A method and related apparatus for calculating object association strength based on multi-source heterogeneous data

By acquiring multi-source heterogeneous data and performing reverse completion when interactive data is missing, and calculating relationship values ​​by combining preset interval mapping rules, the problems of multi-source heterogeneous data fusion and data missing are solved, and the fine quantification and accuracy improvement of the correlation strength between objects are achieved.

CN123118265APending Publication Date: 2026-11-17BANK OF CHINA
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611353692.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-09-02
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing technologies cannot effectively integrate multi-source heterogeneous data in big data analysis, resulting in incomplete characterization of relationships, low data utilization efficiency, and a lack of cross-object reverse data completion mechanisms, leading to computational interruptions or distorted results.

Method used

By acquiring heterogeneous data from multiple sources, we extract the interaction data between the target object and related objects in multiple dimensions. When the interaction data is missing, we perform reverse completion based on the interaction behavior in the same dimension. We use the symmetry principle of interaction behavior to recover the missing features, and combine the preset interval mapping rules to calculate the relationship value, and finally merge them into the association strength.

Benefits of technology

It enables the fusion of heterogeneous information across data sources, improves data utilization and integrity, significantly enhances the accuracy and discriminative power of association strength calculation, and provides a reliable data processing foundation for inter-object association analysis.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123118265A_ABST
    Figure CN123118265A_ABST
Patent Text Reader

Abstract

This application discloses a method and related apparatus for calculating the association strength of objects based on multi-source heterogeneous data, relating to the fields of big data or financial technology. By acquiring multi-source heterogeneous data and extracting interaction data between the target object and related objects in multiple dimensions, heterogeneous information fusion across data sources is achieved. When interaction data in the target dimension is missing, the method performs reverse completion based on the interaction data of related objects in different interaction directions of the same interaction behavior, and recovers the missing features by utilizing the principle of symmetry of interaction behavior. By independently calculating the relationship value for each dimension based on a preset interval mapping rule and merging the relationship values ​​of each dimension into the association strength, the method improves the discriminativeness and accuracy of association strength calculation, and provides a reliable data processing foundation for inter-object association analysis.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This application relates to the field of big data technology, and in particular to a method and related apparatus for calculating the correlation strength of objects based on multi-source heterogeneous data. Background Technology

[0002] In big data analytics scenarios for corporate clients, it is necessary to quantify the degree of correlation between different objects based on various types of interactions between enterprises, so as to realize business applications such as customer link mining, cooperation relationship tracing, and business path identification.

[0003] Therefore, there is an urgent need for a method to calculate the strength of association between different objects. Summary of the Invention

[0004] In view of the above problems, this application provides a method and related apparatus for calculating the association strength between objects based on multi-source heterogeneous data, so as to achieve the purpose of calculating the association strength between different objects. The specific solution is as follows:

[0005] The first aspect of this application provides a method for calculating the object association strength based on multi-source heterogeneous data, including:

[0006] Acquire multi-source heterogeneous data;

[0007] The interaction data of the target object with respect to the associated object in multiple dimensions are obtained from the multi-source heterogeneous data;

[0008] If the interaction data of the target object in the first interaction direction of the target dimension is missing, the interaction data of the first interaction direction is supplemented based on the interaction data of the associated object in the second interaction direction of the target dimension. The first interaction direction and the second interaction direction are opposite interaction directions of the same interaction behavior.

[0009] For each dimension, based on the interaction data of the target object relative to the associated object in that dimension and the preset interval mapping rules, the relationship value between the target object and the associated object in that dimension is calculated.

[0010] Based on the relationship value between the target object and the associated object in each dimension, the association strength value between the target object and the associated object is calculated.

[0011] In one possible implementation, the multiple dimensions include at least two of the following: fund interaction dimension, equity topology dimension, personnel appointment dimension, bill transaction dimension, project bidding cooperation dimension, and organizational hierarchy dimension.

[0012] The step of obtaining interaction data of the target object relative to the associated object in multiple dimensions from the multi-source heterogeneous data includes at least two of the following:

[0013] If the dimension is the fund interaction dimension, obtain the settlement direction of the target object relative to the associated object within a preset period and the total settlement transaction amount corresponding to the settlement direction;

[0014] If the dimension is the equity topology dimension, obtain the shareholding relationship of the target object relative to the associated object within the preset period and the shareholding ratio corresponding to the shareholding relationship;

[0015] If the dimension is the personnel employment dimension, obtain the number of related enterprises of the target object and the related object within the preset period, and the wealth level parameters of the personnel with overlapping employment of the target object and the related object;

[0016] If the dimension is the bill transaction dimension, obtain the seller identifier and the corresponding sales transaction data of the seller identifier relative to the associated object within the preset period, or the buyer identifier and the corresponding input transaction data of the buyer identifier;

[0017] If the dimension is the project bidding cooperation dimension, obtain the bidding identifier and the preset bidding score corresponding to the bidding identifier of the target object relative to the associated object within the preset period, or the winning bid identifier and the preset winning bid score corresponding to the winning bid identifier;

[0018] If the dimension is the organizational hierarchy dimension, obtain the target object's superior identifier and the corresponding preset superior hierarchy score within the preset period, or the subordinate identifier and the corresponding preset subordinate hierarchy score.

[0019] In one possible implementation, obtaining the target object's superior identifier and its corresponding preset superior-level score, or the subordinate identifier and its corresponding preset subordinate-level score, within the preset period, relative to the associated object, includes:

[0020] Based on the subject name of the target object and the subject name of the associated object, determine the hierarchical relationship between the target object and the associated object;

[0021] If the hierarchical relationship indicates that the target object is a subordinate object, configure a subordinate identifier and a corresponding preset subordinate level score for the target object;

[0022] If the hierarchical relationship indicates that the target object is a superior object, then a superior identifier and a corresponding preset superior level score are configured for the target object.

[0023] In one possible implementation, determining the hierarchical relationship between the target object and the associated object based on the subject name of the target object and the subject name of the associated object includes:

[0024] If the main name of the associated object is formed by concatenating the main name of the target object with a preset string, the target object is determined to be the parent object;

[0025] If the main name of the target object is formed by concatenating the main name of the associated object with the preset string, the target object is determined to be a subordinate object.

[0026] In one possible implementation, calculating the relationship value between the target object and the associated object in that dimension based on the interaction data of the target object relative to the associated object and a preset interval mapping rule includes:

[0027] Determine the target numerical range of the interaction data of the target object relative to the associated object in this dimension;

[0028] Find the target gradient level value corresponding to the target numerical range from the correspondence between the numerical range and the gradient level value;

[0029] Based on the target gradient level value and the preset dimension benchmark coefficient, the relationship value of that dimension is calculated.

[0030] In one possible implementation, the step of supplementing the interaction data in the first interaction direction based on the interaction data of the associated object in the second interaction direction of the target dimension includes:

[0031] If the second interaction direction is the second identifier, then the first interaction direction is set to the first identifier;

[0032] The interaction data corresponding to the second interaction direction is used as the interaction data corresponding to the first interaction direction.

[0033] In one possible implementation, the target dimension is the bill transaction dimension, and the interaction data in the second interaction direction includes the seller identifier and the corresponding sales transaction data within a preset period, or the buyer identifier and the corresponding purchase transaction data.

[0034] The step of setting the first interaction direction as the first identifier if the second interaction direction is the second identifier includes:

[0035] If the second identifier is the seller identifier, then the first identifier is set to the buyer identifier;

[0036] If the second identifier is the buyer's identifier, then the first identifier is set to the seller's identifier;

[0037] The step of using the interaction data corresponding to the second interaction direction as the interaction data corresponding to the first interaction direction includes:

[0038] If the second identifier is the seller identifier, the sales transaction data in the interaction data of the second interaction direction shall be used as the input transaction data in the interaction data of the first interaction direction;

[0039] If the second identifier is the buyer's identifier, the input transaction data in the interaction data of the second interaction direction will be used as the output transaction data in the interaction data of the first interaction direction.

[0040] In one possible implementation, the interactive data of the invoice transaction dimension also includes the invoice issuance time and the invoice data entry time;

[0041] If the invoice issuance time is within a preset time window, and the time difference between the current time and the invoice data entry time is less than or equal to a preset storage validity threshold, then the interaction data in the first interaction direction is completed based on the interaction data of the associated object in the second interaction direction of the target dimension.

[0042] The start time of the preset time window is the authorization time of the target object, and the length of the preset time window is a preset duration.

[0043] A second aspect of this application provides an object association strength calculation device based on multi-source heterogeneous data, comprising:

[0044] The data acquisition module is used to acquire heterogeneous data from multiple sources;

[0045] The feature extraction module is used to obtain interaction data of the target object relative to the associated object in multiple dimensions from the multi-source heterogeneous data;

[0046] The data reverse completion module is used to complete the interaction data in the first interaction direction based on the interaction data of the associated object in the second interaction direction of the target dimension if the interaction data of the target object in the first interaction direction of the target dimension is missing. The first interaction direction and the second interaction direction are opposite interaction directions of the same interaction behavior. The target dimension is any one of the multiple dimensions.

[0047] The scoring module is used to calculate the relationship value between the target object and the associated object in each dimension, based on the interaction data of the target object relative to the associated object in the dimension and a preset interval mapping rule.

[0048] The multidimensional fusion module is used to calculate the association strength value between the target object and the associated object based on the relationship value between the target object and the associated object in each dimension.

[0049] A third aspect of this application provides a computer program product, including computer-readable instructions that, when executed on an electronic device, cause the electronic device to implement the object association strength calculation method based on multi-source heterogeneous data described in the first aspect or any implementation thereof.

[0050] A fourth aspect of this application provides an electronic device, including at least one processor and a memory connected to the processor, wherein:

[0051] At least one processor; and,

[0052] A memory communicatively connected to the at least one processor; wherein,

[0053] The memory stores instructions that can be executed by the at least one processor, which enable the at least one processor to implement the object association strength calculation method based on multi-source heterogeneous data as described in the first aspect or any implementation thereof.

[0054] The fifth aspect of this application provides a computer storage medium carrying one or more computer programs, which, when executed by an electronic device, enable the electronic device to implement the object association strength calculation method based on multi-source heterogeneous data described in the first aspect or any implementation thereof.

[0055] By employing the above technical solutions, this application provides a method for calculating the association strength of objects based on multi-source heterogeneous data. By acquiring multi-source heterogeneous data and extracting interaction data between the target object and related objects across multiple dimensions, it achieves cross-data source heterogeneous information fusion, overcoming the problem of incomplete association characterization caused by traditional solutions relying on single-dimensional data. When interaction data in the target dimension is missing, it performs reverse completion based on the interaction data of related objects with different interaction directions under the same dimension, utilizing the symmetry principle of interaction behavior to recover missing features. This effectively avoids computational interruptions caused by data permissions or collection restrictions, improving data utilization and completeness. By independently calculating the relationship value for each dimension based on preset interval mapping rules and fusing the relationship values ​​of each dimension into the association strength, it achieves a technological leap from qualitative judgment to multi-dimensional gradient quantization, significantly improving the discriminative power and accuracy of association strength calculation, and providing a reliable data processing foundation for inter-object association analysis. Attached Figure Description

[0056] The above and other features, advantages, and aspects of the embodiments of this disclosure will become more apparent from the accompanying drawings and the following detailed description. Throughout the drawings, the same or similar reference numerals denote the same or similar elements. It should be understood that the drawings are schematic, and the originals and elements are not necessarily drawn to scale.

[0057] Figure 1 A schematic diagram of a system architecture is provided for this application;

[0058] Figure 2 A flowchart illustrating a method for calculating object association strength based on multi-source heterogeneous data, provided in an embodiment of this application;

[0059] Figure 3 A schematic diagram of the structure of an object association strength calculation device based on multi-source heterogeneous data provided in this application embodiment;

[0060] Figure 4 This is a schematic diagram of the structure of an electronic device provided in this application. Detailed Implementation

[0061] The embodiments of this application are described below with reference to the accompanying drawings. The terminology used in the implementation section of this application is for explaining specific embodiments only and is not intended to limit the scope of this application.

[0062] As will be known to those skilled in the art, with the development of technology and the emergence of new scenarios, the technical solutions provided in the embodiments of this application are also applicable to similar technical problems.

[0063] The terms "first," "second," etc., used in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such terms are interchangeable where appropriate; this is merely a way of distinguishing objects with the same attributes in the embodiments of this application. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion, so that a process, method, system, product, or apparatus that comprises a series of elements is not necessarily limited to those elements, but may include other elements not explicitly listed or inherent to those processes, methods, products, or apparatuses.

[0064] In big data analytics scenarios involving complex enterprise network analysis, corporate customer mining, and assessment of object association strength, it is necessary to accurately quantify the degree of association between different objects based on interaction data of various inter-enterprise interaction behaviors. This enables business scenarios such as customer link mining, cooperative relationship tracing, and business path identification.

[0065] For this type of business scenario, the actual application requirement is to rely on multi-source heterogeneous data to comprehensively and meticulously mine implicit and explicit relationships between objects. Currently, the industry's methods for calculating and judging the strength of object associations are relatively simple. Existing technologies generally rely only on single-dimensional data or static enterprise attributes to complete object screening, association matching, and strength determination, without integrating and applying multi-dimensional and heterogeneous business interaction data.

[0066] Existing technologies have significant technical shortcomings when adapting to the aforementioned business needs. First, the independent storage and analysis of multi-dimensional interaction data creates severe data silos, hindering the deep integration of multi-source heterogeneous data, making it difficult to comprehensively uncover complex relationships between objects, resulting in low data utilization efficiency and a one-sided approach to correlation analysis. Second, some enterprise interaction data is often missing due to issues such as access restrictions, data collection omissions, and data synchronization anomalies. Existing technologies lack mature cross-object reverse data completion mechanisms, which can easily lead to interruptions in the correlation strength calculation process or distortion of the final calculation results, compromising the continuity and reliability of the analysis work.

[0067] In summary, there is an urgent need for a computational method that can integrate multi-source heterogeneous data, support intelligent reverse completion of missing interactive data, and achieve refined quantification of object association strength, in order to solve many technical shortcomings of existing technologies.

[0068] This application can be applied to big data analysis fields such as enterprise network analysis, corporate client risk assessment, business partnership mining, and corporate credit rating.

[0069] This application can be applied, but is not limited to, to applications that have the function of calculating the object association strength based on multi-source heterogeneous data or cloud services provided by cloud-side servers, which will be described in detail below:

[0070] See Figure 1 , Figure 1 A schematic diagram of a system architecture is shown. The system may include a terminal 100 and a server 200. The server 200 can provide the methods provided in the embodiments of this application to one or more terminals.

[0071] The terminal 100 can be equipped with applications for enterprise association analysis, customer relationship assessment, and big data risk control analysis. These applications and web pages can provide an interface. The terminal 100 can receive relevant parameters input by the user on the visual operation interface and send the parameters to the server 200. The server 200 can obtain the processing result based on the received parameters and return the processing result to the terminal 100.

[0072] It should be understood that in some optional implementations, the terminal 100 can also complete the action of obtaining the processing result based on the received parameters on its own, without the need for the server to cooperate. This application embodiment is not limited to this.

[0073] The following description Figure 1 The product form of the mid-terminal 100;

[0074] The terminal 100 in this application embodiment can be a mobile phone, tablet computer, wearable device, vehicle device, augmented reality (AR) / virtual reality (VR) device, laptop computer, ultra-mobile personal computer (UMPC), netbook, personal digital assistant (PDA), etc., and this application embodiment does not impose any restrictions on it.

[0075] Terminal 100 may include a radio frequency unit, memory, input unit, display unit, camera (optional), audio circuitry (optional), speaker (optional), microphone (optional), headphone jack (optional), processor, external interface, power supply, and other components. Those skilled in the art will understand that the above-mentioned components are merely examples and do not constitute a limitation on the terminal or multifunctional device; it may include more or fewer components, or a combination of certain components, or different components.

[0076] The input unit can be used to receive input numeric or character information, and to generate key signal inputs related to user settings and function control of the portable multi-functional device. Specifically, the input unit may include a touchscreen (optional) and / or other input devices. Other input devices may include, but are not limited to, one or more of a physical keyboard, function keys (such as volume control buttons, power buttons, etc.), trackball, mouse, joystick, etc.

[0077] Among them, the input device can receive input data, etc.

[0078] The display unit can be used to display information input by the user or information provided to the user, various menus of the terminal, interactive interfaces, file display, and / or playback of any multimedia file. In the embodiments of this application, the display unit can be used to display interfaces showing the strength of relationships between objects, processing results, etc.

[0079] The memory can be used to store software code related to the object association strength calculation method based on multi-source heterogeneous data. The processor can execute the steps of the object association strength calculation method based on multi-source heterogeneous data, and can also schedule other units (such as the above-mentioned input unit and display unit) to achieve the corresponding functions.

[0080] This radio frequency unit (optional) can be used to receive and send signals during information transmission or calls.

[0081] In this embodiment of the application, the radio frequency unit can send data to the server 200 and receive the processing results sent by the server 200.

[0082] It should be understood that this radio frequency unit is optional and can be replaced with other communication interfaces, such as a network port.

[0083] Terminal 100 also includes a power source (such as a battery) for supplying power to the various components.

[0084] Terminal 100 also includes an external interface, which can be a standard Micro USB interface or a multi-pin connector, which can be used to connect terminal 100 to other devices for communication, or to connect a charger to charge terminal 100.

[0085] Server 200 includes a bus, a processor, a communication interface, and memory. The processor, memory, and communication interface communicate with each other via the bus.

[0086] The memory can be used to store software code related to the object association strength calculation method based on multi-source heterogeneous data. The processor can execute the steps of the chip's object association strength calculation method based on multi-source heterogeneous data, and can also schedule other units to achieve corresponding functions.

[0087] Reference Figure 2 , Figure 2 A flowchart illustrating a method for calculating object association strength based on multi-source heterogeneous data, as provided in this application embodiment, is shown below. Figure 2 As shown in the embodiment of this application, a method for calculating the object association strength based on multi-source heterogeneous data is provided, which may include steps S201 to S205. These steps are described in detail below.

[0088] Step S201: Obtain multi-source heterogeneous data.

[0089] Multi-source heterogeneous data refers to a mixed dataset that originates from different business systems, has different data structures, and represents multi-dimensional attributes of objects. It encompasses both structured and unstructured data.

[0090] The acquisition of multi-source heterogeneous data can be achieved by connecting to external data interfaces or reading internal databases. For example, fund transfer records can be obtained from transaction flow databases, equity topology and personnel appointment information can be obtained from enterprise business information databases, invoice transaction details can be obtained from tax invoice systems, and project bidding information can be obtained from public resource trading platforms.

[0091] After acquiring heterogeneous data from multiple sources, preprocessing operations are required, such as filtering out invalid object data with operating statuses of obsolescence, closure orders, or business suspension, and standardizing and cleaning missing fields to generate a high-quality basic dataset that can be used for subsequent calculations.

[0092] This step integrates cross-domain data sources, breaking down information silos caused by a single data dimension in traditional solutions, and providing underlying data support for comprehensively depicting the relationships between objects.

[0093] Step S202: Obtain the interaction data of the target object relative to the associated object in multiple dimensions from the multi-source heterogeneous data.

[0094] The target object refers to the object for which the correlation strength quantification analysis is to be performed, such as a company or an individual.

[0095] A related object is an object that has a relationship with a target object in at least one dimension. For example, the target object can have one or more related objects.

[0096] A dimension is an independent interactive evaluation perspective used to measure the relationship between objects. For example, multiple dimensions include, but are not limited to: financial interaction dimension, equity topology dimension, personnel appointment dimension, invoice transaction dimension, project bidding cooperation dimension, and organizational hierarchy dimension. Each dimension is explained below.

[0097] Funds Interaction Dimension: This dimension is used to characterize the closeness of two-way fund settlement transactions between the target object and related objects within a preset period. It uses the total amount of settlement transactions received and paid by both parties as the core interaction data, reflecting the activity level and binding depth of fund flow between objects.

[0098] Equity Topology Dimension: An evaluation dimension used to characterize the equity investment binding relationship between target objects and related objects within a preset period. It uses the shareholding relationship identifier between each pair of objects and the corresponding shareholding ratio as the core interactive data to depict the capital relationship topology and binding strength between objects.

[0099] Personnel Appointment Dimension: This is an evaluation dimension used to characterize the degree of personnel association between the target object and related objects within a preset period. It uses the number of related companies, the number of overlapping personnel, the type of public and private relationship of overlapping personnel, and the corresponding wealth level parameters as core interactive data. It reflects the depth of implicit corporate association based on the personnel's appointment links and personal qualification levels.

[0100] The document transaction dimension is an evaluation dimension used to characterize the degree of closeness of market-based transaction cooperation between the target object and related objects within a preset period. It uses the total amount of price and tax corresponding to the purchase and sales invoices of both parties and the direction of document interaction as the core interactive data. It can objectively reflect the frequency and scale of business cooperation between objects through two-way document transaction behavior.

[0101] Project bidding and tendering cooperation dimension: This is an evaluation dimension used to characterize the cooperation behavior of industrial projects between target objects and related objects. It uses the cooperation behavior types that can identify bidding and winning as the core interactive data to identify explicit project cooperation relationships between objects.

[0102] Organizational hierarchy and affiliation dimension: This is an evaluation dimension used to characterize the subordinate relationship of the legal organizational structure between the target object and related objects. It uses the hierarchical affiliation identifier as the core interactive data to distinguish between subordinate entities such as the head office and branch offices, and provides a hierarchical weight basis for calculating the association strength.

[0103] Interaction data refers to quantified fields that, after feature extraction, directly characterize the dynamic relationship between the target object and related objects, distinguishing them from the original static attribute data. Specifically, for each dimension, corresponding interaction features are parsed from the cleaned base dataset.

[0104] It should be understood that the “multiple dimensions” here are not fixed and can be flexibly configured according to data availability in practical applications, as long as at least two dimensions can reflect the interaction behavior between objects.

[0105] Step S203: If the interaction data of the target object in the first interaction direction of the target dimension is missing, the interaction data in the first interaction direction is supplemented based on the interaction data of the associated object in the second interaction direction of the target dimension. The first interaction direction and the second interaction direction are opposite interaction directions of the same interaction behavior. The target dimension is any one of the multiple dimensions.

[0106] Interaction behavior refers to transactional activities between a target object and related objects. For example, interaction behavior in the financial interaction dimension refers to multiple two-way fund receipts and payments settlements. Interaction behavior in the equity topology dimension refers to equity investment interaction behavior.

[0107] The following example illustrates the opposite direction of the same interactive behavior.

[0108] The different interaction directions of multiple two-way fund receipt and payment settlement transactions are {the target object pays to the related object, and the related object receives the payment from the target object}; the different interaction directions of equity investment transactions are {the target object holds shares in the related object, and the related object's shares are held by the target object}.

[0109] In this embodiment, the first and second interaction directions represent bidirectional recording perspectives of the same interaction behavior. Specifically, when the target object lacks interaction data in a certain dimension due to unauthorized access or limited data collection, the principle of symmetry in interaction behavior can be utilized to reverse-engineer and recover the target object's missing data from related objects that already possess interaction data. For example, in the dimension of invoice transactions, if target object B lacks sales transaction data for issuing invoices to related object A (i.e., missing in the B→A direction), but related object A possesses input transaction data for receiving invoices from target object B (i.e., present in the A←B direction), then the input transaction data of related object A can be directly mapped to the sales transaction data of target object B, thereby completing the data completion.

[0110] This reverse completion mechanism is not a simple default value filling, but a deduction and recovery based on the inherent logic of the data, which effectively solves the problem of interruption in the calculation of association strength caused by the lack of interaction data of some objects.

[0111] Step S204: For each dimension, based on the interaction data of the target object relative to the associated object in that dimension and the preset interval mapping rules, calculate the relationship value between the target object and the associated object in that dimension.

[0112] Among them, the preset interval mapping rule is a standardized data processing algorithm that transforms interactive data into numerical values ​​of the same dimension.

[0113] Specifically, the preset interval mapping rules define the correspondence between numerical intervals and gradient level values. For example, for the fund interaction dimension, multiple threshold intervals for total transaction amount can be preset.

[0114] This step normalizes the multi-dimensional interactive data, eliminates the interference of heterogeneity of multi-source data on the comprehensive calculation, and ensures the objectivity and accuracy of the quantification results.

[0115] Step S205: Based on the relationship value between the target object and the associated object in each dimension, calculate the association strength value between the target object and the associated object.

[0116] In this embodiment, the association strength value is a comprehensive quantitative indicator that characterizes the overall degree of association between two objects in the association graph.

[0117] This association strength value can be directly used to update the edge weights in the association graph of an object, or as a threshold for filtering high-value association paths.

[0118] The nodes in the association graph are the objects in this embodiment. If two nodes are associated, there is an edge between the two nodes; the weight of the edge between two nodes is the association strength value between the two nodes.

[0119] Specifically, linear summation models, weighted average models, or other fusion algorithms can be used to fuse the relationship values ​​of various dimensions in a multi-dimensional manner. For example, the relationship values ​​of six dimensions—funding interaction, equity topology, personnel appointment, bill transactions, project bidding cooperation, and organizational hierarchy—can be directly added together to obtain the final association strength value.

[0120] By using multi-dimensional fusion computation, this application overcomes the one-sidedness of single-dimensional evaluation and realizes a three-dimensional and refined measurement of the strength of the relationship between objects, providing a reliable data processing foundation for complex network analysis.

[0121] This application provides a method for calculating object association strength based on multi-source heterogeneous data. By acquiring multi-source heterogeneous data and extracting interaction data between the target object and related objects in multiple dimensions, it achieves cross-data source heterogeneous information fusion, overcoming the problem of incomplete characterization of association relationships caused by traditional solutions relying on single-dimensional data. When interaction data in the target dimension is missing, it performs reverse completion based on the interaction data of related objects with different interaction directions under the same dimension, and restores the missing features by utilizing the symmetry principle of interaction behavior. This effectively avoids calculation interruptions caused by data permissions or collection restrictions, and improves data utilization and completeness. By independently calculating the relationship value for each dimension based on a preset interval mapping rule and fusing the relationship values ​​of each dimension into the association strength, it achieves a technical leap from qualitative judgment to multi-dimensional gradient quantization, significantly improving the discriminativeness and accuracy of association strength calculation, and providing a reliable data processing foundation for inter-object association analysis.

[0122] In one optional implementation method, multiple dimensions include at least two of the following: fund interaction dimension, equity topology dimension, personnel appointment dimension, bill transaction dimension, project bidding cooperation dimension, and organizational hierarchy dimension.

[0123] Specifically, these six dimensions constitute a complete feature space depicting the relationships between objects in the relationship graph, encompassing multi-level information ranging from explicit capital ties and transaction flows to implicit organizational affiliations and personnel networks. In practical deployment, any two or more dimensions can be flexibly selected and combined based on data source availability and computing resource limitations. For example, only the fund interaction dimension and the bill transaction dimension can be selected to construct a transaction relationship subgraph, or all six dimensions can be integrated to construct a panoramic relationship graph. This configurable dimension combination mechanism allows the technical solution to adapt to data analysis needs in different scenarios, avoiding the problem of the overall solution becoming unusable due to the lack of some data sources.

[0124] In an alternative implementation, step S202 includes at least two of the following steps A1 to A6.

[0125] Step A1: If the dimension is the fund interaction dimension, obtain the settlement direction of the target object relative to the associated object within a preset period and the total settlement transaction amount corresponding to the settlement direction.

[0126] The preset period refers to the time window used to capture interaction data, which defines the timeliness boundary for calculating the association strength. For example, the preset period can be configured as a T+1 day rolling window that looks back from the current time, or it can be configured as a fixed duration window (such as 12 months) starting from the authorization time of the target object.

[0127] By setting a unified time benchmark, the alignment of interactive data from different dimensions on the timeline is ensured, avoiding deviations in the calculation of association strength caused by mismatches in data timeliness.

[0128] The settlement direction is an enumerated field representing the vector attribute of fund flow, used to distinguish whether funds flow from the target object to the related object or from the related object to the target object. For example, it can be defined as "inflow direction" or "outflow direction". The total settlement transaction amount refers to the cumulative value of all valid transaction records under this specific settlement direction within a preset period. For example, when calculating the fund dependence of related object A on target object B, records where the payer is A and the payee is B are selected from the transaction log, and their transaction amounts are summarized as the total settlement transaction amount for the "outflow" direction.

[0129] The total settlement transaction amount reflects the strength of economic ties between entities based on real transaction behavior and is a core indicator for quantifying upstream and downstream relationships in a business ecosystem.

[0130] Step A2: If the dimension is the equity topology dimension, obtain the shareholding relationship of the target object relative to the associated object within the preset period and the shareholding ratio corresponding to the shareholding relationship.

[0131] Among these, shareholding relationship is a directed connection attribute characterizing the capital control chain, clearly identifying who the investor and who the investee are. Shareholding ratio is a numerical parameter quantifying the tightness of this control chain, usually stored as a percentage. For example, if the target entity holds 30% of the shares of a related entity, the extracted interaction data includes the shareholding relationship identifier of "target entity → related entity" and the shareholding ratio of 30%. It should be understood that the shareholding ratio includes not only direct shareholding but, in some implementations, indirect shareholding ratios can also be calculated to more accurately reflect actual control.

[0132] Shareholding ratios transform the equity structure into calculable topological weights, providing data support for identifying group factions and actual controllers.

[0133] Step A3: If the dimension is the personnel employment dimension, obtain the number of related enterprises of the target object and the related object within the preset period, and the wealth level parameters of the personnel who hold cross-appointment positions between the target object and the related object.

[0134] If both the target and associated objects are enterprises, the number of associated enterprises is the total number of existing enterprise customers employed, affiliated with, or bound to the same natural person who holds positions in both the target and associated objects (i.e., cross-employment personnel) within a preset period, used to characterize the person's business association coverage.

[0135] For example, if a person is employed by target company A, related company B, company C, and company D, then the number of related companies for that person is N=4.

[0136] Understandably, the more existing corporate clients an individual is associated with, the broader their business network and the stronger their ability to penetrate across enterprises. This also indicates a higher level of credibility in the implicit relationship between the two companies, which is used to calculate the individual's corporate relationship score.

[0137] The wealth level parameter of the cross-appointed personnel of the target object and related objects is a discrete level identifier obtained by comprehensively mapping the asset size, transaction activity and other multi-source financial attributes of the cross-appointed personnel. For example, it can be divided into multiple gradients such as "basic level", "potential level", "financial management level" and "wealth level".

[0138] For example, if the legal representative of the target company also serves as a director in 5 other companies, and its own asset rating is "wealth level", then the extracted interactive data will be "number of related companies = 5" and "wealth level parameter = wealth level".

[0139] The above approach combines unstructured social networks with the individual's own resource endowments, thus overcoming the limitation of relying solely on transaction data to capture implicit interpersonal connections.

[0140] Step A4: If the dimension is the bill transaction dimension, obtain the seller identifier and the corresponding sales transaction data of the target object relative to the associated object within the preset period, or the buyer identifier and the corresponding input transaction data of the buyer identifier.

[0141] Among them, the seller identifier and buyer identifier are unique markers that distinguish the direction of invoice flow, corresponding to the invoice issuer and the recipient, respectively. Sales transaction data and input transaction data are structured fields attached to the corresponding identifiers, including at least key elements such as the total amount including tax, tax amount, and invoice date.

[0142] For example, when the target object issues an invoice to a related object as a seller, the "seller identifier = target object ID" and the corresponding "sales transaction data = {amount: 1 million, date: 2023-10-01}" are extracted; conversely, if the target object receives an invoice as a buyer, the "buyer identifier = target object ID" and the corresponding "input transaction data = {amount: 1 million, date: 2023-10-01}" are extracted.

[0143] In particular, the invoice data {sales transaction data, input transaction data} has strict symmetry, that is, the sales of one party must correspond to the input of the other party. This inherent logic provides a physical basis for the reverse completion of missing data in subsequent embodiments.

[0144] Step A5: If the dimension is the project bidding cooperation dimension, obtain the bidding identifier and the preset bidding score corresponding to the bidding identifier of the target object relative to the associated object within the preset period, or the winning bid identifier and the preset winning bid score corresponding to the winning bid identifier.

[0145] The bidding and winning bid indicators define the roles of the parties involved in the project collaboration, reflecting the intention and outcome of supply and demand matching. The preset bidding score and preset winning bid score are pre-configured quantitative weights based on role differences, used to distinguish the contribution of different roles to the strength of the association.

[0146] For example, if the target entity publishes a tender notice and the associated entity wins the bid, then extract "Tender Identifier = Target Entity ID" and "Preset Tender Score = C111"; if the associated entity publishes a tender and the target entity wins the bid, then extract "Winning Bid Identifier = Target Entity ID" and "Preset Winning Bid Score = C112".

[0147] Typically, the certainty of cooperation implied by winning a bid is higher than that of a mere bidding intention. Therefore, C112 and C111 can be configured with different values ​​to reflect this difference in strength. This dimension transforms discrete bidding events into comparable relational values, effectively capturing the dynamic connections between objects based on project cooperation.

[0148] Step A6: If the dimension is the organizational hierarchy dimension, obtain the superior identifier of the target object relative to the associated object within the preset period and the preset superior level score corresponding to the superior identifier, or the subordinate identifier and the preset subordinate level score corresponding to the subordinate identifier.

[0149] Among them, the superior and subordinate identifiers are hierarchical relationship markers automatically generated based on the organizational structure topology, used to represent the administrative or managerial subordinate relationships between objects. The preset superior level score and preset subordinate level score are differentiated weights set for different subordinate directions.

[0150] For example, if the target object is determined to be the head office of the related object, then "superior identifier = target object ID" and "preset superior level score = D111" are extracted; if the target object is a branch office, then "subordinate identifier = target object ID" and "preset subordinate level score = D112" are extracted.

[0151] Because the head office usually has a stronger control over its branches than the branches have a reverse influence on the head office, the assignment of values ​​to D111 and D112 often exhibits asymmetry.

[0152] This step transforms text-based organization name information into structured topological hierarchy data, solving the technical problems of lagging organizational structure tree maintenance and difficulty in updating in traditional methods.

[0153] By acquiring refined data across the six dimensions described above, this embodiment transforms the originally scattered and heterogeneous raw information into a standardized and semantically clear set of interactive data. This not only provides high-quality input features for subsequent interval mapping and gradient quantization, but more importantly, by clarifying the specific composition and physical meaning of the interactive data in each dimension, it establishes a complete technical link from the underlying data to the upper-level correlation strength calculation, ensuring the reproducibility and objectivity of the entire calculation process.

[0154] In an alternative implementation, step A6 includes steps A61 to A63.

[0155] Step A61: Determine the hierarchical relationship between the target object and the associated object based on the subject name of the target object and the subject name of the associated object.

[0156] The subject name refers to the unique textual identifier used by the object in the legal registration information or business system, such as the full name of a company's business registration or the approved name of a public institution.

[0157] Hierarchical membership is a topological connection attribute that represents whether there is an administrative or managerial subordinate status between objects. Its data representation is usually directed edges in a directed graph model or non-zero elements in an adjacency matrix. Specifically, by comparing the subject names of two objects, it is determined whether they meet preset naming rules or containment relationships, thereby inferring potential hierarchical connections. For example, if the subject name of the target object is "XX Technology Co., Ltd." and the subject name of the associated object is "XX Technology Co., Ltd. Shenzhen Branch", a hierarchical membership relationship can be determined based on the core font and suffix features contained in the subject names.

[0158] The above steps transform unstructured text information into structured topological relationship data, avoiding reliance on externally maintained organizational structures and significantly improving the timeliness and coverage of data processing.

[0159] Step A62: If the hierarchical relationship indicates that the target object is a subordinate object, configure a subordinate identifier and a corresponding preset subordinate level score for the target object.

[0160] The subordinate identifier is an enumerated field or Boolean flag used to clearly mark the subordinate or managed position of a target object in a hierarchical relationship. For example, it can be defined as "SUB" or the value "0". The preset subordinate level score is a quantitative weighting parameter configured for the reverse association contribution of a subordinate object to its superior object, denoted as E112. Specifically, when a target object is determined to be a branch or subsidiary of a related object, a subordinate identifier is automatically generated, and the preset subordinate level score E112 is assigned to that target object. For example, if the target object "Shenzhen Branch" is determined to be a subordinate of the related object "Headquarters", then the target object is assigned the subordinate identifier "SUB" and a score E2=0.5. Since the influence of a subordinate object on its superior object is usually weaker than the control of a superior object over its subordinate object, the value of E112 is generally less than the superior level score. This differentiated assignment accurately reflects the asymmetry of power flow in the organizational topology.

[0161] Step A63: If the hierarchical relationship indicates that the target object is a superior object, configure the superior identifier and the corresponding preset superior level score for the target object.

[0162] The superior identifier is an enumerated field or Boolean flag used to clearly mark the dominant or managerial position of the target object in the hierarchical relationship. For example, it can be defined as "SUP" or the value "1". The preset superior level score is a quantitative weighting parameter configured for the positive correlation contribution of the superior object to the subordinate object, denoted as E111. Specifically, when the target object is determined to be the headquarters or parent company of the related object, a superior identifier is automatically generated, and the preset superior level score E111 is assigned to the target object. For example, if the target object "Headquarters" is determined to be the superior of the related object "Shenzhen Branch", then the target object is assigned the superior identifier "SUP" and the score E111 = 1.0.

[0163] In practical applications, E111 and E112 can be flexibly configured according to business scenarios. For example, in scenarios that emphasize the overall risk transmission of the group, E111 can be set to a higher value to highlight the radiating effect of the core enterprise; while in scenarios that emphasize the activity of grassroots outlets, the weight of E112 can be appropriately increased.

[0164] This step transforms qualitative hierarchical affiliation judgments into quantitative differential scores, enabling hierarchical affiliation relationships to be integrated and calculated with interaction data from other dimensions under the same scale. This solves the problem that traditional methods can only qualitatively determine the existence of affiliation relationships but cannot accurately measure the strength of associations.

[0165] In an alternative implementation, step A1 includes steps A11 to A14.

[0166] Step A11: Filter out all fund settlement interaction records between the target object and related objects within the preset period.

[0167] Step A12: Obtain the transaction fund flow corresponding to each of the multiple fund settlement interaction records.

[0168] Among them, the flow of transaction funds is the settlement direction, which is divided into the expenditure direction of the target object to the related object and the income direction of the related object to the target object.

[0169] Step A13: Classify and aggregate multiple fund settlement interaction records according to different settlement directions to obtain expenditure sets and income sets.

[0170] The expenditure set includes multiple fund settlement interaction records where the settlement direction is payment from the target object to the associated object; the income set includes multiple fund settlement interaction records where the settlement direction is payment from the associated object to the target object.

[0171] Step A14: Calculate the cumulative transaction amount for the expenditure set and the cumulative transaction amount for the income set respectively.

[0172] For example, the total settlement transaction amount = the cumulative transaction amount of the expenditure set + the cumulative transaction amount of the income set.

[0173] For example, the total settlement transaction amount is {the cumulative transaction amount of the expenditure set, the cumulative transaction amount of the income set}.

[0174] In an alternative implementation, step A2 includes steps A21 to A24.

[0175] Step A21: Collect all equity change records, equity filing data, and industrial and commercial equity topology data of the target object and related objects within the preset period.

[0176] Equity change records refer to all historical data on equity changes that the target or related parties have filed with the market supervision department within a preset period, including equity transfers, capital increases, equity transfers, shareholder changes, and other change events.

[0177] For example, the equity change record includes the change time, the shareholding ratio before the change, the shareholding ratio after the change, and the information of the changing entity, which is used to trace the current effective equity status of the enterprise.

[0178] Equity registration data refers to the static equity registration information of the target or related parties that is currently valid and has been formally registered and retained by the industrial and commercial system. It is the latest valid list of legal shareholders, registered shareholding ratios, and capital contribution relationship registration data of the target or related parties, and is the authoritative benchmark data for determining the current shareholding relationship.

[0179] Industrial and commercial equity topology data refers to the overall equity relationship topology data between target objects and related objects, which is compiled based on industrial and commercial investment, shareholding, and controlling relationships. It includes information on direct shareholding, indirect shareholding, cross shareholding, and external investment links of enterprises, and is used to depict the capital relationship topology and hierarchical structure between two enterprises.

[0180] Step A22: Based on industrial and commercial equity topology data and equity registration data, obtain the shareholding relationship between the target object and related objects.

[0181] If the target entity is registered as a direct shareholder of a related party, or if a related party is registered as a direct shareholder of the target entity, then the shareholding relationship indicates that there is a direct equity investment between the target entity and the related party; if neither of the above two conditions is met, then the shareholding relationship indicates that there is no direct equity investment between the target entity and the related party.

[0182] If a shareholding relationship indicates the absence of direct equity investment, it is marked as "no shareholding relationship." If a shareholding relationship indicates the presence of direct equity investment, it is marked as "shareholding relationship."

[0183] Before determining the direct equity investment behavior of the target and related parties, the collected equity filing data and equity change records are first standardized, cleaned and filtered. All registered shareholder names are matched and verified by a pre-set non-related keyword library. Objects whose names contain keywords such as bank, shareholder, and circulating shares are uniformly determined as non-related shareholders and removed.

[0184] Non-related shareholders are mostly secondary market shareholders, financial institutions, and financial investment platforms. They only engage in passive financial investment activities and have no substantial operational ties or industrial synergies with the target companies. By filtering non-related shareholders, invalid equity interference data can be eliminated, and misjudgments of false capital relationships can be avoided. Only valid shareholders with real operational attributes are retained to participate in equity relationship verification, which effectively improves the accuracy and authenticity of shareholding relationship identification and ensures that equity dimension data can objectively reflect the substantial capital binding relationship between enterprises.

[0185] Step A23: If the shareholding relationship indicates a direct equity investment, identify the effective shareholding value of {investor entity target object → investee entity related object} and {investor entity related object → investee entity target object} from the shareholding change record.

[0186] Step A24: Based on the effective shareholding values ​​of {investor entity target object → investee entity related object} and {investor entity related object → investee entity target object}, calculate the shareholding ratio of {investor entity target object → investee entity related object} and the shareholding ratio of {investor entity related object → investee entity target object}.

[0187] In an alternative implementation, step A3 includes steps A31 to A34.

[0188] Step A31: Within a preset period, collect all personnel employment registration data, personnel qualification file data, and personnel employment information of related companies corresponding to the target object, as well as all personnel employment registration data, personnel qualification file data, and personnel employment information of related companies corresponding to the associated object.

[0189] All personnel employment registration data refers to the personnel employment registration data that has been filed with the industry and commerce authorities and publicized by enterprises within a preset period, including official registration information such as personnel name, unique identifier, employing company, job position, and start and end dates of employment.

[0190] Personnel qualification file data refers to the archived personal qualification data of personnel stored in the customer resource system and private banking customer system, which includes structured file information such as customer level, asset qualification, wealth rating, and customer classification.

[0191] Personnel-related corporate employment information refers to all external employment information and affiliated company information of the same person in other companies, excluding the current target and related parties, where they serve as shareholders, senior executives, legal representatives, or key personnel within a preset period.

[0192] Step A32: By comparing the information collected in Step A31 with the unique identifier of the personnel's ID card or real-name information, cross-appointment personnel who simultaneously hold positions with both the target object and related objects can be screened out.

[0193] Step A33: Iterate through each cross-appointed person, count the total number of valid corporate entities that the cross-appointed person holds, controls, or is affiliated with, and obtain the corresponding number of related companies.

[0194] Step A34: Retrieve the personal qualification level and private banking client level data corresponding to the cross-appointed personnel, and match and obtain the wealth level parameters corresponding to the cross-appointed personnel according to the preset wealth level gradient standard.

[0195] In an alternative implementation, step A4 includes steps A41 to A44.

[0196] Step A41: Within a preset period, collect all transaction details of bills between the target object and related objects.

[0197] For example, the detailed data of a bill transaction includes key fields such as the bill issuance time, the seller's information, the buyer's information, the bill type, and the total amount including tax.

[0198] Step A42: Based on the transaction details of the bill, distinguish the seller's identity from the buyer's identity from the seller's identity from the buyer's identity.

[0199] Step A43: If the target object is the invoicing party of the transaction, mark it with the seller identifier and collect all the sales transaction data corresponding to the seller identifier.

[0200] For example, sales transaction data includes the number of sales invoices, the total amount of all types of sales invoices including tax, and the transaction time.

[0201] Step A44: If the target object is the invoice recipient, mark it with a buyer identifier and collect all the input transaction data corresponding to that buyer identifier.

[0202] For example, input transaction data includes the number of input invoices, the total amount of all types of input invoices including tax, and the transaction time.

[0203] In an alternative implementation, step A5 includes steps A51 to A54.

[0204] Step A51: Within a preset period, collect all project bidding and tendering announcement data between the target object and related objects.

[0205] Project bidding and tendering announcement data consists of original project information such as bidding announcements, winning bid announcements, and competitive negotiation announcements released through public channels. It includes key fields such as project name, bidding entity, winning bid entity, announcement time, and project category.

[0206] Step A52: Identify the roles and collaborative behaviors of the target and related parties in the project based on the project bidding and tendering announcement data.

[0207] Step A53: If it is identified that the target object is conducting bidding activities as the project initiator and tender issuer in the project, then assign a bidding identifier to the target object and match and retrieve the preset bidding score.

[0208] Step A54: If the target object is identified as the contractor who successfully won the bid in the project, then assign a winning bid identifier to the target object and retrieve the preset winning bid score.

[0209] Among them, the preset bidding score and the preset winning bid score are fixed quantitative scores pre-configured according to the project cooperation attributes, which are used to characterize the strength of the correlation between the two parties' industrial project cooperation.

[0210] In an alternative implementation, step A61 includes steps A611 to A612.

[0211] Step A611: If the main name of the associated object is formed by concatenating the main name of the target object with a preset string, the target object is determined to be the parent object.

[0212] Among them, the preset string belongs to the set of suffix identifiers used to represent the hierarchical attributes of an organization.

[0213] The suffix identifier set contains a configuration list of multiple enumeration elements, such as "branch office", "branch", "sub-branch", "office", "factory", "branch hospital", etc.

[0214] In actual deployment, system administrators can dynamically expand the set of suffix identifiers according to the industry characteristics of the data to be analyzed. For example, in the financial data analysis scenario, elements such as "business department" and "branch office" can be added, and in the manufacturing scenario, elements such as "workshop" and "work area" can be added.

[0215] The aforementioned configurable set of suffix identifiers allows the technical solution to adapt to changes in organizational naming conventions across different fields and at different times, without requiring modifications to the core algorithm code.

[0216] The "concatenation" operation refers to the concatenation operation that appends characters to the end of the subject name, satisfying the mathematical relationship "subject name + preset string = string B".

[0217] It should be understood that the splicing here is a strict sequential connection, without including the insertion of other characters in the middle, in order to ensure semantic accuracy.

[0218] Step A612: If the main name of the target object is formed by concatenating the main name of the associated object with the preset string, the target object is determined to be a subordinate object.

[0219] For each pair of target and associated objects whose relationship needs to be determined, forward and reverse matching tests are performed. In the forward matching test, each element in the suffix identifier set is traversed and appended to the target object's main name to generate a candidate name sequence. The main name of the associated object is then checked against any element in this sequence. If an equality exists, the associated object's name is determined to be derived from the target object's name, thus identifying the target object as the parent object and the associated object as the child object. Conversely, in the reverse matching test, a pre-defined string is appended to the associated object's main name. If the generated candidate name is identical to the target object's main name, the target object is identified as the child object and the associated object as the parent object. If neither forward nor reverse matching results in a match, it is determined that there is no direct hierarchical relationship between the two objects under the current rules.

[0220] To further improve the accuracy of identification and prevent misjudgments due to name coincidences, this embodiment also introduces exclusion rules as a necessary supplement to the matching logic. Specifically, before performing the above-mentioned concatenation matching, it will detect whether the subject name contains specific interfering features. For example, if the subject name of the associated object starts with the target object's name and contains the suffix "branch company", but its name also contains a specific administrative division prefix (such as "Shenzhen", "Shanghai", etc.) and the prefix is ​​located between the target object's subject name and the suffix, or the object is marked as having independent legal person status in the business registration data, then the matching result should be excluded and it should not be recognized as a hierarchical relationship.

[0221] This is because, in practical applications, subsidiaries with independent geographical prefixes often have independent legal person status, and their relationship with the parent company is based on equity topology rather than organizational hierarchy. By introducing such exclusion rules, this embodiment effectively distinguishes genuine branches from related companies with similar names, ensuring the objectivity and accuracy of subsequent differentiated scoring.

[0222] For example, suppose the set of suffix identifiers is {"branch company", "sub-branch"}, the entity name of associated object A is "Xingchen Technology Co., Ltd.", and the entity name of target object B is "Xingchen Technology Co., Ltd. Beijing Branch". First, a forward matching is performed, concatenating "branch company" to the entity name of associated object A to obtain "Xingchen Technology Co., Ltd. Branch". It is found that this is not completely consistent with the entity name of target object B (it includes the word "Beijing"). At this point, if no exclusion rules or lenient matching mode are configured, the match may fail; however, if a preprocessing rule that allows ignoring regional modifiers is configured, or if the entity name of target object B is actually "Xingchen Technology Co., Ltd. Branch", the match succeeds, and associated object A is determined to be the parent object of target object B. Similarly, if the name of associated object C is "Xingchen Technology Trading Co., Ltd.", although it contains the words "Xingchen Technology", it does not end with any preset string, therefore it will not be mistakenly judged as a subordinate object of associated object A. This series of rigorous logical judgments is entirely executed automatically by the computer program without human intervention.

[0223] This embodiment transforms the abstract membership determination into a standardized string concatenation and verification algorithm, and combines it with exclusion rules for noise filtering, thereby achieving automated and high-precision construction of organizational structure topology. This not only solves the problems of delayed updates and incomplete coverage caused by manual maintenance of organizational structures in traditional methods, but more importantly, it transforms unstructured text naming information into computable and verifiable structured topological data, providing a solid and reliable data foundation for subsequent steps of assigning differentiated association strength based on membership direction.

[0224] In an alternative implementation, step S204 includes steps B1 to B3.

[0225] Step B1: Determine the target numerical range of the interaction data of the target object relative to the associated object in this dimension.

[0226] The numerical range refers to a predefined set of numerical ranges used to segment continuous or discrete interactive data.

[0227] The division of numerical intervals is not arbitrarily set, but rather a piecewise function domain constructed based on the statistical distribution characteristics of the data or thresholds derived from business experience. Specifically, the extracted interactive data can be compared with the boundary values ​​of the numerical intervals to locate the target numerical interval to which the interactive data belongs.

[0228] For example, in the dimension of fund interaction, if the total amount of settlement transactions extracted is S, and the boundary values ​​of multiple numerical intervals are T respectively. 11 T 12 T 13 T 14 T15 And T 11 <T 12 <T 13 <T 14 <T 15 Multiple numerical intervals can be obtained by dividing the boundary values ​​from small to large. The actual settlement transaction amount S is compared with each boundary value in turn to determine the target numerical interval into which S falls.

[0229] Similarly, the interaction data of other dimensions such as equity topology, personnel appointment, bill transactions, project bidding and cooperation, and organizational hierarchy are all determined by using this interval matching logic to complete the determination of the target value range.

[0230] Step B2: Find the target gradient level value corresponding to the target numerical range from the correspondence between the numerical range and the gradient level value.

[0231] The gradient level value is a discretized hierarchical identifier representing the relative strength of the association. Its function is to compress continuous numerical intervals into a finite number of enumerable state labels. The correspondence between numerical intervals and gradient level values ​​is usually stored in the system in the form of a look-up table or a mapping function, forming the core data structure of the preset interval mapping rules.

[0232] Specifically, after determining the target numerical range, the corresponding gradient level value is obtained through index lookup or function calculation.

[0233] Step B3: Based on the target gradient level value and the preset dimension baseline coefficient, calculate the relationship value of that dimension.

[0234] For example, if the total settlement transaction amount S falls within the target value range (-∞, T) 11 ), looking up the table, the corresponding gradient level value is 0, and the relationship value Vs = K1 × 0 (K1 is the preset dimension benchmark coefficient of the capital interaction dimension); if the total settlement transaction amount S falls into the target value range [T 11 , T 12 ), look up the table to get the corresponding gradient level value of 1, and the relationship value Vs = K1 × 1; if the total settlement transaction amount S falls into the target value range [T 12 , T 13 ), looking up the table, the corresponding gradient level value is 1.5, and the relationship value Vs = K1 × 1.5; if the total settlement transaction amount S falls within the target value range [T 13 T 14 ), looking up the table, the corresponding gradient level value is 2, and the relationship value Vs = K1 × 2; if the total settlement transaction amount S falls within the target value range [T 14 , T 15), looking up the table, the corresponding gradient level value is 2.5, and the relationship value Vs = K1 × 2.5; if the total settlement transaction amount S falls within the target value range [T 15 (+∞), look up the table to get the corresponding gradient level value of 3, and the relationship value Vs=K1×3.

[0235] For example, the preset dimensional baseline coefficient K1 can be determined according to the actual situation, such as 0.1, 1, 10, 100, etc. 11 To T 15 The value can be determined based on the actual business requirements.

[0236] For example, if the shareholding ratio falls within the target value range [R1, +∞), the corresponding gradient level value is 3, and the relationship value Vr = K2 × 3 (K2 is the preset dimension baseline coefficient of the equity topology dimension); if the shareholding ratio falls within the target value range [R2, R1), the corresponding gradient level value is 2, and the relationship value Vr = K2 × 2; if the shareholding ratio falls within the target value range [R3, R2), the corresponding gradient level value is 1, and the relationship value Vr = K2 × 1; if the shareholding ratio falls within the target value range (-∞, R3), the corresponding gradient level value is 0, and the relationship value Vr = K2 × 0.

[0237] The boundary values ​​R1, R2, and R3 can be determined according to the actual situation. For example, the boundary values ​​R1, R2, and R3 can be set to 60%, 40%, and 20%, respectively. The preset dimension benchmark coefficient K2 of the equity topology dimension can be determined according to the actual situation.

[0238] For example, if the wealth level parameter corresponding to a cross-appointed employee falls within the target value range (-∞, basic level), the corresponding gradient level value obtained from the table is 1, and the relationship value V p =1×C p (C) p (This refers to the preset baseline coefficient for the personnel's job position dimension). If the wealth level parameter corresponding to the cross-job personnel falls within the target value range [basic level, potential level], the corresponding gradient level value obtained by looking up the table is 1.5, and the relationship value V is... p =1.5×C p If the wealth level parameter corresponding to the cross-appointed personnel falls within the target value range [potential level, financial management level], the corresponding gradient level value obtained by looking up the table is 2, and the relationship value V is... p =2×C p If the wealth level parameter corresponding to the cross-appointed personnel falls within the target value range [financial management level, wealth level], the corresponding gradient level value obtained from the table is 2.5, and the relationship value V is... p =2.5×C pIf the wealth level parameter corresponding to the cross-appointed personnel falls within the target value range [wealth level, +∞), the corresponding gradient level value obtained from the table is 3, and the relationship value V is... p =3×C p .

[0239] The preset benchmark coefficients for personnel job positions can be determined based on the actual situation.

[0240] For example, if the total amount of bill transactions (the sum of sales transactions and input transactions) S inv Falling into the target value range (-∞, T) 21 The corresponding gradient level value is 0, and the relation value V is obtained by looking up the table. inv =K3×0 (K3 is the preset benchmark coefficient for the bill transaction dimension); if the total amount of bill transactions S inv Falling into the target value range [T 21 ,T 22 ), looking up the table, the corresponding gradient level value is 1, and the relation value V inv =K3×1; If the total transaction amount of bills is S inv Falling into the target value range [T 22 ,T 23 The corresponding gradient level value is 1.5, and the relational value V is obtained from the table. inv =K3×1.5; If the total transaction amount of bills is S inv Falling into the target value range [T 23 ,T 24 The corresponding gradient level value is 2, and the relation value V is obtained from the table lookup. inv =K3×2; If the total transaction amount of bills is S inv Falling into the target value range [T 24 ,T 25 The corresponding gradient level value obtained from the table is 2.5, and the relationship value V is... inv =K3×2.5; If the total transaction amount of bills is S inv Falling into the target value range [T 25 (+∞), looking up the table, the corresponding gradient level value is 3, and the relation value V inv =K3×3.

[0241] Wherein, boundary value T 21 T 22 T 23 T 24 T 25 It can be flexibly configured according to the actual business scenario.

[0242] For example, the preset benchmark coefficient for the transaction dimension can be determined based on the actual situation, such as 0.1, 1, 10, 100, etc.

[0243] It is important to emphasize that the gradient level value here is not a subjective score, but a quantitative parameter determined based on data-driven or statistical laws. Its value only represents the ordinal relationship of relative strength.

[0244] Existing technologies, when faced with missing interactive data, typically employ either default value filling or direct discarding of the data. The former introduces human noise, distorting the calculation of association strength, while the latter results in the permanent loss of valid information. This embodiment aims to clarify, from the underlying data processing principles, how to utilize the inherent symmetry of interactive behavior to deduce the missing data of the target object through the existing data logic of the associated objects. This establishes a strong technical barrier distinct from conventional data cleaning techniques, proving that the completion process is based on the recovery of objective data patterns rather than subjective assumptions. In an optional implementation, step S203 includes steps C1 to C2.

[0245] Step C1: If the second interaction direction is the second identifier, set the first interaction direction to the first identifier.

[0246] Among them, the same interaction behavior refers to a single event or state change that objectively exists in the physical world or business logic and involves at least two participating entities, such as a fund transfer, the issuance of an invoice, or the delivery of goods.

[0247] Conversely, the interaction direction refers to the mutually exclusive and complementary state descriptions presented when the same interaction behavior is recorded from the perspectives of different participating subjects.

[0248] Specifically, the interaction direction is typically defined as the edge direction or an enumerated type field in a directed graph. For example, for the interaction behavior of "invoice issuance," if recorded from the perspective of the invoice issuer, the interaction direction is "sales" (i.e., the first interaction direction); while from the perspective of the invoice recipient, the interaction direction for the same invoice must be "income" (i.e., the second interaction direction). These two directions are logically strictly contradictory and unified; there is no intermediate state that is neither sales nor income, nor is there a valid situation where one party has sales while the other party has no corresponding income. Therefore, when it is detected that the target object's interaction data in the first interaction direction, such as income transaction data, is missing, but the related object's interaction data in the second interaction direction, such as sales transaction data, exists, this inherent symmetry principle can be used to automatically flip or remap the sales transaction data in the related object to the income transaction data in the target object. This step realizes the logical transformation of the data perspective, laying the semantic foundation for subsequent data content migration.

[0249] Step C2: Use the interaction data corresponding to the second interaction direction as the interaction data corresponding to the first interaction direction.

[0250] The interaction data corresponding to the first interaction direction refers to the complete set of attributes recorded by the target object from its own perspective regarding the interaction behavior, including but not limited to structured fields such as transaction amount, occurrence time, target object identifier, and tax rate.

[0251] The interaction data corresponding to the second interaction direction refers to the set of corresponding attributes that the associated object should record from its own perspective regarding the same interaction behavior.

[0252] Since both describe the same objective fact, apart from the direction indicators being mirror images of each other, the other core attribute values ​​should theoretically be completely identical or have a definite conversion relationship.

[0253] Specifically, when performing data completion, key numerical fields from the associated object are either directly copied or converted using predefined functions and then assigned to the corresponding fields of the target object. For example, if the sales transaction data of the associated object contains "total price including tax = 1 million yuan", "invoice date = 2023-10-01", and "tax amount = 60,000 yuan", then these three field values ​​will be filled into the newly created input transaction data of the target object.

[0254] It should be understood that this data migration is not a simple copy, but is accompanied by a change in data ownership, that is, the main label of the data is switched from the associated object to the target object, while the timestamp and source mark of the original data are retained for traceability.

[0255] To more clearly illustrate the universality of the above reverse completion logic, we will use two typical interaction scenarios as examples for explanation below.

[0256] In a fund transfer scenario, if target object B is not authorized, resulting in missing "fund inflow" data, but associated object A is authorized and has "fund outflow" data, it can be identified that "A←B" and "B→A" belong to the opposite direction of the same transfer behavior. Then, the outflow amount and outflow time of associated object A can be directly filled in to match the inflow amount and inflow time of target object B.

[0257] Through the above mechanism, this embodiment successfully reconnected the data link that was originally broken due to permission isolation, enabling the correlation strength calculation to be based on complete bidirectional interactive information, significantly improving the coverage of the map construction and the confidence of the calculation results, and fundamentally solving the problem of blind spots in correlation analysis caused by the lack of unilateral data.

[0258] While reverse derivation based on the symmetry of interactive behavior can solve the data missing problem, indiscriminately using all historical data for completion may introduce outdated, invalid, or noisy data irrelevant to the current analysis point in time, leading to distortion of the correlation strength calculation results. The main technical problem addressed in this embodiment is how to establish a data quality control mechanism during automated completion, ensuring through dual timeliness checks that the data used for completion not only meets the time range requirements of business analysis but also possesses sufficient data freshness, thereby improving the rigor and practicality of the overall calculation scheme. In an optional implementation, the interactive data in the invoice transaction dimension also includes the invoice issuance time and the invoice data entry time, further comprising the following steps: if the invoice issuance time is within a preset time window, and the time difference between the current time and the invoice data entry time is less than or equal to a preset storage timeliness threshold, perform the interaction data completion of the first interaction direction based on the second interaction direction of the associated object in the target dimension.

[0259] The start time of the preset time window is the authorization time of the target object, and the length of the preset time window is a preset duration.

[0260] The invoice issuance time refers to the legally mandated date on the invoice or document when the transaction occurred or the invoicing was completed, representing the actual sequence of events at the business level. The invoice data entry time refers to the system timestamp when the invoice data was collected and written to the system database, reflecting the physical sequence of the data entering the computing environment. These two time attributes characterize the timeliness of the data from two orthogonal dimensions: business validity and data freshness.

[0261] Specifically, the preset time window is used to define the effective range of business time periods relevant to the association strength calculation. The start time of this window is defined as the authorization time of the target object, i.e., the moment the target object is formally included in the analysis scope or a data usage agreement is signed; the length of the preset time window is a preset duration, such as 12 months, 6 months, or 24 months. The physical meaning of this design is that only interactions occurring after the target object has been authorized are considered valid sources of association characteristics; historical data prior to authorization is automatically filtered out due to a lack of compliance basis or mismatch with the analysis objectives. For example, if the target object completed authorization on January 1, 2023, and the preset duration is configured as 12 months, then the preset time window would be [2023-01-01, 2024-01-01]. Only when the invoice issuance time of the associated object falls within this range is the invoice data considered to have business-level supplementary value.

[0262] Furthermore, the preset storage time threshold is the maximum tolerable limit for data staleness, defining the longest allowed interval from data entry to its use. Specifically, the system acquires the current time in real time and calculates the difference between it and the invoice data entry time (i.e., ΔT = T). current -T insert Data is considered usable only if the difference is less than or equal to a preset storage expiration threshold. For example, if the preset storage expiration threshold is set to 3 years, and a certain invoice data was entered into the database 5 years ago, even if its issuance date happens to fall within the preset time window (this may be due to batch entry of historical data), it will be rejected for reverse completion because the data is too outdated. This mechanism effectively prevents real-time computing from being interfered with by data backlogged in the underlying data warehouse or delayed synchronization data.

[0263] When processing the interaction data of each related object to be completed, it is necessary to sequentially or in parallel pass the business time window verification and data storage timeliness verification. Only when both verifications pass will the reverse completion process in the aforementioned embodiment be triggered. For example, assuming the current time is October 1, 2023, the authorization time of the target object is January 1, 2023, the preset duration is 12 months, and the preset storage timeliness threshold is 3 years. At this time, if the related object holds an invoice with an issuance date of May 15, 2023 (within the 12-month window after authorization) and the invoice data entry time is May 16, 2023 (less than 1 year from the current time, much less than the 3-year threshold), then the invoice data passes the double verification and can be used to complete the corresponding dimension data of the target object. Conversely, if another invoice is issued in June 2022 (earlier than the authorization time), or the issuance time is compliant but the invoice data entry time is 2019 (more than 3 years ago), they will be automatically removed.

[0264] The above describes a method for calculating object association strength based on multi-source heterogeneous data, provided by embodiments of this application. The following describes the apparatus for performing the above-described method for calculating object association strength based on multi-source heterogeneous data.

[0265] Please see Figure 3 , Figure 3 This is a schematic diagram of a device for calculating the object association strength based on multi-source heterogeneous data, provided as an embodiment of this application. Figure 3 As shown, the object association strength calculation device based on multi-source heterogeneous data includes:

[0266] Data acquisition module 301 is used to acquire multi-source heterogeneous data;

[0267] The feature extraction module 302 is used to obtain interaction data of the target object relative to the associated object in multiple dimensions from the multi-source heterogeneous data;

[0268] The data reverse completion module 303 is used to complete the interaction data of the first interaction direction based on the interaction data of the associated object in the second interaction direction of the target dimension if the interaction data of the target object in the first interaction direction of the target dimension is missing. The first interaction direction and the second interaction direction are opposite interaction directions of the same interaction behavior. The target dimension is any one of the multiple dimensions.

[0269] The scoring module 304 is used to calculate the relationship value between the target object and the associated object in each dimension based on the interaction data of the target object relative to the associated object in the dimension and the preset interval mapping rule.

[0270] The multidimensional fusion module 305 is used to calculate the association strength value between the target object and the associated object based on the relationship value between the target object and the associated object in each dimension.

[0271] In one optional implementation, the multiple dimensions include at least two of the following: fund interaction dimension, equity topology dimension, personnel appointment dimension, bill transaction dimension, project bidding cooperation dimension, and organizational hierarchy dimension; the feature extraction module includes at least two of the following:

[0272] The first acquisition unit is used to acquire, if the dimension is the fund interaction dimension, the settlement direction of the target object relative to the associated object within a preset period and the total settlement transaction amount corresponding to the settlement direction;

[0273] The second acquisition unit is used to acquire, if the dimension is the equity topology dimension, the shareholding relationship of the target object relative to the associated object within the preset period and the shareholding ratio corresponding to the shareholding relationship;

[0274] The third acquisition unit is used to acquire, if the dimension is the personnel employment dimension, the number of related enterprises of the target object and the associated object within the preset period, and the wealth level parameters of the personnel with cross-employment of the target object and the associated object;

[0275] The fourth acquisition unit is used to acquire, if the dimension is the bill transaction dimension, the seller identifier and the corresponding sales transaction data of the target object relative to the associated object within the preset period, or the buyer identifier and the corresponding input transaction data of the buyer identifier;

[0276] The fifth acquisition unit is used to acquire, if the dimension is the project bidding cooperation dimension, the bidding identifier of the target object relative to the associated object within the preset period and the preset bidding score corresponding to the bidding identifier, or the winning bid identifier and the preset winning bid score corresponding to the winning bid identifier;

[0277] The sixth acquisition unit is used to acquire, if the dimension is the organizational hierarchy membership dimension, the superior identifier of the target object relative to the associated object within the preset period and the preset superior level score corresponding to the superior identifier, or the subordinate identifier and the preset subordinate level score corresponding to the subordinate identifier.

[0278] In one optional implementation, the sixth acquisition unit includes:

[0279] The first determining subunit is used to determine the hierarchical relationship between the target object and the associated object based on the subject name of the target object and the subject name of the associated object;

[0280] The first setting subunit is used to configure a lower-level identifier and a corresponding preset lower-level score for the target object if the hierarchical membership relationship indicates that the target object is a lower-level object.

[0281] The second setting subunit is used to configure a superior identifier and a corresponding preset superior level score for the target object if the hierarchical relationship indicates that the target object is a superior object.

[0282] In one alternative implementation, the first determined subunit includes:

[0283] The first determining submodule is used to determine that the target object is a parent object if the main name of the associated object is composed of the main name of the target object and a preset string.

[0284] The second determining submodule is used to determine that the target object is a subordinate object if the main name of the target object is formed by concatenating the main name of the associated object with the preset string.

[0285] In one optional implementation, the scoring module includes:

[0286] The first determining unit is used to determine the target numerical range of the interaction data of the target object relative to the associated object in this dimension;

[0287] The first lookup unit is used to look up the target gradient level value corresponding to the target numerical range from the correspondence between the numerical range and the gradient level value.

[0288] The second determining unit is used to calculate the relationship value of the dimension based on the target gradient level value and the preset dimension benchmark coefficient.

[0289] In one optional implementation, the data inverse completion module includes:

[0290] The first setting unit is configured to set the first interaction direction to the first identifier if the second interaction direction is the second identifier.

[0291] The second setting unit is used to set the interaction data corresponding to the second interaction direction as the interaction data corresponding to the first interaction direction.

[0292] In one optional implementation, the target dimension is the bill transaction dimension, and the interaction data in the second interaction direction includes the seller identifier and the corresponding sales transaction data within a preset period, or the buyer identifier and the corresponding input transaction data; the first setting unit includes:

[0293] The third setting subunit is used to set the first identifier to the buyer identifier if the second identifier is the seller identifier.

[0294] The fourth setting subunit is used to set the first identifier to the seller identifier if the second identifier is the buyer identifier;

[0295] The second setting unit includes:

[0296] The fifth setting subunit is used to, if the second identifier is the seller identifier, take the sales transaction data in the interaction data of the second interaction direction as the input transaction data in the interaction data of the first interaction direction;

[0297] The sixth setting subunit is used to, if the second identifier is the buyer identifier, take the input transaction data in the interaction data of the second interaction direction as the output transaction data in the interaction data of the first interaction direction.

[0298] In an optional implementation, the interactive data of the invoice transaction dimension also includes the invoice issuance time and the invoice data entry time; it also includes:

[0299] The triggering module is used to execute the data reverse completion module if the invoice issuance time is within a preset time window and the time difference between the current time and the invoice data entry time is less than or equal to a preset storage time threshold.

[0300] The start time of the preset time window is the authorization time of the target object, and the length of the preset time window is a preset duration.

[0301] This application also provides an electronic device in its embodiments. (See reference...) Figure 4The diagram illustrates a structural schematic of an electronic device suitable for implementing the object association strength calculation method based on multi-source heterogeneous data in the embodiments of this application. The electronic device in the embodiments of this application may include, but is not limited to, fixed terminals such as mobile phones, laptops, PDAs (personal digital assistants), PADs (tablet computers), desktop computers, etc. Figure 4 The electronic device shown is merely an example and should not impose any limitation on the functionality and scope of use of the embodiments of this application.

[0302] like Figure 4 As shown, the electronic device may include a processing unit (e.g., a central processing unit, a graphics processing unit, etc.) 401, which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 402 or a program loaded from a storage device 408 into a random access memory (RAM) 403. When the electronic device is powered on, the RAM 403 also stores various programs and data required for the operation of the electronic device. The processing unit 401, ROM 402, and RAM 403 are interconnected via a bus 404. An input / output (I / O) interface 405 is also connected to the bus 404.

[0303] Typically, the following devices can be connected to I / O interface 405: input devices 406 including, for example, touchscreens, touchpads, keyboards, mice, cameras, microphones, accelerometers, gyroscopes, etc.; output devices 407 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 408 including, for example, memory cards, hard drives, etc.; and communication devices 409. Communication device 409 allows electronic devices to communicate wirelessly or wiredly with other devices to exchange data. Although Figure 4 Electronic devices with various devices are shown, but it should be understood that it is not required to implement or have all of the devices shown. More or fewer devices may be implemented or have alternatively.

[0304] This application also provides a computer program product, including computer-readable instructions, which, when executed on an electronic device, cause the electronic device to implement any of the object association strength calculation methods based on multi-source heterogeneous data provided in this application.

[0305] This application also provides a computer-readable storage medium that carries one or more computer programs. When the one or more computer programs are executed by an electronic device, the electronic device can implement any of the object association strength calculation methods based on multi-source heterogeneous data provided in this application.

[0306] It should be noted that the object association strength calculation method and related apparatus based on multi-source heterogeneous data provided in this application can be used in the fields of big data, the Internet of Things, or finance. The above are merely examples and do not limit the application areas of the object association strength calculation method and related apparatus based on multi-source heterogeneous data provided in this application.

[0307] It should also be noted that the device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. In addition, in the device embodiment drawings provided in this application, the connection relationship between modules indicates that they have a communication connection, which can be implemented as one or more communication buses or signal lines.

[0308] Through the above description of the embodiments, those skilled in the art can clearly understand that this application can be implemented by means of software plus necessary general-purpose hardware, or it can be implemented by special-purpose hardware including application-specific integrated circuits, special-purpose CPUs, special-purpose memory, special-purpose components, etc. Generally, any function performed by a computer program can be easily implemented by corresponding hardware, and the specific hardware structure used to implement the same function can also be diverse, such as analog circuits, digital circuits, or special-purpose circuits. However, for this application, software program implementation is more often the preferred implementation method. Based on this understanding, the technical solution of this application, in essence, or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product is stored in a readable storage medium, such as a computer floppy disk, USB flash drive, mobile hard disk, ROM, RAM, magnetic disk, or optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, training equipment, or network device, etc.) to execute the methods described in the various embodiments of this application.

[0309] In the above embodiments, implementation can be achieved, in whole or in part, through software, hardware, firmware, or any combination thereof. When implemented in software, it can be implemented, in whole or in part, as a computer program product.

[0310] The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, all or part of the processes or functions described in the embodiments of this application are generated. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, training device, or data center to another website, computer, training device, or data center via wired (e.g., coaxial cable, fiber optic, digital subscriber line (DSL)) or wireless (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium may be any available medium that a computer can store or a data storage device such as a training device or data center that integrates one or more available media. The available media may be magnetic media (e.g., floppy disks, hard disks, magnetic tapes), optical media (e.g., DVDs), or semiconductor media (e.g., solid-state drives (SSDs)).

Claims

1. A method for calculating the object association strength based on multi-source heterogeneous data, characterized in that, include: Acquire multi-source heterogeneous data; The interaction data of the target object with respect to the associated object in multiple dimensions are obtained from the multi-source heterogeneous data; If the interaction data of the target object in the first interaction direction of the target dimension is missing, the interaction data of the first interaction direction is supplemented based on the interaction data of the associated object in the second interaction direction of the target dimension. The first interaction direction and the second interaction direction are opposite interaction directions of the same interaction behavior. The target dimension is any one of the multiple dimensions. For each dimension, based on the interaction data of the target object relative to the associated object in that dimension and the preset interval mapping rules, the relationship value between the target object and the associated object in that dimension is calculated. Based on the relationship value between the target object and the associated object in each dimension, the association strength value between the target object and the associated object is calculated.

2. The method for calculating object association strength based on multi-source heterogeneous data according to claim 1, characterized in that, The multiple dimensions include at least two of the following: fund interaction dimension, equity topology dimension, personnel appointment dimension, bill transaction dimension, project bidding cooperation dimension, and organizational hierarchy dimension. The step of obtaining interaction data of the target object relative to the associated object in multiple dimensions from the multi-source heterogeneous data includes at least two of the following: If the dimension is the fund interaction dimension, obtain the settlement direction of the target object relative to the associated object within a preset period and the total settlement transaction amount corresponding to the settlement direction; If the dimension is the equity topology dimension, obtain the shareholding relationship of the target object relative to the associated object within the preset period and the shareholding ratio corresponding to the shareholding relationship; If the dimension is the personnel employment dimension, obtain the number of related enterprises of the target object and the related object within the preset period, and the wealth level parameters of the personnel with overlapping employment of the target object and the related object; If the dimension is the bill transaction dimension, obtain the seller identifier and the corresponding sales transaction data of the seller identifier relative to the associated object within the preset period, or the buyer identifier and the corresponding input transaction data of the buyer identifier; If the dimension is the project bidding cooperation dimension, obtain the bidding identifier and the preset bidding score corresponding to the bidding identifier of the target object relative to the associated object within the preset period, or the winning bid identifier and the preset winning bid score corresponding to the winning bid identifier; If the dimension is the organizational hierarchy dimension, obtain the target object's superior identifier and the corresponding preset superior hierarchy score within the preset period, or the subordinate identifier and the corresponding preset subordinate hierarchy score.

3. The method for calculating object association strength based on multi-source heterogeneous data according to claim 2, characterized in that, The step of obtaining the target object's superior identifier and its corresponding preset superior level score, or the subordinate identifier and its corresponding preset subordinate level score, within the preset period, relative to the associated object, includes: Based on the subject name of the target object and the subject name of the associated object, determine the hierarchical relationship between the target object and the associated object; If the hierarchical relationship indicates that the target object is a subordinate object, configure a subordinate identifier and a corresponding preset subordinate level score for the target object; If the hierarchical relationship indicates that the target object is a superior object, then a superior identifier and a corresponding preset superior level score are configured for the target object.

4. The method for calculating object association strength based on multi-source heterogeneous data according to claim 3, characterized in that, The step of determining the hierarchical relationship between the target object and the associated object based on the subject name of the target object and the subject name of the associated object includes: If the main name of the associated object is formed by concatenating the main name of the target object with a preset string, the target object is determined to be the parent object; If the main name of the target object is formed by concatenating the main name of the associated object with the preset string, the target object is determined to be a subordinate object.

5. The method for calculating object association strength based on multi-source heterogeneous data according to any one of claims 1 to 4, characterized in that, The step of calculating the relationship value between the target object and the associated object in this dimension based on the interaction data of the target object relative to the associated object and the preset interval mapping rules includes: Determine the target numerical range of the interaction data of the target object relative to the associated object in this dimension; Find the target gradient level value corresponding to the target numerical range from the correspondence between the numerical range and the gradient level value; Based on the target gradient level value and the preset dimension benchmark coefficient, the relationship value of that dimension is calculated.

6. The method for calculating object association strength based on multi-source heterogeneous data according to any one of claims 1 to 4, characterized in that, The step of supplementing the interaction data in the first interaction direction based on the interaction data of the associated object in the second interaction direction of the target dimension includes: If the second interaction direction is the second identifier, then the first interaction direction is set to the first identifier; The interaction data corresponding to the second interaction direction is used as the interaction data corresponding to the first interaction direction.

7. The method for calculating object association strength based on multi-source heterogeneous data according to claim 6, characterized in that, The target dimension is the bill transaction dimension, and the interaction data in the second interaction direction includes the seller identifier and the corresponding sales transaction data within a preset period, or the buyer identifier and the corresponding input transaction data. The step of setting the first interaction direction as the first identifier if the second interaction direction is the second identifier includes: If the second identifier is the seller identifier, then the first identifier is set to the buyer identifier; If the second identifier is the buyer's identifier, then the first identifier is set to the seller's identifier; The step of using the interaction data corresponding to the second interaction direction as the interaction data corresponding to the first interaction direction includes: If the second identifier is the seller identifier, the sales transaction data in the interaction data of the second interaction direction shall be used as the input transaction data in the interaction data of the first interaction direction; If the second identifier is the buyer's identifier, the input transaction data in the interaction data of the second interaction direction will be used as the output transaction data in the interaction data of the first interaction direction.

8. The method for calculating object association strength based on multi-source heterogeneous data according to claim 7, characterized in that, The interactive data in the document transaction dimension also includes the invoice issuance time and the invoice data entry time; If the invoice issuance time is within a preset time window, and the time difference between the current time and the invoice data entry time is less than or equal to a preset storage validity threshold, then the interaction data in the first interaction direction is completed based on the interaction data of the associated object in the second interaction direction of the target dimension. The start time of the preset time window is the authorization time of the target object, and the length of the preset time window is a preset duration.

9. A device for calculating the object association strength based on multi-source heterogeneous data, characterized in that, include: The data acquisition module is used to acquire heterogeneous data from multiple sources; The feature extraction module is used to obtain interaction data of the target object relative to the associated object in multiple dimensions from the multi-source heterogeneous data; The data reverse completion module is used to complete the interaction data in the first interaction direction based on the interaction data of the associated object in the second interaction direction of the target dimension if the interaction data of the target object in the first interaction direction of the target dimension is missing. The first interaction direction and the second interaction direction are opposite interaction directions of the same interaction behavior. The target dimension is any one of the multiple dimensions. The scoring module is used to calculate the relationship value between the target object and the associated object in each dimension, based on the interaction data of the target object relative to the associated object in the dimension and a preset interval mapping rule. The multidimensional fusion module is used to calculate the association strength value between the target object and the associated object based on the relationship value between the target object and the associated object in each dimension.

10. An electronic device, characterized in that, include: At least one processor; as well as, A memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, which, when executed by the at least one processor, enables the at least one processor to perform the object association strength calculation method based on multi-source heterogeneous data as described in any one of claims 1 to 8.