Interactive multi-dimensional data visual display sand table and method for commercial data analysis

Abstract

The invention provides an interactive multi-dimensional data visualization display sand table and method for business data analysis, and the method comprises the steps: carrying out the semantic mapping of state parameters in an interaction control operation when the interaction control operation is received, obtaining a plurality of pieces of visualization intention metadata, and generating a structured query instruction; based on the structured query instruction, extracting a target data subset rendered visually by the heterogeneous commercial data source from a multi-dimensional data cube of the commercial data source, performing visual optimization on the visually rendered chart type, the visual coding channel and the initial visual parameter through the cardinal number of the dimension field in the target data subset and the data distribution type and the numerical range of the measurement field to obtain a visual configuration set of the target data subset; and using the visualization configuration set to render and generate a multi-dimensional data visualization view, and outputting and displaying the data visualization view in a graphical user interface. Based on the scheme, automatic closed loop generation from heterogeneous data to interactive views can be realized.

Description

Technical Field

[0001] This application relates to the field of data visualization technology, and more specifically, to an interactive multi-dimensional data visualization display sandbox and method for business data analysis. Background Technology

[0002] A data visualization sandbox is an enterprise-level analytics platform that integrates multi-source data processing, a real-time analysis engine, and interactive graphics rendering. It allows users to dynamically explore and simulate scenarios using natural interactions such as drag-and-drop, filtering, and drill-down. Its core function is to transform complex data into intuitively operable visual objects, creating an immersive analytics environment with flexible layouts, multi-view linkage, and the ability to record the analysis path. This lowers the barrier to entry for professional analysis and improves decision-making efficiency and collaboration capabilities.

[0003] In traditional technical architectures, building a business data analytics system first faces a high degree of manual reliance on model definition and data pipelines. Data engineers must manually design star / snowflake schema data models based on a prior understanding of the business scenario, explicitly specifying the structure and relationships of dimension tables, fact tables, and so on. Simultaneously, they must write complex scripts to handle format differences, value range unification, and quality cleaning of heterogeneous data sources. This process is not only time-consuming but also leads to the premature solidification of business logic into the code and table structure. When business requirements change, existing technical solutions often require tracing and modifying ETL logic, refactoring data models, rerunning historical data, and even adjusting the semantic layer of downstream reports. This chain of dependencies results in long and costly system response cycles to changes, making it difficult to adapt to agile iterative business analytics scenarios and creating a fundamental contradiction between the evolution of business requirements and the rigidity of technical implementation. Therefore, how to achieve an automated closed loop from heterogeneous data to interactive views, thereby improving the intuitiveness of business data analytics, has become a challenge facing the industry. Summary of the Invention

[0004] This application provides an interactive, multi-dimensional data visualization sandbox and method for business data analysis, which can realize an automated closed loop from heterogeneous data to interactive views, thereby improving the intuitiveness of business data analysis.

[0005] In a first aspect, this application provides an interactive, multi-dimensional data visualization method for business data analysis, including: The system acquires raw data records from heterogeneous business data sources, identifies and standardizes the dimension fields and measure fields in the raw data records, performs multi-level aggregation on all dimension fields through each measure field, and then constructs a multi-dimensional data cube of the heterogeneous business data source in memory. When an interactive control operation is received, the state parameters in the interactive control operation are semantically mapped to obtain multiple visualization intent metadata. Then, all visualization intent metadata is converted into a structured query instruction containing target dimension, target metric, expected chart type and data filtering conditions. Based on the structured query instructions, a subset of target data for visualization rendering of heterogeneous business data sources is extracted from the multidimensional data cube. Visual optimization is performed on the chart type, visual encoding channel and initial visual parameters of the visualization rendering by using the cardinality of the dimension fields, the data distribution type and numerical range of the metric fields in the target data subset, to obtain a visualization configuration set for the target data subset. The visualization configuration set is used to render and generate a multi-dimensional data visualization view, which is then output and displayed in a graphical user interface.

[0006] In some embodiments, identifying and standardizing dimension fields and measure fields in the original data record specifically includes: Read the structured metadata and unstructured sample data of the original data records, and identify candidate dimension fields and measure fields based on the preset field naming rule library and data pattern inference algorithm; Semantic disambiguation and category merging are performed on candidate dimension fields to unify the definitions of fields with the same meaning but different names, and a hierarchical relationship of dimensions is established based on the domain knowledge graph to obtain the dimension fields; The candidate measure fields are validated for data type and standardized for unit conversion. All numerical measures are unified to the benchmark unit of measurement and their aggregate function types are labeled to obtain the measure fields.

[0007] In some embodiments, multi-level aggregation of all dimension fields through various metric fields is performed to construct a multi-dimensional data cube of heterogeneous business data sources in memory, specifically including: Based on the hierarchical relationship of dimensions, multiple aggregated index tables with different granularities are pre-calculated for each dimension field. These aggregated index tables support roll-up and drill-down operations along the hierarchy. Based on the aggregation function type of the metric field, the corresponding metric summary value and statistical feature value are pre-calculated and cached in memory for each dimension combination granularity; By establishing a star-shaped logical model between dimensions and measures through various aggregated index tables, all measure summary values, and statistical feature values, a multidimensional data cube that supports online processing operations is formed.

[0008] In some embodiments, semantic mapping of state parameters in the operation of the interactive control to obtain multiple visual intent metadata specifically includes: Capture the various state parameters during the operation of the interactive control; Each state parameter is mapped to the corresponding dimension field, metric field, and filter condition expression in the multidimensional data cube to form the original intent parameters that the machine can understand. Based on the combination logic of controls and business rules, infer the user's potential analytical intent and supplement the generation of implicit analytical dimensions and metrics; The original intent parameters, implicit analysis dimensions, and analysis metrics are fused to form multiple visualized intent metadata.

[0009] In some embodiments, converting all visualization intent metadata into structured query instructions containing target dimensions, target metrics, desired chart types, and data filtering conditions specifically includes: Conflict detection and priority arbitration are performed on the metadata of each visualization intent. When there are mutually exclusive or redundant intents, they are selected or merged according to preset rules. Convert the intent parameters in the arbitrated visual intent metadata into standard query statement fragments, and clarify the target dimension, target metric, and data filtering conditions of the query. Based on user historical preferences and global visualization settings, multiple recommended expected chart types are specified for the query results, and then the target dimension, target metric, expected chart type and data filtering conditions are encapsulated into structured query instructions.

[0010] In some embodiments, extracting a subset of target data for visualization rendering of heterogeneous business data sources from the multidimensional data cube based on the structured query instruction specifically includes: Apply the filter conditions in the structured query command to the corresponding dimensions of the multidimensional data cube to determine the data range slice for the query; Based on the target dimension and target metric in the structured query instruction, locate the corresponding pre-aggregated data block in the multidimensional data cube and trigger real-time aggregation calculation; The results of the data range slicing and real-time aggregation calculations are organized into a result set according to the format and order required by the query, and contextual metadata is attached to form a target data subset for visualization rendering.

[0011] In some embodiments, visual optimization of the chart type, visual encoding channel, and initial visual parameters for visualization rendering is performed based on the cardinality of the dimension fields, the data distribution type of the metric fields, and the numerical range of the target data subset. This results in a visualization configuration set for the target data subset, specifically including: The role of each dimension field in the chart—faceted, colored, or an axis—is automatically determined based on the cardinality of the data in the target subset. Using the data distribution type and numerical range of each metric field in the target data subset, match the chart type that displays the distribution characteristics; Based on the principle of visual perception effectiveness, appropriate visual encoding channels are assigned to the selected chart type, and the initial visual parameters are optimized based on the numerical range of each metric field to generate a visualization configuration set for the target data subset.

[0012] Secondly, this application provides an interactive, multi-dimensional data visualization sandbox for business data analysis, comprising: The acquisition module is used to acquire the original data records in the heterogeneous business data source, and then identify and standardize the dimension fields and measure fields in the original data records. Through each measure field, all dimension fields are aggregated in multiple layers, and then a multi-dimensional data cube of the heterogeneous business data source is constructed in memory. The processing module is used to perform semantic mapping on the state parameters in the interactive control operation when an interactive control operation is received, to obtain multiple visualization intent metadata, and then convert all the visualization intent metadata into a structured query instruction containing target dimension, target metric, expected chart type and data filtering conditions. The processing module is also used to extract a subset of target data for visualization rendering of heterogeneous business data sources from the multidimensional data cube based on the structured query instruction, and to perform visual optimization on the chart type, visual encoding channel and initial visual parameters of the visualization rendering by using the cardinality of the dimension field, the data distribution type and numerical range of the metric field in the target data subset, so as to obtain a visualization configuration set of the target data subset. The execution module is used to render and generate a multi-dimensional data visualization view using the visualization configuration set, and output and display the data visualization view in the graphical user interface.

[0013] Thirdly, this application provides a computer device, the computer device including a memory and a processor, the memory for storing computer programs, and the processor for calling and running the computer programs from the memory, so that the computer device performs the above-described interactive multi-dimensional data visualization display method for business data analysis.

[0014] Fourthly, this application provides a computer-readable storage medium storing instructions or code that, when executed on a computer, cause the computer to implement the aforementioned interactive multi-dimensional data visualization method for business data analysis.

[0015] The technical solutions provided by the embodiments disclosed in this application have the following beneficial effects: This application provides an interactive multi-dimensional data visualization sandbox and method for business data analysis. The method involves acquiring raw data records from heterogeneous business data sources, identifying and standardizing dimension and metric fields within these records, and then aggregating all dimension fields using various metric fields to construct a multi-dimensional data cube of the heterogeneous business data source in memory. When an interactive control operation is received, the state parameters of the operation are semantically mapped to obtain multiple visualization intent metadata. All visualization intent metadata is then converted into structured query instructions containing target dimensions, target metrics, desired chart types, and data filtering conditions. Based on these structured query instructions, a target data subset for visualization rendering of the heterogeneous business data source is extracted from the multi-dimensional data cube. Visual optimization is performed on the chart type, visual encoding channel, and initial visual parameters of the visualization rendering based on the cardinality of the dimension fields, the data distribution type of the metric fields, and the numerical range of the target data subset, resulting in a visualization configuration set for the target data subset. This visualization configuration set is then used to render a multi-dimensional data visualization view, which is then displayed in a graphical user interface.

[0016] Therefore, in this application, the visualization configuration set is used to render and generate a multi-dimensional data visualization view, which is then displayed in a graphical user interface. First, by determining the structured query instructions, a data request specification that can be accurately executed by the machine is obtained, thus establishing a programmable semantic bridge between the data layer and the interaction layer. This transforms unstructured user interaction intentions into a structured description containing precise dimensions, measures, filtering conditions, and chart preferences, achieving automated mapping from fuzzy business problems to computable data queries. This eliminates the bottleneck of relying on manually written query statements or predefined report models in traditional analysis, enabling the sandbox to dynamically respond to users' immediate exploration needs. This not only improves the real-time performance of the analysis response but, more importantly, liberates users from tedious technical details, allowing them to focus entirely on the construction and verification of business logic. Then, by determining the visualization configuration set, a visualization rendering scheme that conforms to best practices in visual coding and the inherent characteristics of the data is obtained, thus realizing the intelligent transformation from abstract data to effective visual representation. Based on the statistical characteristics of the data subset itself, the system automatically decides on chart types, visual channel mapping, and aesthetic parameters, constructing a data-driven visual design engine. Its core benefit lies in overcoming the cognitive load and design hurdle caused by users manually configuring charts in traditional tools. It ensures that the generated views achieve professional standards in terms of visual effectiveness and information delivery efficiency, enabling the sand table to adaptively present the most suitable data narrative form, intuitively exposing data patterns to observers, significantly enhancing the intuitiveness of analysis and the efficiency of insight discovery, and completing the final automated closed loop from data to insight. In summary, based on the above solution, an automated closed loop for generating interactive views from heterogeneous data can be achieved, thereby improving the intuitiveness of business data analysis. Attached Figure Description

[0017] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0018] Figure 1 This is an exemplary flowchart of an interactive, multi-dimensional data visualization method for business data analysis, as shown in some embodiments of this application. Figure 2 This is a flowchart illustrating the process of determining the visual configuration according to some embodiments of this application; Figure 3 This is a structural schematic diagram of an interactive, multi-dimensional data visualization display sandbox for business data analysis, as shown in some embodiments of this application; Figure 4This is a schematic diagram of the structure of a computer device that implements an interactive multi-dimensional data visualization method for business data analysis, according to some embodiments of this application. Detailed Implementation

[0019] To better understand the technical solution of this application, the technical solution of this application will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0020] refer to Figure 1 The figure is an exemplary flowchart of an interactive multi-dimensional data visualization method for business data analysis, according to some embodiments of this application. This interactive multi-dimensional data visualization method for business data analysis mainly includes the following steps: In step 101, the original data records in the heterogeneous business data source are obtained, and then the dimension fields and measure fields in the original data records are identified and standardized. All dimension fields are aggregated in multiple layers through each measure field, and then a multidimensional data cube of the heterogeneous business data source is constructed in memory.

[0021] It should be noted that, in this application, a business data source refers to a collection of structured or semi-structured data that is generated, stored, or exchanged in business activities and operations and has potential analytical value, including but not limited to: relational databases, NoSQL databases, data warehouses, application programming interfaces, and static data files; raw data records are the smallest units of data that have not undergone business-oriented aggregation and transformation processing, and each record represents an instance of a fact or event, such as: a specific order, a user's page click, a financial transaction, or a snapshot of a customer's attributes.

[0022] In practical implementation, heterogeneous commercial data sources are connected and accessed through a configured adapter. First, metadata probing technology is used to automatically read the table structure, field types, comments, and sample data of the data sources to gain an initial understanding of their patterns. Next, pattern mapping and semantic annotation are performed: based on a predefined commercial ontology or machine learning model, the business meaning of fields is identified. For example, field names are mapped to standard dimensions or measures such as "product category" and "sales amount," and dimension types such as time, region, and category are distinguished. Then, data cleaning and standardization are performed: date formats and currency units are standardized, missing and outlier values ​​are handled, and the value range of category dimensions is normalized. Finally, through incremental or full synchronization mechanisms, the processed raw data records are loaded into a unified data staging area or message queue, providing high-quality, standardized input for subsequent construction of a multidimensional data cube.

[0023] In some embodiments, identifying and standardizing dimension fields and measure fields in the original data records can be achieved by the following steps: Read the structured metadata and unstructured sample data of the original data records, and identify candidate dimension fields and measure fields based on the preset field naming rule library and data pattern inference algorithm; Semantic disambiguation and category merging are performed on candidate dimension fields to unify the definitions of fields with the same meaning but different names, and a hierarchical relationship of dimensions is established based on the domain knowledge graph to obtain the dimension fields; The candidate measure fields are validated for data type and standardized for unit conversion. All numerical measures are unified to the benchmark unit of measurement and their aggregate function types are labeled to obtain the measure fields.

[0024] It should be noted that, in this application, structured metadata is the raw input used to describe the structural information of the data source itself, while unstructured sample data is the raw input used to initially explore the data content and quality of the data source; the preset field naming rule library is a set of mapping rules used to match the raw field names in the data source with known business terms; the domain knowledge graph is a structured knowledge model used to define conceptual entities and their hierarchical and relational relationships in a specific business domain; dimension fields are standardized business perspectives with established clear semantics and hierarchical relationships used to observe, classify, slice, and group business facts; and metric fields are standardized business indicators used for quantitative calculation and analysis.

[0025] In practical implementation, firstly, the structured metadata and unstructured sample data of the original data records are read. Based on a preset field naming rule library and data pattern inference algorithm, candidate dimension fields and measure fields are identified. This can be achieved in the following way: Two types of information from the data source are read in parallel: firstly, structured metadata such as table name, column name, data type, and whether it is nullable is obtained by querying the database's system tables or reading the header information of the data file; secondly, a small number of actual data rows are sampled from the data source as unstructured sample data. The original field names are matched with keywords in the preset field naming rule library. For example, fields containing "date" and "time" are initially marked as time dimension candidates, and fields containing "amount," "quantity," and "rate" are initially marked as measure candidates. Simultaneously, a data pattern inference algorithm is applied to the sample data. This algorithm analyzes statistical characteristics such as the uniqueness ratio (cardinality), numerical distribution, and string patterns of field values. For example, for fields with a high uniqueness ratio and that are numerical, the algorithm infers that they are entity IDs (dimension candidates). For fields with a large number of numerical values ​​and capable of mathematical operations, they are inferred to be candidates for measure fields. A list of candidate dimension fields and measure fields, marked with the combined results of naming rule matching and pattern inference, is output as an initial classification. Then, semantic disambiguation and category merging are performed on the candidate dimension fields to unify the definitions of synonymous but different fields. A hierarchical relationship between dimensions is established based on the domain knowledge graph. Dimension fields can be implemented in the following way: semantic integration of candidate dimension fields is performed. For example, different data tables may name the same concept "product category," "category," or "Prod_Category." These names are merged into a unified standardized name, such as "product category," through a thesaurus or string similarity calculation, completing semantic disambiguation and category merging. A hierarchical relationship is established for the standardized dimension fields based on the pre-constructed domain knowledge graph. For example, under the "region" dimension, the knowledge graph defines a hierarchical structure of "country-province-city." The system assigns this structure to the "region" field, clarifying its drill-down path.After this step, the original, messy candidate fields are transformed into standardized dimension fields with clear business meaning and a clear hierarchical structure. Finally, the candidate metric fields undergo data type validation and unit standardization conversion, unifying all numerical metrics to the benchmark unit of measurement and labeling their aggregate function types. The resulting metric fields can be implemented as follows: First, the candidate metric fields undergo data type validation to ensure their storage or representation type (e.g., integer, decimal) meets the requirements of numerical calculation; those that do not meet the requirements are forcibly converted or marked as anomalies. Then, unit standardization conversion is performed based on a preset conversion rule table, applying this to all... Numerical values ​​are standardized to a benchmark unit of measurement. For example, amounts expressed in different units such as "ten thousand yuan" and "thousand US dollars" are all converted to the benchmark unit of "yuan (RMB)". Then, the aggregation function type of each metric field is labeled. This is done by analyzing the business semantics or configuration information of the field. For example, "sales amount" should use "summation" aggregation, "inventory quantity" may use "last value" or "average value" aggregation, and "gross profit margin" needs to be calculated by summing the numerator and denominator separately before calculating the ratio. After completing the above three steps, the original numerical fields can be transformed into standardized metric fields with reliable data types, comparable units, and clear aggregation logic.

[0026] In some embodiments, the following steps can be used to construct a multidimensional data cube of a heterogeneous business data source in memory by performing multi-level aggregation on all dimension fields through various metric fields: Based on the hierarchical relationship of dimensions, multiple aggregated index tables with different granularities are pre-calculated for each dimension field. These aggregated index tables support roll-up and drill-down operations along the hierarchy. Based on the aggregation function type of the metric field, the corresponding metric summary value and statistical feature value are pre-calculated and cached in memory for each dimension combination granularity; By establishing a star-shaped logical model between dimensions and measures through various aggregated index tables, all measure summary values, and statistical feature values, a multidimensional data cube that supports online processing operations is formed.

[0027] It should be noted that in this application, the dimension hierarchy is a data structure used to describe the parent-child relationship within a single analytical perspective, which can be progressively subdivided from general to specific; the aggregated index table is an inverted index structure used to locate and associate data records at different summary levels; "roll-up" refers to viewing aggregated data from detailed data to summary data, and "drill-down" refers to viewing expanded data from summary data to detailed data; the aggregation function type of the metric field is used to define the operational rules for summarizing numerical business indicators; the granularity of dimension combination refers to the data summary level jointly determined by the values ​​taken by multiple different dimensions at their respective levels; the metric summary value and statistical feature value are used for... This reflects the pre-calculated results of relevant business indicators (e.g., total sales) and key distribution characteristics (e.g., maximum, minimum, number of records) at a specific dimensional combination granularity; aggregated index tables, measure summary values, and statistical feature values ​​are the core data structures and pre-calculated results used to build multidimensional data models; the star schema is a relational model used to organize data, in which the "fact table" at the center stores measure values ​​and foreign keys associated with each dimension, while the surrounding "dimension tables" store detailed descriptions of dimension members; the multidimensional data cube is a data storage and computation structure used to achieve fast, interactive business intelligence analysis.

[0028] In practice, firstly, based on the hierarchical relationship of dimensions, multiple aggregated index tables of various granularities are pre-calculated for each dimension field. The aggregated index tables support roll-up and drill-down operations along the hierarchy, which can be implemented in the following way: based on the established hierarchical relationship of dimensions, pre-calculation is performed independently for each dimension field. For each granularity in the hierarchy (e.g., "country", "province", "city"), the system will scan the corresponding dimension members in the original fact data and create an aggregated index table for each of them. In this table, each unique dimension member value (e.g., "China", "Zhejiang Province", "Hangzhou City") is recorded as an entry, maintaining a set or bitmap index of the row identifiers corresponding to that member across all data records. For hierarchical relationships, the association is also maintained in the index. For example, in the aggregated index table at the "province" granularity, the entry "Zhejiang Province" will record the data rows corresponding to all its cities (e.g., "Hangzhou City", "Ningbo City"), allowing subsequent queries to quickly drill down from the "province" level to the "city" level. A fast lookup and relationship mapping structure covering all summary levels is constructed for each analytical perspective, providing a foundation for efficient data navigation. The dimension member values ​​and their associated data location information are used as aggregated index tables to support fast navigation between different granularities. Then, based on the aggregation function type of the metric field, the corresponding metric summary values ​​and statistical features are pre-calculated and cached in memory for each dimension combination granularity. The values ​​can be implemented in the following way: based on all possible dimension combination granularities, all measure fields are calculated in parallel. According to the aggregation function type of the measure field, the specified aggregation operation is performed on all raw data records falling within each dimension combination granularity (e.g., "Q1 2023 - East China"). For measures defined as "summation", the corresponding values ​​of all records within that granularity are added together. For measures defined as "average", the sum is calculated and divided by the number of records. In addition to these core measure summary values ​​(e.g., total sales, average price), some key statistical feature values ​​are also calculated and cached at the same time, such as the total number of data records under this granularity, the maximum and minimum values ​​of the measure values. The pre-calculated results are stored in memory in the form of key-value pairs, where the key is the unique code of the dimension combination granularity. By sacrificing some storage space, the most time-consuming aggregation calculations in the analysis query are completed in advance, preparing the data for extremely fast query response.Finally, a star-shaped logical model is established between dimensions and measures using various aggregated index tables, all measure summary values, and statistical feature values. This forms a multidimensional data cube that supports online processing. This can be achieved as follows: the aggregated index tables, measure summary values, and statistical feature values ​​for each dimension and its granularity are organized according to a star-shaped logical model. In this model, all measure summary values ​​and statistical feature values, along with their corresponding dimension combination granularity codes, collectively constitute a logical "central fact table," while the set of aggregated index tables for each dimension collectively constitutes the logical "dimension table" for that dimension. By associating dimension member values ​​with granular encoding of dimension combinations in the fact table, when a user issues an analytical query (e.g., querying "sales revenue for each quarter of 2023"), the system can first locate the data range of "2023" and its subordinate "quarters" using the time-dimensional aggregated index table. Then, it can directly retrieve the pre-calculated sum of "sales revenue" corresponding to these granularities from memory, without needing to scan and calculate on the raw, massive dataset in real time. By integrating the pre-calculated index with the summary results, the raw, scattered business data is transformed into a multidimensional data cube supporting online processing.

[0029] In step 102, when an interactive control operation is received, the state parameters in the interactive control operation are semantically mapped to obtain multiple visualization intent metadata, and then all visualization intent metadata are converted into structured query instructions containing target dimension, target metric, expected chart type and data filtering conditions.

[0030] In some embodiments, semantic mapping of state parameters in the operation of the interactive control to obtain multiple visual intent metadata can be achieved by the following steps: Capture the various state parameters during the operation of the interactive control; Each state parameter is mapped to the corresponding dimension field, metric field, and filter condition expression in the multidimensional data cube to form the original intent parameters that the machine can understand. Based on the combination logic of controls and business rules, infer the user's potential analytical intent and supplement the generation of implicit analytical dimensions and metrics; The original intent parameters, implicit analysis dimensions, and analysis metrics are fused to form multiple visualized intent metadata.

[0031] It should be noted that, in this application, the state parameter is a data value used to reflect the specific state of the element after the user interacts with the interactive elements of filters, drop-down lists, sliders, and charts in the graphical user interface; the original intent parameter is an intermediate representation used to initially translate the user interface interaction actions into data query elements and constraints that the computer system can directly process; the combination logic and business rules of the controls are a set of logical judgments used to define the mutual influence relationships between different interface controls and the analysis conventions and constraints within the business domain; the potential analysis intent is used to refer to the deeper analysis goals that the user does not explicitly specify when operating the interface but can reasonably infer based on their operation mode and business common sense; the implicit analysis dimensions and analysis metrics are used to supplement and improve the data analysis perspectives and quantitative indicators required for the user's explicit query; and the visualization intent metadata is a structured data object set used to describe all data elements and their interrelationships required for a visualization analysis request.

[0032] In specific implementation, firstly, capturing the various state parameters in the interactive control operations can be achieved in the following way: by listening to the events of all interactive controls in the graphical user interface (e.g., dimension selector, measure selector, date range selector, chart type selection button), after the user completes an interactive action (e.g., clicking to select, dragging a range), immediately capture and record the current state of these controls. For example, capture the currently selected items as "product category" and "city" in the dimension selector, "sales amount" in the measure selector, and the selected range of the date slider as "2023-01-01 to 2023-12-31". All captured values ​​together constitute the various state parameters reflecting the result of the user's interactive operation; secondly, the various state parameters are... Mapping state parameters to corresponding dimension fields, measure fields, and filter expressions in a multidimensional data cube to form machine-understandable raw intent parameters can be achieved as follows: Based on a pre-configured mapping dictionary, each state parameter is translated into corresponding standardized data elements and query logic in the multidimensional data cube. For example, "Product Category" on the interface is mapped to a dimension field named "prod_category" in the data cube; "Sales Amount" is mapped to a measure field named "sales_amount"; and the date range "2023-01-01 to 2023-12-31" is converted into a filter expression for the "order_date" dimension field, in the form "order_date". "BETWEEN '2023-01-01' AND '2023-12-31'"; Through this mapping, user interface operations are transformed into raw intent parameters that the internal data model can directly recognize and process, completing the initial transformation from interactive actions to data query elements; Then, based on the combination logic of controls and business rules, the user's potential analytical intent is inferred, and implicit analytical dimensions and metrics are generated. This can be achieved in the following way: Analyze the combination logic between raw intent parameters and combine it with business rules within the domain to infer potential analytical intents that the user may not have explicitly stated. For example, if the user only selects a time dimension "year" and a metric "profit", the business rules of "time trend analysis" can infer that the user may want to perform trend analysis, thereby automatically supplementing a more granular implicit time analysis dimension "quarter" so that the chart can show the changing trend; as another example, if the user selects both "sales" and "cost" metrics, the combination logic of "profit analysis" can infer and supplement an implicit analytical metric "gross profit margin" as a derived indicator for calculation, thereby enabling the system to more intelligently understand the user's analytical goals;Finally, the original intent parameters, implicit analysis dimensions, and analysis metrics are fused to form multiple visualization intent metadata. This can be achieved by merging the original intent parameter set with the inferred implicit analysis dimension and analysis metric set. This process is called intent fusion. Intent fusion includes checking and eliminating duplicates (e.g., implicit dimensions are not added if they have been explicitly specified), resolving potential conflicts (e.g., explicit filtering takes priority when it conflicts with inference logic), and sorting the analysis elements according to a preset priority. After fusion, all confirmed and organized dimensions, metrics, filtering conditions, and chart types initially suggested based on these elements are encapsulated together to form one or more complete and self-consistent visualization intent metadata, thus obtaining multiple visualization intent metadata.

[0033] In some embodiments, converting all visualization intent metadata into a structured query instruction containing target dimensions, target metrics, desired chart type, and data filtering conditions can be achieved through the following steps: Conflict detection and priority arbitration are performed on the metadata of each visualization intent. When there are mutually exclusive or redundant intents, they are selected or merged according to preset rules. Convert the intent parameters in the arbitrated visual intent metadata into standard query statement fragments, and clarify the target dimension, target metric, and data filtering conditions of the query. Based on user historical preferences and global visualization settings, multiple recommended expected chart types are specified for the query results, and then the target dimension, target metric, expected chart type and data filtering conditions are encapsulated into structured query instructions.

[0034] It should be noted that, in this application, structured query instructions are standardized instruction pairs used to encapsulate all elements of a complete data analysis request; standard query statements are components used to construct data query requests that conform to grammatical specifications and are issued to a data cube or database engine.

[0035] In practice, the first step is to perform conflict detection and priority arbitration on the metadata of each visualization intent. When mutually exclusive or redundant intents exist, they are selected or merged according to preset rules. This can be achieved in the following way: After receiving multiple visualization intent metadata, the metadata is first traversed, and the dimensions, metrics, and filtering conditions specified within the visualization intent metadata are compared and analyzed. Conflict detection specifically includes identifying mutually exclusive conflicts, such as one intent requiring filtering "product category is electronic products" while another intent requires filtering "product category is clothing," both using mutually exclusive values ​​for the same dimension; and identifying redundant conflicts, such as multiple intents containing the same "annual sales" metric, or one intent's filtering condition (e.g., "year equals 2023") being completely contained by another intent's filtering condition (e.g., "quarter belongs to the first quarter of 2023"). When a conflict or redundancy is detected... At this time, the system processes the data according to preset selection or merging rules. For example, the rules may stipulate that the most recent intent is selected under mutually exclusive filtering conditions, or that redundant identical metrics are automatically deduplicated and merged. Through the above steps, the scattered and potentially contradictory preliminary intents can be organized into a logically clear and element-unique final request specification. Then, the intent parameters in the arbitrated visual intent metadata are converted into standard query statement fragments. The target dimension, target metric, and data filtering conditions can be clearly defined in the following way: the arbitrated visual intent metadata is converted, specifically, the confirmed target dimension name, target metric name, and data filtering condition expression are read; according to the query language specification supported by the backend data cube, each element is converted into a corresponding query statement fragment. For example, the target dimension names "region" and "product category" are converted into "GROUP" in the query language. The query statement fragments collectively define the specific data content and scope required for the query. This is achieved by converting the target metric name "Sales Amount" into an aggregation expression like "SUM(Sales Amount)". The data filtering condition "Time in 2023" is converted into a filter clause like "WHERE Year=2023". Finally, based on user historical preferences and global visualization settings, multiple recommended chart types are specified for the query results. The target dimension, target metric, expected chart type, and data filtering conditions are then encapsulated into a structured query instruction. This can be implemented as follows: after determining the data elements to be queried, the query is not executed immediately. Instead, the display method of the results is determined first. A knowledge base is queried, which integrates user historical preferences and global visualization settings. User historical preferences record the chart type that the user most frequently selected or spent the longest time on when faced with similar data features (e.g., including a time dimension and a single metric). The global visualization settings include some general rules, such as "when the number of dimensions is 1 and it is a time type, and the number of metrics is 1, a line chart is preferred."The system integrates these two aspects of information to calculate one or more recommended chart types and their confidence rankings, which are then used as the desired chart types. The target dimension, target metric, data filtering conditions, and desired chart types are encapsulated together into a data structure with a fixed format, which is the structured query instruction.

[0036] In step 103, a subset of target data for visualization rendering of heterogeneous business data sources is extracted from the multidimensional data cube based on the structured query instruction. Visual optimization is performed on the chart type, visual encoding channel and initial visual parameters of the visualization rendering by using the cardinality of the dimension fields, the data distribution type and numerical range of the metric fields in the target data subset, to obtain a visualization configuration set of the target data subset.

[0037] In some embodiments, extracting the target data subset for visualization rendering of heterogeneous business data sources from the multidimensional data cube based on the structured query instructions can be achieved through the following steps: Apply the filter conditions in the structured query command to the corresponding dimensions of the multidimensional data cube to determine the data range slice for the query; Based on the target dimension and target metric in the structured query instruction, locate the corresponding pre-aggregated data block in the multidimensional data cube and trigger real-time aggregation calculation; The results of the data range slicing and real-time aggregation calculations are organized into a result set according to the format and order required by the query, and contextual metadata is attached to form a target data subset for visualization rendering.

[0038] It should be noted that, in this application, the target data subset is the data set used to drive the front-end chart component to perform graphic drawing and visual encoding; the data range slice is used to identify the subset of data to be analyzed that meets all constraints after being processed by the filtering conditions; and the pre-aggregated data block is a unit of metric summary data used to quickly respond to common analytical queries.

[0039] In practical implementation, firstly, the filtering conditions in the structured query instruction are applied to the corresponding dimensions of the multidimensional data cube. Determining the data range slice for the query can be achieved as follows: Parse the received structured query instruction and extract the filtering conditions. Each filtering condition typically consists of a dimension name, a comparison operator, and one or more values. Match these conditions one-to-one with the corresponding dimensions defined in the multidimensional data cube. For example, when the filtering condition is "product category belongs to ['electronic products', 'office supplies']", locate the dimension named "product category" in the cube and use the pre-built aggregation index table or bitmap index on this dimension to quickly find the set of memory locations of all data records that satisfy the condition. All filtering conditions work together according to logical relationships (usually "AND") to ultimately determine a set of data records that logically satisfy all constraints. This set of data records constitutes the data range slice for this query. Then, based on the target dimension and target metric in the structured query instruction, locate the corresponding pre-aggregated data block in the multidimensional data cube and trigger real-time aggregation calculation. This can be achieved as follows: Simultaneously extract the target dimension from the structured query instruction. The system searches the multidimensional data cube for pre-aggregated data blocks that perfectly match the granularity of the target dimension combination. These pre-aggregated data blocks are pre-computed and cached summary values ​​of metrics for a specific dimension combination. If a matching pre-aggregated data block is found, the system reads the data directly from it. If no perfectly matching block is found, for example, if the queried dimension combination is not pre-computed, real-time aggregation calculation is triggered. Real-time aggregation calculation takes data range slices as input and, based on the aggregation function type defined by the target metric, performs instant aggregation on the original data within the slice or pre-aggregated data with finer granularity. For example, if the target metric is "to calculate..." The summation function adds up the corresponding values ​​of all records within the slice, ensuring correct calculation results regardless of changes in query conditions. Finally, the results of the data range slicing and real-time aggregation calculations are organized into a result set according to the format and order required by the query, and contextual metadata is appended to form a target data subset for visualization rendering. This can be achieved by organizing the results of the data range slicing and aggregation calculations, arranging the results according to the format and order specified in the structured query instruction, such as sorting in ascending order by a target dimension or organizing a crosstab according to a specified row and column structure. Simultaneously, the system generates and appends rich contextual metadata, including but not limited to: the complete filtering conditions used in this query, the dimensional granularity level on which the data aggregation is based, the latest update time of the data, and the units of measurement used in the calculation. Through the above organization and encapsulation, the original numerical values ​​and dimension labels are integrated into a clearly structured and complete structured dataset.Ultimately, this series of steps transforms an abstract, structured query instruction into a subset of target data containing data and metadata, which can directly drive the chart engine to generate visual mappings and renderings.

[0040] In some embodiments, visual optimization is performed on the chart type, visual encoding channel, and initial visual parameters of the visualization rendering based on the cardinality of the dimension fields, the data distribution type of the metric fields, and the numerical range of the target data subset, to obtain the visualization configuration of the target data subset. (Refer to...) Figure 2 The diagram is a flowchart illustrating the process of determining the visualization configuration in some embodiments of this application. In this embodiment, determining the visualization configuration can be achieved through the following steps: In step 1031, the role of each dimension field in the chart is automatically determined as faceting, coloring, or coordinate axis based on the cardinality of each dimension field in the target data subset. In step 1032, the data distribution type and numerical range of each metric field in the target data subset are used to match the chart type that displays the distribution characteristics; In step 1033, based on the principle of visual perception effectiveness, appropriate visual encoding channels are assigned to the selected chart type, and the initial visual parameters are optimized based on the numerical range of each metric field to generate a visualization configuration set for the target data subset.

[0041] It should be noted that in this application, faceting refers to dividing the chart into multiple parallel sub-charts, coloring refers to using color to classify data points or represent continuous values, and coordinate axes refer to mapping dimension values ​​to the X-axis or Y-axis of the chart as tick labels; data distribution type and numerical range are data feature descriptions used to determine the data characteristics of the visualization chart form that displays the data characteristics of the metric field; and visualization configuration set is a structured configuration file used to fully define and drive the visualization view.

[0042] In practice, the role of each dimension field in the chart (faceted, colored, or axis) is automatically determined based on its cardinality in the target data subset. This can be achieved as follows: First, analyze the number of unique values ​​(cardinality) of each dimension field in the target data subset. Then, automatically assign roles according to preset threshold rules: For dimensions with a cardinality of less than 5 (e.g., the "Gender" field with only two values, "Male" and "Female"), it is suitable to use faceted roles, dividing the data into several parallel sub-charts for comparison; for dimensions with a cardinality between 5 and 20 (e.g., the "Product Category" field with 5 to 20 different values), it is suitable to use colored roles, using different colors to distinguish different categories of data within the same chart. According to the series; for cardinality greater than 20 or for categorical or continuous dimensions that are themselves ordered (e.g., a "city" field or a "time" field with dozens of different values), it is suitable to directly map them to the X-axis or Y-axis of the chart as the primary categorical or metric coordinates; then, using the data distribution type and numerical range of each metric field in the target data subset, matching the chart type displaying the distribution characteristics can be achieved in the following way: analyze the data characteristics of each metric field in the target data subset, calculate the statistical characteristics of each metric field to determine its data distribution type (e.g., whether it is a time series, whether it contains multiple parts that need to be compared, whether it is geographical data); at the same time, the system obtains the numerical range (maximum) of each metric field. The process begins by matching each data feature (value, minimum value) to the most suitable chart type according to a predefined rule mapping table. For example, for a single time dimension and a single measure, with data points ordered by time, a line chart is matched to show the trend; for a categorical dimension and a measure, a bar chart is matched for comparison; for two measures, a scatter plot is matched to show the relationship; and for data consisting of multiple parts, a pie chart or stacked bar chart is matched. This step ensures that the selected chart format effectively reveals the core patterns in the data. Finally, based on the principle of visual perception effectiveness, appropriate visual encoding channels are assigned to the selected chart type, and the initial visual parameters are optimized based on the numerical range of each measure field to generate the target data sub-data. The visualization configuration set can be implemented as follows: After determining the chart type, based on the principle of visual perception effectiveness, assign appropriate visual encoding channels to each data dimension or metric in the chart. For example, for the most important comparison dimensions, they are usually assigned to the position or length channel (e.g., the height of a bar chart); for categorical information, they are assigned to the color hue channel; for secondary quantitative information, they may be assigned to the color brightness or point size channel. Then, based on the actual numerical range of each metric field, automatically optimize the initial visual parameters of the chart; for example, automatically set the start and end values ​​of the axes according to the maximum and minimum values ​​of the metrics to ensure that the data can make full use of the plotting area; and select an appropriate color gradient scheme according to the numerical distribution.Based on the number of data points, the system automatically adjusts graphic elements (e.g., bar width, point size) to ensure readability. All chart types, role assignments, code mappings, and parameter settings are combined into a visualization configuration set for the target data subset.

[0043] In step 104, the visualization configuration set is used to render and generate a multi-dimensional data visualization view, and the data visualization view is output and displayed in the graphical user interface.

[0044] In practice, the generated visualization configuration set drives the front-end visualization rendering engine to construct the view. The rendering engine first parses the chart type, visual encoding mapping relationship, and initial visual parameters in the configuration set, instantiates the corresponding chart components (e.g., bar chart, line chart components), and binds dimension field data to axis scales, faceted panels, or color legend generators according to role assignments. Simultaneously, it binds metric field data to the core visual attributes of the graphic elements (e.g., bar length, point position, color value) based on the encoding channel mapping. Next, the engine performs style configuration and scale calibration on the components based on the optimization parameters in the configuration (e.g., axis range, color palette, graphic size), performs geometric calculations and visual attribute filling from data to graphics, and completes the drawing of static graphics. Subsequently, the engine injects interactive logic (e.g., hover tooltips, click filters) into the graphic elements and embeds the rendered graphic container into the specified layout area of ​​the graphical user interface. Finally, the complete data visualization view, integrating data, visual encoding, and interactive capabilities, is displayed on the user's terminal screen as a graphical object that can be intuitively perceived and interactively analyzed.

[0045] Furthermore, in another aspect of this application, in some embodiments, this application provides an interactive, multi-dimensional data visualization sandbox for business data analysis, referencing... Figure 3 The figure is a schematic diagram of the structure of an interactive multi-dimensional data visualization display sandbox for business data analysis according to some embodiments of this application. The interactive multi-dimensional data visualization display sandbox for business data analysis includes: an acquisition module 201, a processing module 202, and an execution module 203, which are described below: The acquisition module 201 in this application is mainly used to acquire the original data records in the heterogeneous business data source, and then identify and standardize the dimension fields and measure fields in the original data records. Through each measure field, all dimension fields are aggregated in multiple layers, and then a multi-dimensional data cube of the heterogeneous business data source is constructed in memory. The processing module 202 in this application is used to perform semantic mapping on the state parameters in the interactive control operation when an interactive control operation is received, to obtain multiple visualization intent metadata, and then convert all the visualization intent metadata into a structured query instruction containing target dimension, target metric, expected chart type and data filtering conditions. It should be noted that the processing module 202 is also used to extract the target data subset for visualization rendering of heterogeneous business data sources from the multidimensional data cube based on the structured query instruction, and to perform visual optimization on the chart type, visual encoding channel and initial visual parameters of the visualization rendering by using the cardinality of the dimension field, the data distribution type and numerical range of the metric field in the target data subset, so as to obtain the visualization configuration set of the target data subset. The execution module 203 in this application is mainly used to render and generate a multi-dimensional data visualization view using the visualization configuration set, and to output and display the data visualization view in the graphical user interface.

[0046] The foregoing has detailed examples of interactive multi-dimensional data visualization sandboxes and methods for business data analysis provided in embodiments of this application. It is understood that the corresponding apparatus, in order to achieve the above functions, includes hardware structures and / or software modules corresponding to the execution of each function. Those skilled in the art should readily recognize that, based on the units and algorithm steps of the examples described in conjunction with the embodiments disclosed herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0047] In some embodiments, this application also provides a computer device, the computer device including a memory and a processor, the memory for storing a computer program, and the processor for calling and running the computer program from the memory, causing the computer device to perform the above-described interactive multi-dimensional data visualization method for business data analysis.

[0048] In some embodiments, reference Figure 4 The dashed lines in the figure indicate that the unit or module is optional. This figure is a structural schematic diagram of a computer device implementing an interactive multi-dimensional data visualization method for business data analysis according to an embodiment of this application. The interactive multi-dimensional data visualization method for business data analysis described in the above embodiments can be... Figure 4The computer device shown is used to implement this, and the computer device includes at least one processor 301, a memory 302 and at least one communication unit 305. The computer device may be a terminal device, a server or a chip.

[0049] Processor 301 can be a general-purpose processor or a special-purpose processor. For example, processor 301 can be a central processing unit (CPU), which can be used to control computer devices, execute software programs, and process data from software programs. The computer device may also include a communication unit 305 for inputting (receiving) and outputting (transmitting) signals.

[0050] For example, the computer device may be a chip, and the communication unit 305 may be the input and / or output circuit of the chip, or the communication unit 305 may be the communication interface of the chip, which may be a component of a terminal device, network device or other device.

[0051] For example, the computer device may be a terminal device or a server, and the communication unit 305 may be a transceiver of the terminal device or the server, or the communication unit 305 may be a transceiver circuit of the terminal device or the server.

[0052] The computer device may include one or more memories 302 storing a program 304. The program 304 can be executed by a processor 301 to generate instructions 303, causing the processor 301 to execute the method described in the above method embodiments according to the instructions 303. Optionally, the memory 302 may also store data (such as a target audit model). Optionally, the processor 301 may also read data stored in the memory 302, which may be stored at the same storage address as the program 304, or it may be stored at a different storage address than the program 304.

[0053] The processor 301 and memory 302 can be configured separately or integrated together, for example, integrated on the system on chip (SOC) of the terminal device.

[0054] It should be understood that each step of the above method embodiment can be completed by hardware logic circuits or software instructions in the processor 301. The processor 301 can be a CPU, a digital signal processor (DSP), an application-specific integrated circuit (ASIC), a field-programmable gate array (FPGA), or other programmable logic devices, such as discrete gate, transistor logic devices, or discrete hardware components.

[0055] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0056] For example, in some embodiments, this application also provides a computer-readable storage medium storing instructions or code that, when executed on a computer, cause the computer to implement the above-described interactive multi-dimensional data visualization method for business data analysis.

[0057] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0058] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. An interactive, multi-dimensional data visualization method for business data analysis, characterized in that, Includes the following steps: The system acquires raw data records from heterogeneous business data sources, identifies and standardizes the dimension fields and measure fields in the raw data records, performs multi-level aggregation on all dimension fields through each measure field, and then constructs a multi-dimensional data cube of the heterogeneous business data source in memory. When an interactive control operation is received, the state parameters in the interactive control operation are semantically mapped to obtain multiple visualization intent metadata. Then, all visualization intent metadata is converted into a structured query instruction containing target dimension, target metric, expected chart type and data filtering conditions. Based on the structured query instructions, a subset of target data for visualization rendering of heterogeneous business data sources is extracted from the multidimensional data cube. Visual optimization is performed on the chart type, visual encoding channel and initial visual parameters of the visualization rendering by using the cardinality of the dimension fields, the data distribution type and numerical range of the metric fields in the target data subset, to obtain a visualization configuration set for the target data subset. The visualization configuration set is used to render and generate a multi-dimensional data visualization view, which is then output and displayed in a graphical user interface.

2. The method as described in claim 1, characterized in that, Identifying and standardizing the dimension fields and measure fields in the original data records specifically includes: Read the structured metadata and unstructured sample data of the original data records, and identify candidate dimension fields and measure fields based on the preset field naming rule library and data pattern inference algorithm; Semantic disambiguation and category merging are performed on candidate dimension fields to unify the definitions of fields with the same meaning but different names, and a hierarchical relationship of dimensions is established based on the domain knowledge graph to obtain the dimension fields; The candidate measure fields are validated for data type and standardized for unit conversion. All numerical measures are unified to the benchmark unit of measurement and their aggregate function types are labeled to obtain the measure fields.

3. The method as described in claim 1, characterized in that, By aggregating all dimensional fields through various metrics, a multidimensional data cube of heterogeneous business data sources is constructed in memory. This specifically includes: Based on the hierarchical relationship of dimensions, multiple aggregated index tables with different granularities are pre-calculated for each dimension field. These aggregated index tables support roll-up and drill-down operations along the hierarchy. Based on the aggregation function type of the metric field, the corresponding metric summary value and statistical feature value are pre-calculated and cached in memory for each dimension combination granularity; By establishing a star-shaped logical model between dimensions and measures through various aggregated index tables, all measure summary values, and statistical feature values, a multidimensional data cube that supports online processing operations is formed.

4. The method as described in claim 1, characterized in that, Semantic mapping is performed on the state parameters in the operation of the interactive control to obtain multiple visual intent metadata, specifically including: Capture the various state parameters during the operation of the interactive control; Each state parameter is mapped to the corresponding dimension field, metric field, and filter condition expression in the multidimensional data cube to form the original intent parameters that the machine can understand. Based on the combination logic of controls and business rules, infer the user's potential analytical intent and supplement the generation of implicit analytical dimensions and metrics; The original intent parameters, implicit analysis dimensions, and analysis metrics are fused to form multiple visualized intent metadata.

5. The method as described in claim 1, characterized in that, Convert all visualization intent metadata into structured query instructions that include target dimensions, target metrics, desired chart types, and data filtering conditions. Specifically, this includes: Conflict detection and priority arbitration are performed on the metadata of each visualization intent. When there are mutually exclusive or redundant intents, they are selected or merged according to preset rules. The intent parameters in the arbitrated visual intent metadata are converted into standard query statement fragments, clarifying the target dimension, target metric, and data filtering conditions of the query. Based on user historical preferences and global visualization settings, multiple recommended expected chart types are specified for the query results, and then the target dimension, target metric, expected chart type and data filtering conditions are encapsulated into structured query instructions.

6. The method as described in claim 1, characterized in that, The specific subset of target data for visualization rendering of heterogeneous business data sources extracted from the multidimensional data cube based on the structured query instructions includes: Apply the filter conditions in the structured query command to the corresponding dimensions of the multidimensional data cube to determine the data range slice for the query; Based on the target dimension and target metric in the structured query instruction, locate the corresponding pre-aggregated data block in the multidimensional data cube and trigger real-time aggregation calculation; The results of the data range slicing and real-time aggregation calculations are organized into a result set according to the format and order required by the query, and contextual metadata is attached to form a target data subset for visualization rendering.

7. The method as described in claim 1, characterized in that, Visual optimization of the chart type, visual encoding channel, and initial visual parameters for visualization rendering is performed based on the cardinality of the dimension fields, the data distribution type of the metric fields, and the numerical range of the target data subset. This results in a visualization configuration set for the target data subset that specifically includes: The role of each dimension field in the chart—faceted, colored, or an axis—is automatically determined based on the cardinality of the data in the target subset. Using the data distribution type and numerical range of each metric field in the target data subset, match the chart type that displays the distribution characteristics; Based on the principle of visual perception effectiveness, appropriate visual encoding channels are assigned to the selected chart type, and the initial visual parameters are optimized based on the numerical range of each metric field to generate a visualization configuration set for the target data subset.

8. An interactive, multi-dimensional data visualization sandbox for business data analysis, characterized in that, include: The acquisition module is used to acquire the original data records in the heterogeneous business data source, and then identify and standardize the dimension fields and measure fields in the original data records. Through each measure field, all dimension fields are aggregated in multiple layers, and then a multi-dimensional data cube of the heterogeneous business data source is constructed in memory. The processing module is used to perform semantic mapping on the state parameters in the interactive control operation when an interactive control operation is received, to obtain multiple visualization intent metadata, and then convert all the visualization intent metadata into a structured query instruction containing target dimension, target metric, expected chart type and data filtering conditions. The processing module is also used to extract a subset of target data for visualization rendering of heterogeneous business data sources from the multidimensional data cube based on the structured query instruction, and to perform visual optimization on the chart type, visual encoding channel and initial visual parameters of the visualization rendering by using the cardinality of the dimension field, the data distribution type and numerical range of the metric field in the target data subset, so as to obtain a visualization configuration set of the target data subset. The execution module is used to render and generate a multi-dimensional data visualization view using the visualization configuration set, and output and display the data visualization view in the graphical user interface.

9. A computer device, characterized in that, The computer device includes a memory and a processor, the memory for storing computer programs, and the processor for calling and running the computer programs from the memory, causing the computer device to perform the interactive multi-dimensional data visualization method for business data analysis as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions or code that, when executed on a computer, cause the computer to implement the interactive multi-dimensional data visualization method for business data analysis as described in any one of claims 1 to 7.

Images