A method and system for multi-dimensional analysis and visualization of agency business ledger data

By constructing a dual-channel architecture and a multi-dimensional mapping space, the problems of data flow relationship loss and data protection separation in the existing multi-dimensional analysis of agency business ledgers are solved, realizing efficient data analysis and secure visualization.

CN123115263APending Publication Date: 2026-11-17HENGGEBAO TECHNOLOGY (HEBEI) CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611317563.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-28
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

Existing multidimensional analysis methods for agency business ledgers lose the flow relationships between data entities during the aggregation process, failing to describe the reasons for the generation of data entities and the relationships between participants in the business process. Furthermore, data protection and visualization are separated, making it difficult to ensure data security while providing a consistent analytical experience.

Method used

A dual-channel architecture consisting of anchor channels and event mapping channels is constructed. The anchor channels store data entities and generate data anchor numbers, while the event mapping channels establish the relationships and prefix numbers between data entities. Combined with a multi-dimensional mapping space and an observation window, a visual structure output is generated, realizing the association model and visualization of data entities.

Benefits of technology

It effectively improves the efficiency of ledger data analysis, ensures data integrity and traceability, and embeds data protection mechanisms in the visualization process to ensure data security and provide a consistent analysis experience.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123115263A_ABST
    Figure CN123115263A_ABST
Patent Text Reader

Abstract

This invention discloses a method and system for multidimensional analysis and visualization of agency business ledger data, belonging to the field of ledger data analysis technology. It involves collecting multidimensional ledger data, performing dimensional segmentation on the multidimensional ledger data to generate a dimension-based ledger dataset; constructing a dual-channel architecture, registering the ledger dataset based on the dual-channel architecture to generate a correlation model of the multidimensional ledger data; constructing a multidimensional mapping space, mapping the correlation model into the multidimensional mapping space, pre-setting multiple observation windows, and generating multiple visualization structure outputs based on the multiple observation windows; combining the visualization structure outputs from different observation directions based on the visualization structure outputs to generate and output a visualization view. This invention, through a dual-channel architecture and a probe factor mechanism, achieves the integration of ledger data protection and visualization while preserving complete correlation relationships.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of ledger data analysis technology, specifically to a method and system for multidimensional analysis and visualization of agency business ledger data. Background Technology

[0002] The existing multidimensional analysis of agency business ledgers mainly relies on OLAP multidimensional datasets to aggregate ledger data according to preset dimensions and realize the query of metric values ​​under different dimension combinations through operations such as slicing, dicing, roll-up, and drill-down. However, existing technologies have certain shortcomings. Existing OLAP methods lose the flow relationships between data entities during the aggregation process. Each metric in the ledger data has multiple attribution relationships across multiple dimensions such as customer, contract, service personnel, region, and business type. Existing methods split the attribution of the same metric across different dimensions into independent aggregation operations, and the relationship information between the dimensions cannot be reflected after aggregation. At the same time, data lineage tracing only records the technical flow between tables and does not carry business semantics, so it cannot describe the reasons for the generation of data entities in the business process and the relationships between participants. Furthermore, the analysis results of different dimensions are displayed in multiple independent charts, which fragments the relationship structure of the same business problem from different perspectives. In addition, data protection and visualization are separated, making it difficult to provide a consistent data analysis process while ensuring data security. Therefore, it is of great significance to develop a method for multidimensional analysis and visualization of agency business ledger data. Summary of the Invention

[0003] The purpose of this invention is to provide a method and system for multidimensional analysis and visualization of agency business ledger data, so as to solve the problems in the background technology.

[0004] To achieve the above objectives, the present invention provides the following technical solution: a method for multi-dimensional analysis and visualization of agency business ledger data, comprising: Collect multidimensional ledger data, perform dimensional segmentation on the multidimensional ledger data, and generate a dimension-based ledger dataset; A dual-channel architecture is constructed, and the ledger dataset is registered based on the dual-channel architecture to generate a multi-dimensional ledger data association model; Construct a multidimensional mapping space, map the associated model to the multidimensional mapping space, preset multiple observation windows, and generate multiple visualization structure outputs based on the multiple observation windows; Based on the visualization structure output, the visualization structure outputs from different observation directions are combined to generate and output a visualization view.

[0005] In a preferred embodiment, the step of dimensionally segmenting the multidimensional ledger data to generate a dimension-based ledger dataset is as follows: The feature extraction algorithm is used to identify the dimension labels of multidimensional ledger data. The dimension labels include time labels, region labels, business labels, customer type labels, and service personnel labels. Multidimensional ledger data is segmented based on dimension labels, and multidimensional ledger data with the same dimension label are grouped into a subset of dimension data. Each dimensional data subset is labeled with a dimensional identifier, and the dimensional identifier corresponds one-to-one with the dimensional label of the dimensional data subset; Integrate all dimensional data subsets to generate a ledger dataset.

[0006] In a preferred embodiment, the step of constructing a dual-channel architecture, registering the ledger dataset based on the dual-channel architecture, and generating a multi-dimensional ledger data association model is as follows: Construct a dual-channel architecture, which includes an anchor channel and an event mapping channel; Among them, the anchor channel is used to store the data entities in the ledger dataset and assign a data anchor number to each data entity. The data entities in the anchor channel cannot be modified after being written. The event mapping channel is used to generate the relationships between data entities and the prefix number of new data entities based on preset key business events; Each data record in the ledger dataset is treated as a data entity and registered in the anchor channel. During registration, the anchor channel assigns a data anchor number to the data entity and establishes a storage unit. The storage unit carries the data anchor number, entity type, original metric value, and dimension identifier. Configure event mapping points in the event mapping channel. Each event mapping point corresponds to a key business event type. Event mapping points are connected through event channels. The direction of the event channels is determined by the flow order of key business events in the agent business ledger. Numbering rules are pre-stored in the event mapping points. When a critical business event is triggered, the corresponding event mapping point generates a prefix number for the new data entity according to the numbering rules, and assigns a data anchor number to the corresponding new data entity based on the prefix number. When generating a prefix number, the event mapping point generates a probe factor, and weak connection segments between data entities are constructed based on the probe factor; Record the data anchor number of the data entity participating in the key business event as the source entity identifier, record the data anchor number of the new data entity as the target entity identifier, record the event type code as the association type, and generate association edges; A multidimensional ledger data association model is generated by combining all associated edges and all data entities based on associated edges and weak connection segments.

[0007] In a preferred embodiment, the step of assigning data anchor point numbers to the corresponding new data entities based on the preceding number is as follows: The event mapping point obtains the data anchor number of all data entities participating in the current key business event, forming a set of data anchor numbers; According to the numbering segmentation rules, extract number fragments from the specified positions of each data anchor number in the data anchor number set; The number segmentation rules include the definition of the starting position of the truncation, the definition of the truncation length, the definition of the event type encoding, and the definition of the combination order. All extracted number fragments are combined with the event type codes of the current key business events and concatenated based on the combination order definition to serve as the pre-encoding of the new data entity; After the event mapping point generates a prefix number, it sends a new data entity registration request to the anchor channel. The registration request includes the prefix number, the entity type of the new data entity, the original metric value, and the dimension identifier. Based on the registration request, the anchor channel assigns a unique data anchor number to the new data entity and establishes a corresponding storage unit. The storage unit carries the preceding number, data anchor number, entity type, original metric value, and dimension identifier.

[0008] In a preferred embodiment, the step of constructing weak links between data entities based on the probe factor is as follows: When generating a prefix number, the feature segment of the prefix number is extracted, the feature segment is encapsulated into multiple independent detection factors, and the detection factors are released into the event mapping channel. When the detection factor detects the intersection event channel, the prefix number of the new data entity in the event channel is detected based on the detection factor. The feature segment consists of the event type code carried in the prefix number and the data anchor number fragment of at least one data entity participating in the key business event. The feature segment is used to characterize the source entity characteristics and event type characteristics of the prefix number. If a new data entity with a number similarity higher than a preset threshold is detected, the new data entity submits a weak connection request to the data entity or new data entity that sent the probe factor and generates a weak connection segment.

[0009] In a preferred embodiment, the step of generating multiple visualization structure outputs based on multiple observation windows is as follows: A multidimensional mapping space is constructed. The outer wall of the multidimensional mapping space is composed of multiple dimensional walls. Each dimensional wall corresponds to a dimension identifier of the ledger dataset. Access areas and isolation areas are deployed inside the multidimensional mapping space. Randomly generated chaotic data is stored in the access areas. Among them, the dimension wall includes the time dimension wall, the regional dimension wall, the business type dimension wall, the customer type dimension wall, and the service personnel dimension wall; For each data entity in the association model, spatial coordinates are assigned to an isolated region in the multidimensional mapping space according to the dimension identifier carried by the data entity and the association relationship of the data entity. For a data entity that carries a dimension identifier, the coordinate components of the data entity in that dimension are determined by the dimension identifier. For a dimension in which a data entity does not carry a dimension identifier, the coordinate components of the data entity in that dimension are determined by the position of the dimension identifiers of other data entities that have associated edges with that data entity; The associated edges and weak connection segments in the association model are mapped to the connection lines between data entities in the multidimensional mapping space. The associated edges are mapped to solid connection lines, and the weak connection segments are mapped to dashed connection lines. Multiple observation windows are set on each dimensional wall. The observation windows are located on the surface of the dimensional wall, and the observation direction of the observation windows is from the dimensional wall where the observation window is located to the interior of the multidimensional mapping space. Only one observation window on each dimension wall is the correct observation window. The correct observation window stores the access address path of the isolated region, while the other observation windows store the access address path of the accessed region. For a specific dimensional wall, the correct observation window generates the observation direction. Based on the observation direction, the association model of the isolated area is observed, and the association relationship between data entities in the corresponding dimensional wall is generated as a single-dimensional visualization structure output. At the same time, the correlation model under this dimension is projected onto the dimension wall in the opposite direction of observation to generate the correlation relationship under the dimension of the opposing dimension wall, which serves as the output of a multi-dimensional visualization structure. The single-dimensional visualization structure output and the multi-dimensional visualization structure output are integrated into a single visualization structure output.

[0010] In a preferred embodiment, the step of combining the visualization structure outputs from different observation directions to generate and output a visualization view based on the visualization structure output is as follows: Obtain the visualization structure output generated from the observation windows on the multi-dimensional wall. Each visualization structure output includes data entities and the relationships between data entities. Each data entity in the output of the visualization structure is converted into a visualization node, and the display position of the visualization node is determined by the spatial coordinate projection of the data entity in the multidimensional mapping space. Solid lines are converted into first visual lines, and dashed lines are converted into second visual lines. The first visual lines represent associated edges, and the second visual lines represent weak connection segments. All visualization nodes, the first visualization connection, and the second visualization connection are arranged according to the dimensional wall where the corresponding observation window is located, generating a multi-view visualization view. Each display area of ​​the multi-view visualization view corresponds to the visualization structure output of the observation window of a dimensional wall.

[0011] This invention also provides a multi-dimensional analysis and visualization system for agency business ledger data, including: Data partitioning module: Collects multidimensional ledger data, performs dimensional partitioning on the multidimensional ledger data, and generates a dimension-based ledger dataset; Model building module: Connected to the data partitioning module, it constructs a dual-channel architecture, performs registration processing on the ledger dataset based on the dual-channel architecture, and generates a multi-dimensional ledger data association model; Mapping Space Module: Connects to the model building module to construct a multi-dimensional mapping space, mapping related models to the multi-dimensional mapping space, pre-setting multiple observation windows, and generating multiple visualization structure outputs based on the multiple observation windows; Visualization module: Connected to the mapping space module, based on the visualization structure output, it combines the visualization structure outputs from different observation directions to generate and output a visualization view.

[0012] The technical effects and advantages provided by the present invention in the above technical solution are as follows: 1. This invention constructs a dual-channel architecture comprising an anchor channel and an event mapping channel, completely separating the original storage of ledger data from business flow processing at the physical level. The anchor channel, acting as a read-only storage area, ensures the integrity and traceability of the original ledger data. The event mapping channel models business events through event mapping points and event channels, and constructs pre-encoding, ensuring that each newly generated data entity naturally carries the genetic information of its source entity. Simultaneously, the detection factor mechanism, by extracting feature segments from the pre-encoding and autonomously diffusing along the event channel, can automatically discover implicit associations between data entities belonging to different flow paths and generate weak connection segments. Based on this mechanism, the data flow process itself becomes the process of forming an association model, effectively improving the analysis efficiency of ledger data. 2. This invention embeds a data protection mechanism into the generation process of the visualization structure by mapping a multidimensional space and deploying isolated and access areas within that space. The isolated area stores real relational model data, while the access area stores randomly generated chaotic data. In each dimension wall, only one of the multiple observation windows stores the access address path of the isolated area, while the others store the access address path of the access area. When an authorized user observes through the correct observation window, the observation direction is perpendicular to the dimension wall, obtaining the real single-dimensional relational relationships in the isolated area and the multi-dimensional relational relationships projected onto the opposite dimension wall. When an unauthorized user observes through other observation windows, they obtain chaotic data with no business meaning in the access area under the same visualization interface. Based on this mechanism, the data security of the ledger data is effectively guaranteed. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0014] Figure 1 This is a flowchart of the method of the present invention.

[0015] Figure 2 This is a system block diagram of the present invention. Detailed Implementation

[0016] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0017] Example 1, please refer to Figure 1 As shown in this embodiment, a method for multidimensional analysis and visualization of agency business ledger data includes: S1. Collect multidimensional ledger data, perform dimensional segmentation on the multidimensional ledger data, and generate a dimension-based ledger dataset; S2. Construct a dual-channel architecture, and register the ledger dataset based on the dual-channel architecture to generate a multi-dimensional ledger data association model. S3. Construct a multidimensional mapping space, map the associated model to the multidimensional mapping space, preset multiple observation windows, and generate multiple visualization structure outputs based on the multiple observation windows; S4. Based on the visualization structure output, combine the visualization structure outputs from different observation directions to generate and output a visualization view; As described in steps S1-S4 above, the existing multidimensional analysis of agency business ledgers mainly relies on OLAP multidimensional datasets to aggregate ledger data according to preset dimensions and realize the query of metric values ​​under different dimension combinations through operations such as slicing, dicing, roll-up, and drill-down. However, existing technologies have certain shortcomings. Current OLAP methods lose the flow relationships between data entities during the aggregation process. Each metric in the ledger data simultaneously has attribution relationships across multiple dimensions such as customer, contract, service personnel, region, and business type. Existing methods split the attribution of the same metric across different dimensions into independent aggregation operations, resulting in the loss of the relationship information between dimensions after aggregation. Furthermore, data lineage tracing only records the technical flow between tables, lacking business semantics and failing to describe the reasons for the generation of data entities and the relationships between participants in the business process. Moreover, the analysis results from different dimensions are displayed through multiple independent charts, fragmenting the relational structure of the same business problem from different perspectives. In addition, the separation of data protection and visualization makes it difficult to provide a consistent analytical experience while ensuring data security. This invention constructs a dual-channel architecture comprising an anchor channel and an event mapping channel, completely separating the original storage of ledger data from business flow processing at the physical level. The anchor channel, acting as a read-only storage area, ensures the integrity and traceability of the original ledger data. The event mapping channel models business events through event mapping points and event channels, and constructs pre-encoding, ensuring that each newly generated data entity naturally carries the genetic information of its source entity. Simultaneously, the detection factor mechanism, by extracting feature segments from the pre-encoding and autonomously diffusing along the event channel, can automatically discover implicit associations between data entities belonging to different flow paths and generate weak connection segments. Based on this mechanism, the data flow process itself becomes the process of forming an association model, effectively improving the analysis efficiency of ledger data. By mapping a multidimensional space and deploying isolated and access areas within that space, a data protection mechanism is embedded in the generation process of the visualization structure. The isolated areas store real relational model data, while the access areas store randomly generated, chaotic data. Among the multiple observation windows set on each dimension wall, only one correct observation window stores the access address path of the isolated area, while the others store the access address paths of the access areas. When an authorized user observes through the correct observation window, the observation direction is perpendicular to the dimension wall, obtaining the real single-dimensional relational relationships in the isolated area and the multi-dimensional relational relationships projected onto the opposite dimension wall. Conversely, when unauthorized users observe through other observation windows, they obtain chaotic data with no business meaning from the access areas under the same identical visualization interface. Based on this mechanism, the data security of the ledger data is effectively guaranteed.

[0018] In one embodiment, step S1, which involves dimensionally segmenting the multidimensional ledger data to generate a dimension-based ledger dataset, includes: S11. Identify the dimension labels of multidimensional ledger data based on feature extraction algorithms. The dimension labels include time labels, region labels, business labels, customer type labels, and service personnel labels. S12. Segment the multidimensional ledger data based on dimension labels, and group the multidimensional ledger data with the same dimension label into a single dimension data subset. S13. Label each dimensional data subset with a dimensional identifier, and the dimensional identifier corresponds one-to-one with the dimensional label of the dimensional data subset; S14. Integrate all dimensional data subsets to generate a ledger dataset; As described in steps S11-S14 above, a feature extraction algorithm combining rule matching based on field names and pattern recognition based on data content is used to scan each data record of the multidimensional ledger data field by field. When a field name successfully matches a time-type field, region-type field, business-type field, customer-type field, or service personnel-type field in the preset dimension field dictionary, the value of that field is extracted as the corresponding dimension label. Dimension labels include time labels, region labels, business labels, customer-type labels, and service personnel labels. At the same time, for field content where the field name does not directly match but the field value conforms to the preset regular expression pattern, the implicit dimension labels are extracted through pattern recognition. After completing the dimension label recognition, all data records are traversed using the dimension label as the delimiter. Data records carrying the same dimension label are grouped into the same dimension data subset. One dimension data subset corresponds to one dimension label. Because the same data record carries multiple dimensions... A data set can be categorized into multiple dimensional data subsets simultaneously. Each dimensional data subset is labeled with a dimensional identifier. The generation rule for the dimensional identifier is to concatenate the category prefix of the dimensional label with the specific value of the dimensional label. The category prefixes include TIME, REGION, BUSINESS, CUSTTYPE, and SERVER, which correspond to time labels, region labels, business labels, customer type labels, and service personnel labels, respectively. The specific value is a fixed-length string after hash encoding. The concatenated dimensional identifier serves as the globally unique identifier for that dimensional data subset. All dimensional data subsets are integrated, and a key-value mapping structure is constructed with the dimensional identifier as the primary key and the dimensional data subset as the value. At the same time, an inverted index is built for each dimensional data subset. The inverted index records the correspondence between each data entity identifier and the list of dimensional identifiers of the dimensional data subset to which the data entity belongs. The key-value mapping structure and the inverted index together constitute the ledger dataset.

[0019] In one embodiment, step S2, which involves constructing a dual-channel architecture, performing registration processing on the ledger dataset based on the dual-channel architecture, and generating a multi-dimensional ledger data association model, includes: S21. Construct a dual-channel architecture, which includes an anchor channel and an event mapping channel; S22. Among them, the anchor channel is used to store the data entities in the ledger dataset and assign a data anchor number to each data entity. The data entities in the anchor channel cannot be modified after being written. S23. The event mapping channel is used to generate the association between data entities and the prefix number of new data entities based on preset key business events. S24. Treat each data record in the ledger dataset as a data entity and register it in the anchor channel. During registration, the anchor channel assigns a data anchor number to the data entity and establishes a storage unit. The storage unit carries the data anchor number, entity type, original metric value and dimension identifier. S25. Configure event mapping points in the event mapping channel. Each event mapping point corresponds to a key business event type. Event mapping points are connected through event channels. The direction of the event channels is determined by the flow order of key business events in the agent business ledger. S26. Pre-store number segmentation rules in the event mapping point. When a key business event is triggered, the corresponding event mapping point generates a prefix number of the new data entity according to the number segmentation rules, and assigns a data anchor number to the corresponding new data entity based on the prefix number. S27. When generating the preceding number, the event mapping point generates a probe factor, and a weak connection segment between data entities is constructed based on the probe factor. S28. Record the data anchor number of the data entity participating in the key business event as the source entity identifier, record the data anchor number of the new data entity as the target entity identifier, record the event type code as the association type, and generate the association edge; S29. Combine all associated edges and all data entities based on associated edges and weak connection segments to generate a multidimensional ledger data association model. As described in steps S21-S29 above, a dual-channel architecture is constructed, comprising an anchor channel and an event mapping channel. The anchor channel is implemented based on an immutable key-value storage engine, using data anchor numbers as keys and storage units as values. Once data is written, modification and deletion are prohibited. The event mapping channel consists of event mapping points and event channels. Event mapping points are configurable processing nodes, and event channels are message queues connecting event mapping points. The anchor channel stores data entities in the ledger dataset and assigns a data anchor number to each data entity. The data anchor number is composed of the entity type code, registration timestamp, and incrementing sequence number. The event mapping channel is used to generate associations between data entities based on preset key business events. The new data entities are assigned prefix numbers, and key business events include, but are not limited to, contract signing events, service execution events, payment receipt events, contract modification events, and contract termination events. Each data record in the ledger dataset is treated as a data entity and registered in the anchor channel. During registration, the anchor channel assigns a data anchor number to the data entity and establishes a storage unit. The storage unit carries the data anchor number, entity type, original metric, and dimension identifier. The dimension identifier is directly taken from the dimension identifier in the ledger dataset. Event mapping points in the event mapping channel are deployed as independent processing units. Each event mapping point contains three components: a rule loader, an interceptor executor, and a registration request builder. The rule loader is responsible for... The configuration center loads the number segmentation rule corresponding to the event mapping point. The truncation executor is responsible for performing number segmentation operations on the input data anchor number according to the number segmentation rule. The registration request builder is responsible for concatenating the truncation result with the event type code to form a prefix number and constructing a new data entity registration request. The event channel is implemented in the form of a message queue topic. The upstream event mapping point acts as a message producer and sends the mapping packet to the topic. The downstream event mapping point acts as a message consumer and subscribes to the mapping packet from the topic. The directionality of the event channel is guaranteed by the subscription relationship of the topic. The number segmentation rule is pre-existing in the rule loader of the event mapping point in the form of a structured configuration object. The truncation start position is defined as a non-negative integer value representing the first character of the data anchor number. The starting offset for counting, the truncation length is defined as a positive integer value representing the number of consecutive characters truncated from the starting position, the event type code is defined as a fixed-length unique code for this critical business event type, and the combination order is defined as an ordered list specifying the order in which the truncated number fragments and the event type code are arranged during concatenation; when a critical business event is triggered, the corresponding event mapping point obtains the data anchor point numbers of all data entities participating in the critical business event, forming a set of data anchor point numbers, and truncates a number fragment of a specified length from the specified starting position of each data anchor point number according to the numbering segmentation rules, and concatenates all the truncated number fragments with the event type code according to the combination order, and the concatenation result is used as the prefix number of the new data entity;After generating a prefix number, the event mapping point sends a new data entity registration request to the anchor channel. The registration request carries the prefix number, the entity type of the new data entity, the original metric value, and the dimension identifier. The anchor channel assigns a data anchor number to the new data entity and establishes a corresponding storage unit. While generating the prefix number, the event mapping point generates a probe factor and constructs weak connection segments between data entities based on the probe factor. After completing the prefix number generation and weak connection segment construction, the event mapping point records the data anchor number of the data entity participating in the key business event as the source entity identifier, records the data anchor number of the new data entity as the target entity identifier, records the event type code as the association type, and generates an association edge. When all key business events are processed, all association edges and all data entities are combined based on the association edges and weak connection segments, using data entities as nodes and association edges and weak connection segments as edges, to form a graph-structured multidimensional ledger data association model.

[0020] In one embodiment, step S26, which assigns a data anchor number to the corresponding new data entity based on the preceding number, includes: S261. Obtain the data anchor point numbers of all data entities participating in the current key business event from the event mapping point, and form a set of data anchor point numbers; S262. According to the numbering segmentation rules, extract number fragments from the specified positions of each data anchor number in the data anchor number set; S263, where the numbering segmentation rules include the definition of the truncation start position, the definition of the truncation length, the definition of the event type encoding, and the definition of the combination order; S264. Combine all the extracted number fragments with the event type codes of the current key business events, and concatenate them based on the combination order definition to serve as the pre-encoding of the new data entity. S265. After the event mapping point generates the preceding number, a new data entity registration request is sent to the anchor channel. The registration request includes the preceding number, the entity type of the new data entity, the original metric value, and the dimension identifier. S266. Anchor channel assigns a unique data anchor number to a new data entity based on the registration request and establishes a corresponding storage unit. The storage unit carries the preceding number, data anchor number, entity type, original metric value and dimension identifier. As described in steps S261-S266 above, when a critical business event is triggered, the corresponding event mapping point obtains the data anchor point numbers of all data entities participating in the critical business event, forming a data anchor point number set. The event mapping point reads the number segmentation rules corresponding to the critical business event type from the pre-stored configuration. The number segmentation rules include the definition of the truncation start position, the truncation length, the event type code, and the combination order. The truncation start position is defined as an offset counting from zero, indicating the position of the character after the first character of the data anchor point number from which to start truncation. The truncation length is defined as a positive integer, indicating the number of consecutive characters to be truncated from the truncation start position. The event type code is defined as a fixed-length unique code for the critical business event type. The combination order is defined as an ordered sequence, where each element in the sequence specifies the arrangement position of a number fragment or event type code during concatenation. The event mapping point traverses the data anchor point number set, and for each data anchor point number in the set, according to the truncation start position... The definition and truncation length definition extract continuous character substrings from the data anchor number string, which serve as the number fragment corresponding to that data anchor number. The event mapping point concatenates all extracted number fragments and event type codes in the order specified in the combination order definition, and the concatenation result serves as the prefix number of the new data entity. After generating the prefix number, the event mapping point sends a new data entity registration request to the anchor channel. This registration request includes the prefix number, the entity type of the new data entity, the original metric value, and the dimension identifier. The entity type field indicates the entity type of the new data entity, the original metric value field carries the business metric value corresponding to the new data entity, and the dimension identifier field carries the dimension affiliation information of the new data entity. After receiving the registration request, the anchor channel assigns a unique data anchor number to the new data entity. The assignment method is to concatenate the entity type code of the new data entity, the current timestamp millisecond value of the anchor channel, and the auto-incrementing sequence number maintained internally by the anchor channel using a hyphen. Anchor channels establish storage units for new data entities. Each storage unit carries the data anchor number, preceding number, entity type, original metric value, and dimension identifier. Once a storage unit is written into an anchor channel, it is marked as unmodifiable, and any subsequent modification requests to that storage unit will be rejected by the anchor channel.

[0021] In one embodiment, step S27, which involves constructing weak links between data entities based on a probe factor, includes: S271. When generating the prefix number, extract the feature segment of the prefix number, encapsulate the feature segment into multiple independent detection factors, release the detection factors into the event mapping channel, and when the detection factor detects the intersection event channel, detect the prefix number of the new data entity in the event channel based on the detection factor. S272, wherein the feature segment consists of the event type code carried in the preceding number and the data anchor number fragment of at least one data entity participating in the key business event, and the feature segment is used to characterize the source entity characteristics and event type characteristics of the preceding number; S273. If a new data entity with a number similarity higher than a preset threshold is detected, the new data entity submits a weak connection request to the data entity or new data entity that sent the probe factor and generates a weak connection segment. As described in steps S271-S273 above, when generating a new data entity's prefix number, the event mapping point synchronously extracts a feature segment from the prefix number. The extraction of the feature segment is based on the position of the event type code defined in the numbering segmentation rules and the occupancy order of the number fragments of the data entities participating in the critical business event in the prefix number. The feature segment consists of the event type code carried in the prefix number and at least one data anchor number fragment of the data entity participating in the critical business event. The event mapping point splits the feature segment into multiple independent fragment units, each fragment unit corresponding to a data anchor number fragment or event type code in the feature segment. Each fragment unit is encapsulated as a probe factor. Each probe factor carries the content string of the fragment unit, the identifier of the event mapping point that generated the probe factor, and the maximum propagation hop count. The initial value of the maximum propagation hop count is preset according to the size of the event mapping channel. The event mapping point releases all probe factors one by one into the current event channel. The probe factors propagate downstream along the direction of the event channel. When they reach the intersection of the event channel and other event channels, the routing component at the intersection point copies and forwards the probe factors to all other connected event channels. When a probe factor propagates in an event channel, if there is a mapping point in that event channel... As the packet flows through, the probe factor intercepts the mapping packet and reads the prefix number of the new data entity carried in the packet. The probe factor extracts a comparison fragment of the same length as its own fragment unit from the same offset position of this prefix number. It then performs character-by-character matching between its own fragment unit string and the comparison fragment string, dividing the number of matched characters by the total number of characters in the fragment unit to obtain the number similarity. If the number similarity is higher than a preset threshold, the probe factor triggers a weak connection establishment process. The new data entity corresponding to the mapping packet, acting as the weak connection request initiator, submits a weak connection request to the set of source entities associated with the event mapping point that sent the probe factor. The data entity includes existing data entities participating in the key business event corresponding to the probe factor, as well as new data entities generated by the key business event. The weak connection request is transmitted to the anchor channel in the form of a message. The anchor channel records a weak connection segment in the association model. The source entity identifier of the weak connection segment is the data anchor number of the selected entity in the source entity set associated with the probe factor. The target entity identifier is the data anchor number of the new data entity that initiated the weak connection request. The association type is marked as the weak connection identifier. If the number similarity does not exceed the preset threshold, the probe factor continues to spread along the event channel until the maximum spread hop count is exhausted and it automatically dies without generating a weak connection segment.

[0022] In one embodiment, step S3, which generates multiple visualization structure outputs based on multiple observation windows, includes: S31. Construct a multidimensional mapping space. The outer wall of the multidimensional mapping space is composed of multiple dimensional walls. Each dimensional wall corresponds to a dimension identifier of the ledger dataset. Inside the multidimensional mapping space, access areas and isolation areas are deployed. Randomly generated chaotic data is stored in the access areas. S32. Among them, the dimension wall includes the time dimension wall, the regional dimension wall, the business type dimension wall, the customer type dimension wall, and the service personnel dimension wall. S33. Assign spatial coordinates to each data entity in the association model in the isolated region of the multidimensional mapping space according to the dimension identifier carried by the data entity and the association relationship of the data entity. S34. For a dimension in which a data entity carries a dimension identifier, the coordinate components of the data entity in that dimension are determined by the dimension identifier. S35. For a dimension in which a data entity does not carry a dimension identifier, the coordinate components of the data entity in that dimension are determined by the position of the dimension identifiers of other data entities that have an associated edge with the data entity. S36. Map the associated edges and weak connection segments in the association model to the connection lines between data entities in the multidimensional mapping space. Map the associated edges to solid connection lines and the weak connection segments to dashed connection lines. S37. Set multiple observation windows on each dimensional wall. The observation windows are located on the surface of the dimensional wall and the observation direction of the observation windows is from the dimensional wall where the observation window is located to the interior of the multidimensional mapping space. S38. On each dimensional wall, only one observation window is the correct observation window. The correct observation window stores the access address path of the isolated region, while the other observation windows store the access address path of the accessed region. S39. For a specific dimension wall, generate the observation direction through the correct observation window, observe the association model of the isolated area based on the observation direction, generate the association relationship between data entities in the dimension to which the corresponding dimension wall belongs, and output it as a single-dimensional visualization structure. S310. At the same time, the correlation model under this dimension is projected onto the dimension wall opposite to the observation direction to generate the correlation relationship under the dimension of the opposing dimension wall, as the output of the multi-dimensional visualization structure. S311. Integrate single-dimensional visualization structure output and multi-dimensional visualization structure output into a single visualization structure output; As described in steps S31-S311 above, when constructing the multidimensional mapping space, the outer wall of the multidimensional mapping space is enclosed by the time dimension wall, the region dimension wall, the business type dimension wall, the customer type dimension wall, and the service personnel dimension wall. The interior of the multidimensional mapping space is divided into an access area and an isolation area. The access area is pre-set with randomly generated disordered data, and the isolation area is used to store the real data entities and association structures mapped from the association model. When assigning spatial coordinates to each data entity in the association model in the isolation area of ​​the multidimensional mapping space, for the dimension carrying the dimension identifier, the coordinate components of the data entity in that dimension are determined by a fixed offset calculated by the dimension identifier through a hash mapping function, so that the coordinate components of the data entity carrying the same dimension identifier are determined by the fixed offset calculated by the dimension identifier through a hash mapping function. Data entities with dimension identifiers cluster along the direction of the dimension wall. For dimensions where data entities do not carry dimension identifiers, the coordinate components of the data entity in that dimension are derived by averaging the corresponding dimension coordinate components of other data entities with associated edges. The derivation process is as follows: traverse all directly associated edges of the data entity, find the coordinate components of the entity at the other end of the associated edge in that dimension, sum all valid coordinate components and divide by the number of valid associated entities, and use the average value as the coordinate component of the data entity in the missing dimension. In the association model, associated edges are mapped as solid connecting lines, and weak connecting segments are mapped as dashed connecting lines. Connecting lines are stored in the isolation area as line segments and are located synchronously with the determination of the coordinates of the associated entities. An observation window is set on each dimensional wall, located on its surface. The observation direction of the window points vertically inward from the dimensional wall to the interior of the multidimensional mapping space. Of the multiple observation windows on each dimensional wall, only one is marked as the correct observation window. The configuration of the correct observation window stores the access address path of the isolated region, which points to the spatial coordinate index table within the isolated region. The configurations of the remaining observation windows store the access address paths of the access regions, which point to the disordered data index table within the access regions. When generating the observation direction for the correct observation window of a given dimensional wall, the system follows the observation direction of that correct observation window. The system samples data entities and their connecting lines from the isolated area in the observation direction. The sampling depth is determined by the spatial range traversed by the observation direction. The relationships between data entities belonging to the dimension identifier of the dimension wall where the correct observation window is located are extracted as a single-dimensional visualization structure output. At the same time, the system projects the sampled data entity set back along the observation direction onto the opposite dimension wall. The system generates the data entity relationships under the dimension identifier of the opposite dimension wall as a multi-dimensional visualization structure output. The steps are as follows: determine the starting dimension wall where the current correct observation window is located and the opposite dimension wall that the observation direction passes through along the straight line. The opposite dimension wall is the projection target dimension wall.Using the observation point of the correct observation window on the starting dimension wall as the origin and the observation direction vector as the projection reference axis, a perspective projection transformation is performed on the spatial coordinates of each sampled data entity in the isolated area. The formula for the perspective projection transformation is that the projection coordinates equal the spatial coordinates of the data entity minus the component of the data entity's spatial coordinates in the direction of the projection reference axis. The resulting projection coordinates are the projection positions of the data entities on the opposing dimension wall. The solid and dashed connecting lines between the sampled data entities are also projected along with the data entities. The projection of the connecting lines maintains their topological connection relationship, and only the projection coordinates of the data entities at both ends of the connecting lines are used as the new endpoint coordinates. After projection, a set of projected data entities and projected connecting lines are obtained on the opposing dimension wall. This set of projected data entities and projected connecting lines represents the association relationship between data entities under the corresponding dimension identifier of the opposing dimension wall. For multiple data entities projected to the same projection position, an offset arrangement strategy is adopted, and overlapping data entities are sequentially slightly shifted along the normal direction of the opposing dimension wall. The shift interval is a preset fixed value to ensure that the data entities on the projected view do not overlap. The relationships under the dimensions of the projected opposing dimension wall are used as components of the multi-dimensional visualization structure output, and merged with the single-dimensional visualization structure output of the starting dimension wall to form a visualization structure output that includes the relationships between the starting dimension and the opposing dimension. The single-dimensional visualization structure output is merged and packaged with the multi-dimensional visualization structure output to form the visualization structure output corresponding to the correct observation window. For incorrect observation windows, the system uses the access address path of its stored access area to read random and disordered data from the access area and generates a visualization structure output with the same appearance format but no business meaning in the content. The visualization structure outputs generated by all observation windows on all dimension walls constitute multiple visualization structure outputs.

[0023] In one embodiment, step S4, which combines the visualization structure outputs from different observation directions to generate and output a visualization view based on the visualization structure output, includes: S41. Obtain the visualization structure output generated by the observation windows on the multi-dimensional wall. Each visualization structure output includes data entities and the relationships between data entities. S42. Convert the data entities in each visualization structure output into visualization nodes. The display position of the visualization nodes is determined by the spatial coordinate projection of the data entities in the multidimensional mapping space. S43. Convert solid line connections into first visual connections and dashed line connections into second visual connections. The first visual connection represents associated edges, and the second visual connection represents weak connection segments. S44. Arrange all visualization nodes, the first visualization connection and the second visualization connection according to the dimension wall where the corresponding observation window is located to generate a multi-view visualization view. Each display area of ​​the multi-view visualization view corresponds to the visualization structure output of the observation window of a dimension wall. As described in steps S41-S44 above, all visualization structure outputs generated by the observation windows on multiple dimension walls are collected. Each visualization structure output includes a set of data entities, a set of solid line connections between data entities, and a set of dashed line connections. When converting the data entities in each visualization structure output into visualization nodes, the system reads the entity type field of the data entity and assigns the display shape of the visualization node according to the entity type: customer entities correspond to circular nodes, contract entities correspond to square nodes, service personnel entities correspond to triangular nodes, and payment record entities correspond to diamond nodes. The display position of the visualization node is determined by the two-dimensional screen coordinates obtained by calculating the spatial coordinates of the data entity in the multidimensional mapping space using a perspective projection algorithm. The perspective projection algorithm uses the observation position of the observation window as the viewpoint, the observation direction as the line of sight, and the surface of the dimension wall as the projection plane to calculate the projection abscissa and ordinate of the data entity on the projection plane. The display size of the visualization node is determined by the node radius obtained by transforming the original metric value carried by the data entity through a linear mapping function. When converting the solid line connection into the first visualization connection, the projection position coordinates of the visualization nodes of the data entities corresponding to the two endpoints of the solid line connection are extracted, and the nodes are drawn using these two projection position coordinates as the start and end points. For continuous line segments, the line width of the first visual connection is determined by the association weight of the associated edges through proportional mapping, and the color of the first visual connection is a fixed dark color. When converting the dashed connection to the second visual connection, the system calculates the projected coordinates of the start and end points in the same way, and draws the line segment using alternating dashed segments of equal length and equally spaced blanks. The line width of the second visual connection is a fixed value, and the color of the second visual connection is a lighter color different from that of the first visual connection. When all visual nodes, the first visual connection, and the second visual connection are arranged according to the dimensional wall where the corresponding observation window is located, the system divides the display area of ​​the multi-view visualization view. The system is divided into multiple rectangular sub-regions, each corresponding to an observation window on a dimension wall. The rectangular sub-regions are arranged in the order of the dimension wall types, starting with the time dimension wall, followed by the region dimension wall, business type dimension wall, customer type dimension wall, and service personnel dimension wall. Within each rectangular sub-region, the system renders the visualization node, the first visualization connection, and the second visualization connection corresponding to the observation window. The system labels the corresponding dimension wall name and observation window number for each rectangular sub-region at the top of the multi-view visualization view. The rendered multi-view visualization view is output to the display terminal in a scalable vector graphics format through the front-end visualization component.

[0024] Example 2, please refer to Figure 2As shown in this embodiment, a multi-dimensional analysis and visualization system for agency business ledger data includes: Data partitioning module: Collects multidimensional ledger data, performs dimensional partitioning on the multidimensional ledger data, and generates a dimension-based ledger dataset; Model building module: Connected to the data partitioning module, it constructs a dual-channel architecture, performs registration processing on the ledger dataset based on the dual-channel architecture, and generates a multi-dimensional ledger data association model; Mapping Space Module: Connects to the model building module to construct a multi-dimensional mapping space, mapping related models to the multi-dimensional mapping space, pre-setting multiple observation windows, and generating multiple visualization structure outputs based on the multiple observation windows; Visualization module: Connected to the mapping space module, based on the visualization structure output, it combines the visualization structure outputs from different observation directions to generate and output a visualization view.

[0025] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A method for multidimensional analysis and visualization of agency business ledger data, characterized in that, include: Collect multidimensional ledger data, perform dimensional segmentation on the multidimensional ledger data, and generate a dimension-based ledger dataset; A dual-channel architecture is constructed, and the ledger dataset is registered based on the dual-channel architecture to generate a multi-dimensional ledger data association model; Construct a multidimensional mapping space, map the associated model to the multidimensional mapping space, preset multiple observation windows, and generate multiple visualization structure outputs based on the multiple observation windows; Based on the visualization structure output, the visualization structure outputs from different observation directions are combined to generate and output a visualization view.

2. The method of claim 1, wherein, The steps for performing dimensional segmentation on the multidimensional ledger data to generate a dimension-based ledger dataset are as follows: The feature extraction algorithm is used to identify the dimension labels of multidimensional ledger data. The dimension labels include time labels, region labels, business labels, customer type labels, and service personnel labels. Multidimensional ledger data is segmented based on dimension labels, and multidimensional ledger data with the same dimension label are grouped into a subset of dimension data. Each dimensional data subset is labeled with a dimensional identifier, and the dimensional identifier corresponds one-to-one with the dimensional label of the dimensional data subset; Integrate all dimensional data subsets to generate a ledger dataset.

3. The method of claim 1, wherein, The steps for constructing a dual-channel architecture, registering the ledger dataset based on the dual-channel architecture, and generating a multi-dimensional ledger data association model are as follows: Construct a dual-channel architecture, which includes an anchor channel and an event mapping channel; Among them, the anchor channel is used to store the data entities in the ledger dataset and assign a data anchor number to each data entity. The data entities in the anchor channel cannot be modified after being written. The event mapping channel is used to generate the relationships between data entities and the prefix number of new data entities based on preset key business events; Each data record in the ledger dataset is treated as a data entity and registered in the anchor channel. During registration, the anchor channel assigns a data anchor number to the data entity and establishes a storage unit. The storage unit carries the data anchor number, entity type, original metric value, and dimension identifier. Configure event mapping points in the event mapping channel. Each event mapping point corresponds to a key business event type. Event mapping points are connected through event channels. The direction of the event channels is determined by the flow order of key business events in the agent business ledger. Numbering rules are pre-stored in the event mapping points. When a critical business event is triggered, the corresponding event mapping point generates a prefix number for the new data entity according to the numbering rules, and assigns a data anchor number to the corresponding new data entity based on the prefix number. When generating a prefix number, the event mapping point generates a probe factor, and weak connection segments between data entities are constructed based on the probe factor; Record the data anchor number of the data entity participating in the key business event as the source entity identifier, record the data anchor number of the new data entity as the target entity identifier, record the event type code as the association type, and generate association edges; A multidimensional ledger data association model is generated by combining all associated edges and all data entities based on associated edges and weak connection segments.

4. The method of claim 3, wherein, The step of assigning data anchor point numbers to the corresponding new data entities based on the preceding number is as follows: The event mapping point obtains the data anchor point numbers of all data entities participating in the current key business event, forming a set of data anchor point numbers; According to the numbering segmentation rules, extract number fragments from the specified positions of each data anchor number in the data anchor number set; The number segmentation rules include the definition of the starting position of the truncation, the definition of the truncation length, the definition of the event type encoding, and the definition of the combination order. All extracted number fragments are combined with the event type codes of the current key business events, and then concatenated based on the combination order definition to serve as the pre-encoding of the new data entity. After the event mapping point generates a prefix number, it sends a new data entity registration request to the anchor channel. The registration request includes the prefix number, the entity type of the new data entity, the original metric value, and the dimension identifier. Based on the registration request, the anchor channel assigns a unique data anchor number to the new data entity and establishes a corresponding storage unit. The storage unit carries the preceding number, data anchor number, entity type, original metric value, and dimension identifier.

5. The method of claim 3, wherein, The steps for constructing weak links between data entities based on the probe factor are as follows: When generating a prefix number, the feature segment of the prefix number is extracted, the feature segment is encapsulated into multiple independent detection factors, and the detection factors are released into the event mapping channel. When the detection factor detects the intersection event channel, the prefix number of the new data entity in the event channel is detected based on the detection factor. The feature segment consists of the event type code carried in the prefix number and the data anchor number fragment of at least one data entity participating in the key business event. The feature segment is used to characterize the source entity characteristics and event type characteristics of the prefix number. If a new data entity with a number similarity higher than a preset threshold is detected, the new data entity submits a weak connection request to the data entity or new data entity that sent the probe factor and generates a weak connection segment.

6. The method of claim 1, wherein, The steps for generating multiple visual structure outputs based on multiple observation windows are as follows: A multidimensional mapping space is constructed. The outer wall of the multidimensional mapping space is composed of multiple dimensional walls. Each dimensional wall corresponds to a dimension identifier of the ledger dataset. Access areas and isolation areas are deployed inside the multidimensional mapping space. Randomly generated chaotic data is stored in the access areas. Among them, the dimension wall includes the time dimension wall, the regional dimension wall, the business type dimension wall, the customer type dimension wall, and the service personnel dimension wall; For each data entity in the association model, spatial coordinates are assigned to an isolated region in the multidimensional mapping space according to the dimension identifier carried by the data entity and the association relationship of the data entity. For a data entity that carries a dimension identifier, the coordinate components of the data entity in that dimension are determined by the dimension identifier. For a dimension in which a data entity does not carry a dimension identifier, the coordinate components of the data entity in that dimension are determined by the position of the dimension identifiers of other data entities that have associated edges with that data entity; The associated edges and weak connection segments in the association model are mapped to the connection lines between data entities in the multidimensional mapping space. The associated edges are mapped to solid connection lines, and the weak connection segments are mapped to dashed connection lines. Multiple observation windows are set on each dimensional wall. The observation windows are located on the surface of the dimensional wall, and the observation direction of the observation windows is from the dimensional wall where the observation window is located to the interior of the multidimensional mapping space. Each dimension wall has only one observation window that is the correct observation window. The correct observation window stores the access address path of the isolated region, while the other observation windows store the access address path of the access region. For a specific dimensional wall, the correct observation window generates the observation direction. Based on the observation direction, the association model of the isolated area is observed, and the association relationship between data entities in the corresponding dimensional wall is generated as a single-dimensional visualization structure output. At the same time, the correlation model under this dimension is projected onto the dimension wall in the opposite direction of observation to generate the correlation relationship under the dimension of the opposing dimension wall, which serves as the output of a multi-dimensional visualization structure. The single-dimensional visualization structure output and the multi-dimensional visualization structure output are integrated into a single visualization structure output.

7. The method of claim 1, wherein the method further comprises: The steps for combining the visualization structure outputs from different observation directions to generate and output a visualization view based on the visualization structure output are as follows: Obtain the visualization structure output generated from the observation windows on the multi-dimensional wall. Each visualization structure output includes data entities and the relationships between data entities. Each data entity in the output of the visualization structure is converted into a visualization node, and the display position of the visualization node is determined by the spatial coordinate projection of the data entity in the multidimensional mapping space. Solid lines are converted into first visual lines, and dashed lines are converted into second visual lines. The first visual lines represent associated edges, and the second visual lines represent weak connection segments. All visualization nodes, the first visualization connection, and the second visualization connection are arranged according to the dimension wall where the corresponding observation window is located, generating a multi-view visualization view. Each display area of ​​the multi-view visualization view corresponds to the visualization structure output of the observation window of a dimension wall.

8. A multi-dimensional analysis and visualization system for agency business ledger data, used to implement the multi-dimensional analysis and visualization method for agency business ledger data as described in any one of claims 1-7, characterized in that, Data partitioning module: Collects multidimensional ledger data, performs dimensional partitioning on the multidimensional ledger data, and generates a dimension-based ledger dataset; Model building module: Connected to the data partitioning module, it constructs a dual-channel architecture, performs registration processing on the ledger dataset based on the dual-channel architecture, and generates a multi-dimensional ledger data association model; Mapping Space Module: Connects to the model building module to construct a multi-dimensional mapping space, mapping related models to the multi-dimensional mapping space, pre-setting multiple observation windows, and generating multiple visualization structure outputs based on the multiple observation windows; Visualization module: Connected to the mapping space module, based on the visualization structure output, it combines the visualization structure outputs from different observation directions to generate and output a visualization view.