1 and C3 by 0.03. This difference does not exceed the allowable decrease in connection effectiveness of 0.05 in the preset replacement conditions. The weight allocation module reallocates 0.0031 of the connection capacity of D1 that was already allocated to C3 to C4, adjusting the weight share of D1 allocated to C3 from 0.0372 to 0.0341. It also allocates the 0.0031 of connection capacity released from C3 to D2, adjusting the weight share of D2 allocated to C3 from 0.0228 to 0.0259.

[0079] After the reassignment within the same layer, C3 has accumulated a total of 0.0341 and 0.0259, which is 0.0600. C5 has accumulated 0.0150. The total accumulated capacity of the shared capacity group for the shoe upper remains at 0.0750, which does not exceed the capacity limit. Neither D1 nor D2 has any remaining demand weight.

[0080] like Figure 4 As shown, the D1 receiving channel corresponds to the forefoot space requirement weight item, the D2 receiving channel corresponds to the shoe cavity height requirement weight item, and the dashed lines from D1 to C4 represent the receiving channels formed by the redistribution within the same layer. Figure 4 The shared capacity groups on the right side correspond to C3 and C5, indicating that C3 and C5 are both subject to the upper capacity limit of the shared capacity group. C1, C2, and C4 are not included in the cumulative capacity of this shared capacity group.

[0081] Candidate A's heel stability design has a load-bearing capacity of 0.1050, resulting in a remaining demand weight of 0.0070. The outsole anti-slip design has a load-bearing capacity of 0.1620, with no remaining demand weight. The long-term cushioning design has a load-bearing capacity of 0.1080, resulting in a remaining demand weight of 0.0300. The combined load-bearing capacity of outsole anti-slip and long-term cushioning is 0.2700, reaching the capacity limit of the shared capacity group for candidate A's outsole.

[0082] Candidate A's dark exterior weight share of 0.1170 was fully absorbed, while the low trim received 0.0700, resulting in a residual demand weight of 0.0130. The price range weight share of 0.0584 and the delivery time limit weight share of 0.0416 were both fully absorbed.

[0083] After the demand for footwear structure is allocated, the occupied capacity is locked. The sum of the remaining weights of the demand is 0.0070, the level remaining rate is 0.0070 ÷ 0.40, which is 0.0175, and the maximum single item remaining rate is 0.0070 ÷ 0.1120, which is 0.0625.

[0084] The sum of the residual weights for scenario function requirements is 0.0300, the hierarchical residual rate is 0.0300 ÷ 0.30, which is 0.1000, and the maximum single-item residual rate is 0.0300 ÷ 0.1380, recorded as 0.2174. The sum of the residual weights for style and aesthetic requirements is 0.0130, the hierarchical residual rate is 0.0130 ÷ 0.20, which is 0.0650, and the maximum single-item residual rate is 0.0130 ÷ 0.0830, recorded as 0.1566. Transaction preference requirements have no residual weights; both the hierarchical residual rate and the maximum single-item residual rate are 0.

[0085] The matching and filtering module sets the hierarchical residual rate threshold for footwear structural requirements to 0.12 and the individual residual rate threshold to 0.25; the hierarchical residual rate threshold for scenario functional requirements to 0.15 and the individual residual rate threshold to 0.25; and the hierarchical residual rate threshold for style aesthetic requirements and transaction preference requirements to 0.20 and the individual residual rate threshold to 0.30. Each threshold is determined based on the distribution of residual demand weights between acceptable footwear and footwear that exhibits issues such as pinching, insufficient stability, inadequate slip resistance, or discomfort during prolonged standing, based on historical valid wearing records.

[0086] Candidate B has a residual weight of 0.1400 within the footwear structure requirement, a hierarchical residual rate of 0.3500, and a maximum single-item residual rate of 0.5384, and did not enter the scenario functional requirement screening. Candidate C has a hierarchical residual rate of 0.0750 within the footwear structure requirement, and a maximum single-item residual rate of 0.1964, and entered the scenario functional requirement screening; however, Candidate C also has a hierarchical residual rate of 0.3333 within the scenario functional requirement, and a maximum single-item residual rate of 0.5062, and did not enter the style aesthetic requirement screening. Candidate D has a hierarchical residual rate of 0.1750 within the footwear structure requirement, and a maximum single-item residual rate of 0.1795, and was eliminated because its hierarchical residual rate exceeded the hierarchical residual rate threshold.

[0087] like Figure 5As shown in the figure, U represents the hierarchical surplus rate, and M represents the maximum single-item surplus rate. U and M are only used as reference numerals in the figure. Candidate A is filtered sequentially based on footwear structural requirements, scenario functional requirements, style aesthetic requirements, and transaction preference requirements; Candidates B and D are eliminated based on footwear structural requirements; and Candidate C is eliminated based on scenario functional requirements.

[0088] After completing the screening of all requirement levels, the matching and screening module writes the sample number S-A01 corresponding to candidate A into the footwear matching results. The footwear matching results also retain the remaining requirement weights corresponding to heel stability (0.0070), long-term standing cushioning (0.0300), and low embellishment (0.0130), used to indicate specific requirements that candidate A has not yet fully addressed. The results of addressing lower requirement levels are not used to offset the remaining requirement weights in footwear structural requirements or scenario functional requirements.

[0089] like Figure 6 As shown, a footwear style intelligent matching system based on multi-level weight constraints includes the following modules.

[0090] Figure 6 Each module in the matching processing server is implemented collaboratively by its processor, memory, communication interface, and result output interface. The memory stores the program instructions and processing data used to implement the functions of each module. The processor executes the program instructions to complete the data processing corresponding to each module. Each module can share the processor and memory.

[0091] The data acquisition module is implemented by a communication interface, a processor, and a memory. The communication interface receives user requirement data and footwear attribute data, the processor performs integrity verification and format standardization on the received data, and the memory temporarily stores the verified data.

[0092] The weight construction module is implemented by a processor and a memory. The memory stores the demand hierarchy, the target object identifier, and the hierarchy weight budget, while the processor determines the weight share and generates the demand weight set.

[0093] The carrier construction module is implemented by a processor and a memory. The memory stores footwear attribute data and shared capacity group configuration, while the processor determines the target object identifier, the carrying capacity, and the shared capacity group, and generates a style carrier set.

[0094] The weight allocation module is implemented by a processor and a memory. The processor establishes the receiving channel, performs weight share allocation under capacity constraints, and locks the occupied capacity layer by layer. The memory stores the receiving channel, the occupied capacity, and the remaining weight required.

[0095] The matching and filtering module is implemented by a processor, a memory, and a result output interface. The processor performs filtering based on the remaining weights required and generates footwear matching results. The memory stores the filtering process data, and the result output interface outputs the footwear matching results.

[0096] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0097] The various embodiments in this invention are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the system embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions in the method embodiments.

Claims

1. A footwear style intelligent matching method based on multi-level weight constraints, characterized in that, The method includes: Obtain user demand data and footwear attribute data; The user demand data is converted into a demand weight set, and the demand weight item includes the demand level, the target object identifier, and the weight share constrained by the level weight budget. The footwear attribute data is converted into a style carrying set. The style carrying item includes the target identifier and carrying capacity. Some of the style carrying items form a shared capacity group. Establish a receiving channel from the demand weight item to the style carrying item according to the target identifier, allocate the weight share through the receiving channel in hierarchical order, the allocation amount is constrained by the receiving capacity and the capacity limit of the shared capacity group, lock the occupied receiving capacity of the style carrying item and the cumulative receiving amount of the shared capacity group layer by layer, and determine the unallocated weight share as the demand remaining weight. Footwear is filtered based on the remaining weights according to the stated requirements, and footwear matching results are generated.

2. The method according to claim 1, characterized in that, The step of converting the user demand data into a demand weight set includes: determining footwear structural demands, scenario functional demands, style aesthetic demands, and transaction preference demands based on the user demand data; dividing the footwear structural demands, scenario functional demands, style aesthetic demands, and transaction preference demands into different demand levels in sequence, determining the target identifier of each demand, and generating the demand weight item.

3. The method according to claim 2, characterized in that, The determination of the weight share includes: determining the basic importance of the demand, the confidence level of the evidence for the demand, and the scenario reinforcement coefficient for each demand weight item based on the user demand data. The basic importance of the demand is used to indicate the degree of impact of demand mismatch on actual footwear wearing, the confidence level of the evidence for the demand is used to indicate the degree of support of the user demand data for the corresponding demand, and the scenario reinforcement coefficient is used to indicate the degree of adjustment of the corresponding demand weight by the current usage scenario; the product of the basic importance of the demand, the confidence level of the evidence for the demand, and the scenario reinforcement coefficient is determined as the initial weight of each demand; the hierarchical weight budget is allocated according to the proportion of each initial demand weight within the same demand level to obtain the weight share.

4. The method according to claim 1, characterized in that, The step of converting the footwear attribute data into a style carrying set includes: determining footwear structural attributes, footwear functional attributes, footwear appearance attributes, and transaction attributes from the footwear attribute data; determining the target identifier and attribute value of each attribute to generate the style carrying item; determining the data credibility of the attribute value based on the reliability of the data source of the attribute value and the consistency between multiple data sources; and determining the carrying capacity based on the attribute value and the data credibility.

5. The method according to claim 4, characterized in that, The determination of the shared capacity group includes: when multiple style carrying items are generated by the same footwear structural attribute, classifying the multiple style carrying items into the same shared capacity group, and determining the upper limit of the capacity of the shared capacity group according to the footwear structural attribute; so that the cumulative weight share of each style carrying item in the shared capacity group does not exceed the upper limit of the capacity.

6. The method according to claim 1, characterized in that, The step of establishing a connection channel from the demand weight item to the style carrying item based on the target identifier includes: determining the demand target of the demand weight item based on the user demand data, and determining the attribute value of the style carrying item based on the footwear attribute data; determining a preset interaction relationship, which is used to indicate the satisfaction or compensation relationship between the attribute of the style carrying item and the demand represented by the demand weight item; when the target identifier of the demand weight item and the style carrying item are the same or the preset interaction relationship is satisfied, determining the connection validity based on the deviation between the demand target and the attribute value, and establishing the connection channel when the connection validity is greater than zero.

7. The method according to claim 6, characterized in that, The allocation of the weight share through the receiving channel includes: determining the allowable relaxation level and demand evidence confidence level of each demand weight item based on the user demand data, wherein the allowable relaxation level indicates the degree to which the demand target can be adjusted, and the demand evidence confidence level indicates the degree of support of the user demand data for the demand represented by the demand weight item; counting the number of receiving channels for each demand weight item; within the same demand level, determining the allocation order of demand weight items according to the allowable relaxation level from low to high, the number of receiving channels from few to many, the weight share from large to small, and the demand evidence confidence level from high to low; and allocating the weight share to style carrying items that have remaining receiving capacity and meet the capacity upper limit constraint of the shared capacity group according to the allocation order and the receiving effectiveness from high to low.

8. The method according to claim 7, characterized in that, If any demand weight item within the same demand level has an unallocated weight share, and all style carrying items with a receiving channel to the aforementioned demand weight item have no remaining carrying capacity, then another demand weight item is identified to occupy the carrying capacity of any of the style carrying items, and the occupied style carrying item is designated as the current style carrying item. If there is an alternative style carrying item with a receiving channel to the other demand weight item, has remaining carrying capacity, and satisfies the capacity upper limit constraint of the shared capacity group, and the decrease in the carrying effectiveness of the alternative style carrying item relative to the carrying effectiveness of the current style carrying item is not greater than the allowed decrease in carrying effectiveness, then the weight share of the other demand weight item already allocated to the current style carrying item is reallocated to the alternative style carrying item, and the released carrying capacity is used to allocate the unallocated weight share of the aforementioned demand weight item.

9. The method according to claim 1, characterized in that, The step of filtering footwear according to the remaining demand weights includes: determining the ratio of the sum of the remaining demand weights within the same demand level to the level weight budget as the level remaining rate; determining the maximum value of the ratio of the remaining demand weights within the same demand level to their respective weight shares as the maximum single-item remaining rate; and, according to the level order, if the level remaining rate of any footwear at the current demand level is not greater than the level remaining rate threshold corresponding to the current demand level, and the maximum single-item remaining rate is not greater than the single-item remaining rate threshold corresponding to the current demand level, then retaining any footwear to the next demand level for filtering, until all demand levels are completed, and obtaining the footwear matching result.

10. A footwear style intelligent matching system based on multi-level weight constraints, used to implement the footwear style intelligent matching method based on multi-level weight constraints as described in any one of claims 1 to 9, characterized in that, The system includes: The data acquisition module is used to acquire user demand data and footwear attribute data; The weight construction module is used to convert the user demand data into a set of demand weights. The demand weight items include the demand level, the target object identifier, and the weight share constrained by the level weight budget. The carrier construction module is used to convert the footwear attribute data into a style carrier set. The style carrier item includes the target identifier and the carrying capacity. Some of the style carrier items form a shared capacity group. The weight allocation module is used to establish a receiving channel from the demand weight item to the style carrying item according to the target object identifier, allocate the weight share through the receiving channel in hierarchical order, the allocation amount is constrained by the receiving capacity and the capacity limit of the shared capacity group, lock the occupied receiving capacity of the style carrying item and the cumulative receiving amount of the shared capacity group layer by layer, and determine the unallocated weight share as the demand remaining weight. The matching and filtering module is used to filter footwear according to the remaining weight of the requirements and generate footwear matching results.

Images