Virtual commodity intelligent uploading and management system for financial institution credit mall

Abstract

The application discloses a virtual commodity intelligent shelving and management system for a financial institution credit mall, relates to the technical field of virtual commodity intelligent shelving and management, and is aimed at the sudden shortage of a single product in a combination product, automatically screens the best substitute based on a dynamically constructed equity relation graph, and performs real-time rule conflict detection through graph weight fusion, so that the substitute scheme can meet the user preference and the cost limit of the financial institution at the same time, compared with the existing scheme that needs manual checking of the inventory and rules, the system can compress the processing time limit from hours to seconds, and completely avoid customer complaints caused by manual errors; meanwhile, a double protection mechanism through an intelligent compliance fuse is introduced, the matching degree of a user risk label and a KYC strategy is verified in the commodity shelving stage, and high-risk exchange is intercepted.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of virtual commodity intelligent shelving and management, in particular to a virtual commodity intelligent shelving and management system for a financial institution credit mall. BACKGROUND

[0002] Currently, the financial institution credit mall generally uses virtual commodity combination packages to improve user activity, such as mixed benefits of video membership and travel coupons, and such combination commodities generally interface with real-time inventories of multiple third-party platforms.

[0003] When a single product in the combination commodity suddenly fails, such as when a certain ride coupon inventory is zero, the system mostly cancels the entire order and refunds, but this will cause the user to lose the combination discount and trigger complaints, or manual intervention to replace the single product, but this is time-consuming and prone to errors; in addition, some replacement solutions only rely on simple rules, such as when a user exchanges a video season card and a certain ride coupon, if the certain ride coupon is out of stock, but the system will mistakenly replace it with a ride coupon from another platform plus a music software monthly card, which violates the user's needs and causes the institution to overspend; and when the rules of cross-platform commodity benefits conflict, such as video membership being limited to new users, this will trigger a second failure.

[0004] To solve the above problems, some solutions use a dynamic decision engine to optimize the process, pre-load a replacement list based on user profiles and historical behavior, and introduce real-time compliance verification; for example, when a certain ride coupon is detected to be out of stock, the system automatically matches other platform ride coupons commonly used by the user as a replacement, while verifying whether the user meets the new customer rules of a certain platform; but such solutions still have obvious shortcomings: they rely on manual pre-set replacement relationships and are difficult to cover a large number of commodity combinations; and they do not consider institutional cost constraints, which may result in losses due to replacing high-priced commodities; and they cannot perceive changes in hidden rules of third-party platforms, such as sudden restrictions on exchange areas; these defects force the institution to still retain a manual review channel, which drags down processing efficiency. SUMMARY

[0005] In view of the above existing problems, the present application is proposed.

[0006] The present application provides a virtual commodity intelligent shelving and management system for a financial institution credit mall to solve the problem that existing solutions rely on manual replacement or entire order refunds when a single product in a combination commodity fails, which triggers user complaints and uncontrolled costs.

[0007] To solve the above technical problems, the present application provides the following technical solutions:

[0008] The present application provides a virtual commodity intelligent shelving and management system for a financial institution credit mall, which includes,

[0009] Commodity atomization packaging module, used for disassembling third-party virtual commodities into minimum equity units, and generating unique blockchain storage identification with digital watermark of a financial institution;

[0010] Cross-platform rule library, dynamically accessing financial institution compliance strategies and third-party platform commodity rules;

[0011] Dynamic equity replacement engine, responding to combination commodity single product failure events, and outputting replacement schemes based on an equity relationship graph;

[0012] Intelligent compliance fuse, real-time calling of a rule library for operation interception.

[0013] As a preferred scheme of the virtual commodity intelligent listing and management system for the financial institution credit mall, the dynamic equity replacement engine comprises:

[0014] Equity relationship graph library, storing a network structure with commodities as nodes and replacement relationships as edges, and dynamically calculating edge weights;

[0015] Real-time inventory monitoring unit, interacting with a third-party platform API through a polling mechanism, updating inventory states and marking abnormal thresholds every 15 seconds.

[0016] As a preferred scheme of the virtual commodity intelligent listing and management system for the financial institution credit mall, in the equity relationship graph library, the step of dynamically calculating edge weights comprises:

[0017] When a user exchange behavior data set D t satisfies ΔD=KL(D t ||D t-1 )>θ, weight updating is triggered, D t represents a user exchange behavior data set at time t, containing a user ID, a commodity ID, and a redemption timestamp triplet, KL(p||q) represents a Kullback-Leibler divergence, calculating a difference amount of behavior distributions of two continuous time windows, θ=0.15 represents a threshold value calibrated through historical data mutation detection, and the unit is information entropy bits;

[0018] A three-dimensional weight calculation model is constructed, and a model formula is represented as:

[0019]

[0020] wherein, represents a replacement weight of commodity a→b at time t, is dimensionless, and the range is [0, 1], α represents a semantic similarity weight coefficient, and the constraint condition is 0.2≤α≤0.5, S ab represents a semantic similarity of commodities a and b, β represents a co-redemption probability weight coefficient, and the constraint condition is 0.2≤β≤0.5, The formula represents the co-exchange probability of commodity a→b at time t, and γ represents the rule compatibility weight coefficient, which is automatically calculated by γ = 1-α-β,

[0021] The semantic similarity component in the formula is represented as:

[0022]

[0023] wherein, The BERT word vector of the title of commodity a is represented as: The knowledge graph embedding vector of commodity a is represented as:

[0024] ||·||2 represents the L2 norm calculator, and G represents the set of all commodity nodes in the current graph,

[0025] The co-exchange probability component in the formula is represented as:

[0026]

[0027] wherein, σ(x) = 1 / (1+e -x ) is a sigmoid function that maps values to (0,1), N u represents the total number of active users in the last 30 days, and a user is defined as having completed at least one exchange, represents the decay factor of user k, n k is the number of exchanges of user k in the last 7 days, represents the indicator function, which is 1 when a and b are both included in the exchange record of user k, and 0 otherwise, t k is the time of the last exchange of a or b by user k, in Unix timestamp format, t represents the timestamp of the current calculation time, and λ = 0.05 represents the time decay coefficient, in seconds -1 ,

[0028] The rule compatibility component in the formula is represented as:

[0029]

[0030] wherein, R a represents the set of rule clauses applicable to commodity a, and the elements are cross-platform rule library item IDs, is the rule conflict coefficient, n is the total number of related clauses in the cross-platform rule library, w k is the severity weight of clause k, δ k ∈{0,1} indicates whether clause k is violated when a→b is replaced, c a represents the unit cost of commodity a for the financial institution to settle with the supplier, in yuan, and the denominator 5 and 0.9 represent the S-curve steepness coefficient and cost proportion threshold, which are determined by fitting historical financial data.

[0031] As a preferred scheme of the virtual commodity intelligent listing and management system for the financial institution credit mall, wherein the construction of the interest relationship graph library comprises:

[0032] The node is associated with the user historical behavior label, including the consumption frequency and the category preference;

[0033] The edge weight calculation fuses the following dimensions:

[0034] The cross-platform rule conflict coefficient is 0-1 scale, the user group substitute acceptance, and the financial institution preset cost constraint function.

[0035] As a preferred scheme of the virtual commodity intelligent listing and management system for the financial institution credit mall, wherein in the construction process of the interest relationship graph library, the construction step of the cost constraint function comprises:

[0036] Obtaining the original commodity cost c o , the candidate substitute cost c c , and the financial institution floating coefficient δ, δ∈[0.05, 0.2];

[0037] Calculating the absolute cost deviation Δ c =|c c -c o | and the relative cost deviation

[0038] The fuse determination function is calculated as:

[0039]

[0040] Wherein, c o represents the unit cost of the original commodity, unit yuan c c represents the unit cost of the candidate substitute, unit yuan, δ represents the preset floating coefficient of the financial institution, dimensionless, and is calibrated through historical financial data, η c represents the relative cost deviation, dimensionless, Γ(c c , c o ) is the cost adaptation degree function, and ∨ represents the operator logical or;

[0041] In the formula, the cost adaptation degree function is represented as:

[0042]

[0043] Wherein, k=12 represents the relative deviation sensitivity coefficient, μ=0.08 is the absolute cost safety coefficient, unit: yuan -1 ;

[0044] represents the market average price of similar commodities, ci The settlement cost of the ith same kind of commodity is yuan, N is the number of the same kind of commodity in the past 30 days, e is the natural constant;

[0045] The dynamic threshold adjustment mechanism is defined as follows: when M=5 consecutive fuses are detected, the floating coefficient is triggered to adaptively adjust:

[0046]

[0047] Wherein, represents the floating coefficient of the mechanism at time t sys , ζ=0.02 is the adjustment step, erf(x) represents the Gaussian error function, Q req represents the number of alternative scheme requests per unit time, Q avl represents the total amount of available alternative goods, and 0.8 is the supply-demand balance threshold coefficient.

[0048] As a preferred scheme of the virtual commodity intelligent listing and management system for the financial institution credit mall, wherein: the intelligent compliance fuse:

[0049] In the listing stage of the commodity, the matching degree of the user risk level label in the evidence storage identifier and the KYC strategy of the financial institution is verified;

[0050] When the matching degree is lower than the preset threshold, the listing process is blocked and a violation flag report is generated.

[0051] As a preferred scheme of the virtual commodity intelligent listing and management system for the financial institution credit mall, wherein: the dynamic benefit substitution engine executes substitution:

[0052] Traverse the associated edge of the invalid single product in the benefit relationship graph;

[0053] Filter the candidate nodes that meet the following conditions at the same time:

[0054] The real-time inventory monitoring unit returns the available state,

[0055] There is no rule conflict with the remaining goods of the original combination, and the conflict is defined as violating the cross-platform rule library items,

[0056] The total cost after substitution is located in the preset floating interval of the financial institution.

[0057] As a preferred scheme of the virtual commodity intelligent listing and management system for the financial institution credit mall, wherein: in the commodity atomization packaging module:

[0058] The financial institution digital watermark is generated by splicing the institution code and the user credit account hash value;

[0059] The blockchain storage certificate identifier is associated with a supplier key fragment for the regulatory party to decrypt and verify.

[0060] As a preferred scheme of the virtual commodity intelligent listing and management system for the financial institution credit mall, the intelligent compliance fuse is further configured to:

[0061] Listen to third-party platform rule change events;

[0062] When it is detected that the listed goods are in violation of the rules due to rule updates, the migration scheme generated by the replacement engine is automatically triggered.

[0063] As a preferred scheme of the virtual commodity intelligent listing and management system for the financial institution credit mall, the blockchain storage certificate identifier is generated in the financial institution private chain, and the chain node includes a supplier node and a regulatory audit node.

[0064] The regulatory audit node has read-only permission and data tampering alarm function.

[0065] The beneficial effects of the present application are: when a single product in the combination commodity suddenly runs out of stock, the system can automatically select the best replacement product based on the dynamically constructed interest relationship graph; the graph weight fusion realizes real-time rule conflict detection, ensures that the replacement scheme meets the user's preferences and the financial institution's cost restrictions, and compared with the existing scheme, the system compresses the processing time from hours to seconds, completely avoiding customer complaints caused by human errors; At the same time, the intelligent compliance fuse double protection mechanism is introduced to check the matching degree of user risk label and KYC strategy during the listing stage of the goods, and intercept high-risk exchanges such as underage exchange of restricted goods; real-time monitoring of third-party rule changes during the existence period, automatic migration of goods due to policy unlisting, this mechanism changes the traditional ex-post compliance review to pre-blocking, significantly reducing the risk of financial institution violation penalties.

[0066] The present application introduces a dynamic threshold adjustment in the cost constraint function, when it is detected that the replacement scheme is triggered for cost fuse for several times in succession, the floating coefficient is automatically relaxed, avoiding replacement failure caused by stock shortage, and combining market average price to calculate absolute cost safety boundary, preventing systematic loss caused by replacing high-priced goods, the institution can accurately control the integral exchange cost rate; the blockchain storage certificate identifier adopts a sharded design, the institution watermark is generated by public coding and user hash value for fast traceability; the supplier key fragment is used for the regulatory party to decrypt and verify, realizing partial transparent audit, the regulatory node has data tampering alarm function, ensuring that the virtual commodity ownership change leaves traces throughout the process, meeting the financial industry data governance requirements. BRIEF DESCRIPTION OF DRAWINGS

[0067] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed to be used in the embodiments will be briefly introduced as follows. Obviously, the drawings in the following description only constitute some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor on the basis of these drawings.

[0068] Figure 1 The framework schematic diagram of the virtual commodity intelligent shelving and management system in the embodiment 1. DETAILED DESCRIPTION

[0069] In order to make the above-mentioned purposes, features and advantages of the present application more apparent and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the drawings of the specification.

[0070] In the following description, a large number of specific details are set forth in order to facilitate a thorough understanding of the present application, but the present application can also be implemented in other ways different from those described herein, and those skilled in the art can make similar generalizations without departing from the connotation of the present application, therefore the present application is not limited to the specific embodiments disclosed below.

[0071] Secondly, the "one embodiment" or "embodiment" referred to herein means that the specific features, structures or characteristics can be included in at least one implementation of the present application. "In one embodiment" appearing in different places in the specification does not mean the same embodiment, nor is it an independent or alternative embodiment that excludes other embodiments.

[0072] Embodiment 1, refer to Figure 1 The embodiment provides a virtual commodity intelligent shelving and management system for a financial institution points mall, comprising:

[0073] A commodity atomization packaging module is used to disassemble a third-party virtual commodity into a minimum right unit, and generate a unique blockchain storage identification with a financial institution digital watermark;

[0074] In the commodity atomization packaging module:

[0075] The financial institution digital watermark is generated by splicing an institution code and a user points account hash value;

[0076] The blockchain storage identification is associated with a supplier key fragment for the regulatory party to decrypt and verify;

[0077] The blockchain storage identification is generated in a financial institution private chain, and the chain node includes a supplier node and a regulatory audit node;

[0078] The regulatory audit node has a read-only permission and a data tampering alarm function;

[0079] Cross-platform rule library, dynamically access financial institution compliance strategy and third-party platform commodity rules;

[0080] Dynamic equity replacement engine, responds to combination commodity single product failure events, and outputs replacement schemes based on equity relationship graph;

[0081] The dynamic equity replacement engine comprises:

[0082] Equity relationship graph library, storing a network structure with commodities as nodes and replacement relationships as edges, and dynamically calculating edge weights;

[0083] Real-time inventory monitoring unit, interacts with third-party platform API through polling mechanism, updates inventory state every 15 seconds and marks abnormal threshold;

[0084] In the equity relationship graph library, the step of dynamically calculating the edge weight comprises:

[0085] When the user exchange behavior data set D t satisfies ΔD=KL(D t ||D t-1 )>θ, the weight update is triggered, D t represents the user exchange behavior data set at time t, containing user ID, commodity ID and exchange timestamp triplets, KL(p||q) represents the Kullback-Leibler divergence, which calculates the difference between the behavior distributions of two consecutive time windows, θ=0.15 represents the threshold calibrated by historical data mutation detection, and the unit is information entropy bits;

[0086] A three-dimensional weight calculation model is constructed, and the model formula is represented as:

[0087]

[0088] Among them, represents the replacement weight of commodity a→b at time t, dimensionless, range [0,1], α represents the semantic similarity weight coefficient, constraint condition 0.2≤α≤0.5, S ab represents the semantic similarity of commodities a and b, β represents the co-exchange probability weight coefficient, constraint condition 0.2≤β≤0.5, represents the co-exchange probability of commodity a→b at time t, γ represents the rule compatibility weight coefficient, which is automatically calculated by γ=1-α-β,

[0089] The semantic similarity component in the formula is represented as:

[0090]

[0091] Among them, represents the BERT word vector of the title of commodity a, dimension 768, fine-tuned by financial corpus, Knowledge graph embedding vector of commodity a, generated by TransE algorithm, dimension 256,

[0092] ||·||2 represents an L2 norm calculator, G represents a set of all commodity nodes in the current graph,

[0093] The co-exchange probability component in the formula is represented as:

[0094]

[0095] where σ(x) = 1 / (1+e -x ) is a sigmoid function that maps values to (0, 1), N u represents the total number of active users in the past 30 days, and a user is defined as having completed at least one exchange, represents the decay factor of user k, n k is the number of exchanges of user k in the past 7 days, represents an indicator function, which is 1 when user k's exchange record contains both a and b, and 0 otherwise, t k is the time of the last exchange of a or b by user k, in Unix timestamp format, t represents the timestamp of the current calculation time, and λ = 0.05 represents the time decay coefficient, in seconds -1 ,

[0096] The rule compatibility component in the formula is represented as:

[0097]

[0098] where R a represents the set of rule clauses applicable to commodity a, with elements being cross-platform rule library item IDs, is the rule conflict coefficient, n is the total number of related clauses in the cross-platform rule library, w k is the severity weight of clause k, δ k ∈{0,1} indicates whether clause k is violated when a→b is replaced, c a represents the unit cost of commodity a for the financial institution to settle with the supplier, in yuan, and the denominator 5 and 0.9 represent the S-curve steepness coefficient and cost proportion threshold, determined by fitting historical financial data;

[0099] Specifically, the calculation system here dynamically couples multiple parameters to achieve precise quantification of replacement weights. The fusion of semantic vectors and knowledge graphs enhances the representation of commodity features, solving the problem of matching non-standard commodities. The time decay mechanism ensures that the model continuously tracks user preference shifts, and the KL divergence trigger condition avoids invalid calculations.

[0100] When the dynamic equity replacement engine performs replacement:

[0101] Traverse the associated edges of the failed single product in the equity relationship graph;

[0102] Screen the candidate nodes that meet the following conditions simultaneously:

[0103] The real-time inventory monitoring unit returns an available state,

[0104] There is no irregular conflict with the remaining goods of the original combination, and the conflict is defined as violating the cross-platform rule library entry,

[0105] The total cost after substitution is within the preset floating interval of the financial institution;

[0106] The construction of the equity relationship graph library includes:

[0107] The node is associated with the user historical behavior label, including the consumption frequency and category preference;

[0108] The calculation of the edge weight integrates the following dimensions:

[0109] Cross-platform rule conflict coefficient, 0-1 scale, user group substitute acceptance, and financial institution preset cost constraint function;

[0110] In the construction process of the equity relationship graph library, the construction steps of the cost constraint function include:

[0111] Obtain the original commodity cost c o , the cost of the candidate substitute c c , and the financial institution floating coefficient δ, δ ∈ [0.05, 0.2];

[0112] Calculate the absolute cost deviation Δ c = |c c -c o | and the relative cost deviation

[0113] Calculate the fuse determination function:

[0114]

[0115] Where c o represents the unit cost of the original commodity, unit yuan c c represents the unit cost of the candidate substitute, unit yuan, δ represents the preset floating coefficient of the financial institution, dimensionless, and is calibrated through historical financial data η c represents the relative cost deviation, dimensionless, Γ(c c , c o ) is the cost adaptation degree function, ∨ represents the operator logical or;

[0116] In the formula, the cost adaptation degree function is represented as:

[0117]

[0118] wherein k = 12 represents a relative deviation sensitivity coefficient, determined by gradient descent method optimization, μ = 0.08 is an absolute cost safety coefficient, unit: yuan -1 , based on 100,000 historical transaction data calibration;

[0119] represents the market average price of the same kind of goods, c i is the settlement cost of the i-th same kind of goods, unit: yuan, N is the number of similar goods in the past 30 days, e is the natural constant;

[0120] The dynamic threshold adjustment mechanism is defined as: when M = 5 consecutive fuses are detected, the floating coefficient is triggered to adaptively adjust:

[0121]

[0122] wherein, represents the floating coefficient of the agency at time t sys , ζ = 0.02 is the adjustment step, calibrated by control experiment, erf(x) represents the Gaussian error function, Q req represents the number of alternative scheme requests per unit time, Q avl represents the total amount of available alternative goods, and 0.8 is the supply and demand balance threshold coefficient;

[0123] Specifically, this function realizes accurate control of cost risk through a double constraint mechanism. The relative deviation term prevents cost over-limit caused by single replacement, and the absolute cost term of market average price avoids systemic risk of high-priced goods. The dynamic threshold mechanism automatically relaxes the constraint when inventory is tight, improving the coverage of the scheme. The error function adjustment ensures a smooth adjustment process and avoids parameter mutation.

[0124] Intelligent compliance fuse, real-time call rule base to perform operation interception;

[0125] Intelligent compliance fuse:

[0126] In the goods listing stage, verify the matching degree of the user risk level label in the evidence storage identifier and the financial institution KYC strategy;

[0127] When the matching degree is lower than the preset threshold, block the listing process and generate a violation flag report;

[0128] The intelligent compliance fuse is also configured to:

[0129] Listen to third-party platform rule change events;

[0130] When it is detected that the listed goods are in violation due to rule updates, automatically trigger the alternative engine to generate a migration scheme;

[0131] The calibration method of system parameters in this embodiment includes:

[0132] a) regression analysis based on historical data sets,

[0133] b) control variable method experiment,

[0134] c) industry experience value verification;

[0135] Establish a parameter configuration matrix Π:

[0136]

[0137] When deploying the system, load the parameters in any of the following ways:

[0138] (1) read the preset JSON configuration file;

[0139] (2) call the financial institution parameter management API to obtain real-time configuration.

[0140] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present application and are not limiting. Although the present application has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solutions of the present application can be modified or replaced equivalently without departing from the spirit and scope of the technical solutions of the present application, which should be covered in the scope of the claims of the present application.

Claims

1. A virtual commodity intelligent listing and management system for a financial institution points mall, characterized in that, Comprising, a commodity atomization packaging module for disassembling third-party virtual commodities into the smallest equity unit, generating a unique blockchain storage identification with a financial institution digital watermark; a cross-platform rule library dynamically accessing financial institution compliance strategies and third-party platform commodity rules; a dynamic equity replacement engine responding to combination commodity single product failure events and outputting replacement solutions based on an equity relationship graph; an intelligent compliance fuse for real-time calling of the rule library to perform operation interception; when the dynamic equity replacement engine performs replacement: traversing the associated edges of the failed single product in the equity relationship graph; screening candidate nodes that simultaneously satisfy the following conditions: the real-time inventory monitoring unit returns an available state, there is no rule conflict with the remaining commodities of the original combination, and the conflict is defined as a violation of the cross-platform rule library entries, the total cost after replacement is within the preset floating interval of the financial institution; the dynamic equity replacement engine includes an equity relationship graph library, and the construction of the equity relationship graph library includes: node association user historical behavior labels, including consumption frequency and category preference; the calculation of edge weights integrates the following dimensions: a cross-platform rule conflict coefficient, a 0-1 scale, a user group replacement product acceptance, and a financial institution preset cost constraint function; in the construction process of the dynamic equity replacement engine, the construction steps of the cost constraint function include: obtaining the unit cost of the original commodity, the unit cost of the candidate replacement product, and the preset floating coefficient of the financial institution; Computing absolute cost deviation With relative cost deviation ; calculating the fuse decision function: ， wherein, denotes the original commodity unit cost, unit Yuan, denotes the candidate substitute unit cost, unit Yuan, denotes the preset floating coefficient of the financial institution, dimensionless, calibrated by historical financial data, denotes the relative cost deviation, dimensionless, is a cost adaptation degree function, denotes the operator logic or; in the formula, the cost adaptation degree function is expressed as: ， wherein, represents a relative deviation sensitivity coefficient, is an absolute cost safety coefficient; represents the market average price of the same kind of goods, is the settlement cost of the ith same kind of goods, in yuan, is the number of the same kind of goods in the past 30 days, is a natural constant; The dynamic threshold adjustment mechanism is defined as follows: when consecutive fusing is detected, the floating coefficient adaptive adjustment is triggered: ， wherein, denotes the mechanism floating coefficient at the moment, is the adjustment step, calibrated by controlling experiments, denotes the Gaussian error function, denotes the alternative request amount per unit time, denotes the total amount of available alternative goods, is the supply-demand balance threshold coefficient.

2. The virtual commodity intelligent listing and management system for the financial institution credit mall according to claim 1, wherein, the dynamic equity replacement engine includes: an equity relationship graph library storing a network structure with commodities as nodes and replacement relationships as edges, and dynamically calculating edge weights; a real-time inventory monitoring unit that interacts with the third-party platform API through a polling mechanism, updates the inventory state every 15 seconds, and marks the abnormal threshold.

3. The virtual commodity intelligent listing and management system for financial institution credit mall according to claim 2, wherein, in the equity relationship graph library, the steps of dynamically calculating the edge weight include: When the user exchange behavior dataset is satisfied , the weight update is triggered, represents the user exchange behavior dataset at time t, containing user ID, commodity ID and exchange timestamp triplets, represents the Kulback-Leibler divergence, which calculates the difference between the behavior distributions of two consecutive time windows, represents the threshold calibrated by the history data mutation detection, with the unit of information entropy bits; constructing a three-dimensional weight calculation model, and the model formula is expressed as: ， wherein, represents the substitution weight of commodity a→b at time t, dimensionless, range [0, 1], represents the semantic similarity weight coefficient, constraint condition 0.2≤a≤0.5, represents the semantic similarity of commodity a and b, represents the co-exchange probability weight coefficient, constraint condition 0.2≤β≤0.5, represents the co-exchange probability of commodity a→b at time t, represents the rule compatibility weight coefficient, by automatic calculation, in the formula, the semantic similarity component is expressed as: ， wherein, represents a BERT word vector of the title of the product a, represents a knowledge graph embedding vector of the product a, denotes an L2 norm calculator, denotes a set of all commodity nodes in the current graph, in the formula, the co-exchange probability component is expressed as: ， where, sigmoid function, mapping values to (0, 1), denotes the total number of active users in the last 30 days, with a user defined as having completed at least one exchange, denotes the decay factor for user k, denotes the number of exchanges for user k in the last 7 days, denotes the indicator function, which is 1 if both a and b are included in the exchange record of user k, and 0 otherwise, denotes the time of the last exchange of a or b by user k, in Unix timestamp format, denotes the timestamp of the current calculation time, denotes the time decay coefficient; in the formula, the rule compatibility component is expressed as: ， wherein, represents the set of rule clauses applicable to item a, the elements being cross-platform rule base entry IDs, is a rule conflict coefficient, is the total number of related clauses in the cross-platform rule base, is a clause severity weight, indicates whether the clause is violated when a→b is replaced, indicates the unit cost of the financial institution to the supplier for goods a, unit yuan, 5 and 0.9 in the denominator represent the S-shaped curve steepness coefficient and the cost proportion threshold value, which are determined by fitting historical financial data.

4. The virtual goods intelligent listing and management system for financial institution credit mall according to claim 1, wherein, the intelligent compliance fuse: in the commodity listing stage, verify the matching degree of the user risk level label in the storage identification and the financial institution KYC strategy; when the matching degree is lower than the preset threshold, block the listing process and generate a violation report.

5. The virtual goods intelligent listing and management system for financial institution credit mall according to claim 1, wherein, in the commodity atomization packaging module: the financial institution digital watermark is generated by splicing the institution code and the user integral account hash value; the blockchain storage identification is associated with the supplier key fragments for supervision parties to decrypt and verify.

6. The virtual goods intelligent listing and management system for financial institution credit mall according to claim 1, wherein, the intelligent compliance fuse is further configured to: listen to third-party platform rule change events; when it is detected that the listed commodity is in violation due to rule updates, automatically trigger the replacement engine to generate a migration solution.

7. The virtual goods intelligent listing and management system for financial institution credit mall according to claim 1, wherein, the blockchain storage identification is generated in the financial institution private chain, and the chain nodes include supplier nodes and supervision and audit nodes; the supervision and audit nodes have read-only permission and data tampering alarm function.

Images