Financial data collaborative integration and visualization processing method and system based on structured data

Abstract

The invention discloses a financial data collaborative integration and visualization processing method and system based on structured data, and relates to the field of data integration and visualization, and the method comprises the steps: S1, collecting structured original financial data from a plurality of heterogeneous financial data sources, and adding a data gene identifier containing a source, a timestamp and a version to each data item; s2, cleaning and standardizing the original financial data, and generating a standard data unit with a multi-dimensional quality score label; s3, inputting the standard data unit into a dynamic integration engine for processing. The method is used for solving the problems that in the prior art, a data integration rule is static and rigid and is difficult to adapt to service changes, a visualization result lacks automatic explanation of an internal story line and an abnormal root cause of data, and the data integration efficiency is high. And cooperative operation and underlying integration logic are disjointed, and the process lacks credible audit tracking.

Description

Technical Field

[0001] This invention relates to the field of data integration and visualization, specifically to a method and system for collaborative integration and visualization of financial data based on structured data. Background Technology

[0002] In the fields of corporate financial management and decision support, efficiently and accurately integrating multi-source, heterogeneous financial data and transforming it into intuitive, collaborative insights is crucial for improving the operational efficiency and risk management capabilities of modern enterprises. With the increasing number of enterprise information systems, financial data is typically scattered across ERP, CRM, banking systems, and various business databases. These data differ significantly in structure, format, update frequency, and semantics, leading to severe data silos. How to achieve the automatic collaborative integration of this structured data and generate visual results that support in-depth analysis and team collaboration has become a pressing technical challenge.

[0003] Currently, there are several existing technologies and systems for integrating and visualizing financial data. Common practices include: first, extracting data from different data sources using ETL (Extract, Transform, Load) tools or customized interfaces, and performing basic cleaning and format standardization; second, using predefined data mapping rules and association logic to perform integrated modeling in a data warehouse or data lake; and finally, based on the integrated data, creating static or interactive dashboards and reports using business intelligence (BI) tools for display. Regarding collaboration, existing solutions largely rely on shared databases or cloud-based document and view sharing, allowing different users to view the same data or engage in simple annotation and communication.

[0004] However, the aforementioned existing technical solutions have several drawbacks. First, the data integration process typically relies on pre-defined, static rules and mapping models, making it difficult to adapt to dynamic changes in business logic and the evolution of the data sources themselves. It lacks self-learning and adaptive capabilities, resulting in poor flexibility and high maintenance costs when facing new data sources or complex relationships. Second, existing visualization solutions mostly present the integration results in direct charts, with logically fragmented charts lacking automatic discovery and coherent narration of the underlying storylines and root causes of anomalies. Users still need to rely on their own experience for interpretation, leading to insufficient depth in decision support. Finally, existing collaboration mechanisms mostly remain at the level of "sharing" data or views. Collaborative operations (such as hypothesis analysis and data correction) themselves fail to effectively feedback and optimize the data integration logic, and the collaboration process lacks tamper-proof audit trails, making it difficult to meet stringent financial compliance and data traceability requirements. Therefore, there is an urgent need for a collaborative processing solution that can achieve intelligent adaptive integration, generate narrative visualizations, and form a closed-loop learning and reliable audit. Summary of the Invention

[0005] Based on this, the purpose of this invention is to provide a method and system for collaborative integration and visualization of financial data based on structured data, in order to solve the technical problems in the prior art, such as static and rigid data integration rules that are difficult to adapt to business changes, lack of automatic interpretation of the data’s internal storyline and root causes of anomalies in visualization results, and disconnect between collaborative operation and underlying integration logic, and lack of reliable audit trails in the process.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a method and system for collaborative integration and visualization of financial data based on structured data, comprising the following steps: S1: collecting structured raw financial data from multiple heterogeneous financial data sources, and attaching a data gene identifier containing source, timestamp, and version to each data item; S2: cleaning and standardizing the raw financial data to generate standard data units with multi-dimensional quality scoring labels; S3: inputting the standard data units into a dynamic integration engine for processing; wherein, the operation of the dynamic integration engine is based on an evolvable integration knowledge network, which uses data entities as nodes and financial logical relationships and historical collaborative consensus between entities as edges, with the weight of the edges determined by the frequency of use of the association, the success rate of conflict resolution, and cross-application... S4: Based on the integrated data set output by the integrated knowledge network and the current user role, the narrative visualization generator is invoked. The generator dynamically arranges the type, order, and focus of the visualization charts according to the preset financial analysis framework and the real-time detected data anomaly patterns or key inflection points, and generates explanatory narrative text to form an interactive data narrative sequence. S5: The data narrative sequence is displayed on the collaborative interface, and user collaborative instructions are received. In response to the instructions, the system not only updates the data and views, but also feeds the instructions, their context, and execution results back to the integrated knowledge network as a new training sample to adjust the weights of the relevant relationships, and generates a collaborative decision block with a digital signature, which is recorded in the local audit chain.

[0007] The present invention is further configured as a financial data collaborative integration and visualization processing system based on structured data, characterized in that it is used to implement the method as described in any one of claims 1 to 6, the system comprising: a data gene management module for collecting data from heterogeneous data sources and attaching data gene identifiers; a quality assessment and standardization module for cleaning data and generating standard data units with quality score labels; an evolvable integrated knowledge network module for storing and dynamically adjusting an evolvable integrated knowledge network composed of data entities and financial logical relationships, and driving the data integration process; a narrative visualization engine for dynamically generating a data narrative sequence containing visual charts and narrative text based on integration results and anomaly detection; and an intelligent collaboration and audit chain module for managing collaborative interactions, processing user collaborative instructions, recording collaborative decision blocks to the audit chain, and transmitting collaborative feedback back to the evolvable integrated knowledge network module.

[0008] In summary, the present invention has the following main beneficial effects: This invention achieves intelligent and adaptive financial data integration rules by constructing a self-evolving integrated knowledge network, thereby improving the model's flexibility and accuracy. It utilizes a narrative visualization generator to automatically transform data anomalies and key inflection points into a coherent data narrative sequence, significantly enhancing the depth and interpretability of visual insights. It establishes an intelligent closed loop from collaborative operation feedback to integration logic optimization, and uses a blockchain-style audit chain to ensure credible traceability and compliant auditing throughout the entire process. Finally, through an effect feedback learning cycle, it links the actual effectiveness of business decisions with the quality of data integration, driving the system to continuously improve itself, forming a complete value loop from data integration and collaborative analysis to decision verification. Attached Figure Description

[0009] Figure 1 This is a flowchart illustrating the overall method of the present invention; Figure 2 This is a flowchart illustrating the operation of the evolvable and integrated knowledge network module of the present invention. Figure 3 This is a flowchart illustrating the operation of the intelligent collaboration and audit chain module of the present invention. Detailed Implementation

[0010] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings, so that those skilled in the art can implement the present invention. It should be understood that the specific embodiments described herein are only for explaining the present invention and are not intended to limit the scope of protection of the present invention.

[0011] This invention provides a method for collaborative integration and visualization of financial data based on structured data. like Figure 1The diagram shown is a flowchart of a preferred embodiment of the method of the present invention. This method constructs a complete closed-loop system from data collection, intelligent integration, narrative presentation to collaborative auditing, and its core lies in "evolvability", "narrative presentation" and "traceability".

[0012] Specifically, the following steps are included: Data collection and genetic labeling This step is the source of data processing, aiming to achieve unified access and native traceability of heterogeneous data. The specific implementation is as follows: First, the system configuration manager allows administrators to define the various heterogeneous financial data sources that need to be accessed. These data sources are physically dispersed and logically heterogeneous, typically including: ERP system databases deployed on-premises or in the cloud, CRM system databases, independent budget management databases, reconciliation file interfaces provided by bank systems through security gateways, and electronic invoice data interfaces from tax platforms, etc. Each data source needs to have its connection parameters, authentication methods, and data extraction scope configured.

[0013] Secondly, the system has a built-in data acquisition scheduler. This scheduler supports two modes: timed polling mode and event-triggered mode. Timed polling mode automatically initiates data extraction tasks at preset time intervals; event-triggered mode listens for system messages or file changes in the data source and immediately triggers acquisition when new data is detected. During acquisition, the data source is accessed through the corresponding adapter, predefined query statements are executed, or API calls are made to obtain structured raw financial data, such as accounting entries, sales order tables, bank transaction records, and invoice details.

[0014] The key innovation lies in the fact that for every key data item in each collected record, the system generates and binds an immutable data identifier in real time. This identifier is globally unique within the system, and its data structure includes a composite key. Simultaneously, the metadata of this identifier also records the timestamp of successful data collection by the system, as well as a version number maintained by the system. For example, the identifier for a sales revenue record from an SAP ERP system might be recorded as a composite structure containing the source system, table name, primary key, field name, and version information for the "Sales Amount Including Tax" field. This identifier, like the data's "DNA," will be permanently retained and used throughout all subsequent processing stages, providing the foundation for accurate end-to-end traceability.

[0015] Data cleaning, standardization, and multi-dimensional quality label generation This step aims to improve the quality and consistency of the raw data and assign a quantifiable credibility rating to each data unit. It is a multi-stage processing flow: Stage 1: Standardized Cleaning. The cleaning rule base pre-constructs various cleaning rules for financial data. The system executes these rules sequentially: structure mapping and field matching, format unification, handling of invalid and missing values, and verification of basic business rules. Stage 2: Intelligent Anomaly Detection and Isolation. The system embeds an anomaly detection model. This model is trained as follows: the system continuously collects manually confirmed dirty and normal data cases from historical cleaning processes to form a training sample set. The model learns the distribution patterns of normal data and the characteristics of common erroneous data. During runtime, the model scans the standardized cleaned data and outputs an anomaly probability score. Simultaneously, the system uses a set of strong business rules as hard constraints. Finally, combining the model's output score and rule violations, the system classifies the data into three categories: "credible," "pending review," and "abnormal." The latter two types of data will be isolated in a special area and highlighted, awaiting manual review and adjudication by users with the appropriate permissions, rather than being processed automatically by the system. This ensures data security and avoids automated misjudgments.

[0016] Phase Three: Standardized Encapsulation and Quality Tag Generation. Based on a global financial data meta-model, the system encapsulates the cleaned and inspected data into standard data units. The meta-model defines the standard attribute sets of core entities in the financial domain. The encapsulation process involves filling the raw data from various sources into the corresponding standard data structures according to the meta-model's definition.

[0017] Simultaneously, the system automatically calculates and assigns a multi-dimensional quality score label to each standard data unit. This label is a structured evaluation object, including but not limited to the following dimensions and their scores: source authority score, processing confidence score, timeliness score, and consistency score. The scores for each dimension can be combined into an overall quality score through weighted averaging or other algorithms. This label will serve as an important reference for subsequent integration, visualization, and decision-making.

[0018] Dynamic weighted integration based on an evolvable knowledge network This step is the core of data logic fusion. Its goal is to weave discrete, standard data units into an interconnected network based on business logic, and this network is capable of self-learning and evolution. First, the system initializes or maintains an evolvable, integrated knowledge network. Internally, this network is represented as an attribute graph: nodes represent specific financial data entity instances; edges represent relationships between nodes; and the weight of each edge is a dynamically changing value. The initial value can be defined by rules, but the core lies in its adaptive adjustment mechanism.

[0019] The dynamic integration engine utilizes this network for data integration. For example, when a "product line profit and loss statement" needs to be generated, the engine locates all relevant nodes in the network, traverses the edges connecting these nodes, and determines how to merge, associate, or calculate the data carried by these nodes based on the edge type and current weight, and finally performs calculations to generate an integrated data view.

[0020] The network's evolvability (adaptability) is reflected in the dynamic adjustment of edge weights. Specific mechanisms include: positive reinforcement based on usage frequency and result stability: the system records the number of times each edge is integrated and referenced in calculations. When a financial logic edge is frequently used to generate various reports within a preset statistical period, and the calculation results are never subsequently corrected or questioned by the user, the system automatically increases the weight of that edge. Negative weakening and reassessment based on conflict and falsification: when data integrated based on a certain edge is marked as suspicious by the user through the system's "questioning" function, or when a later-imported official audit report proves the integration result to be incorrect, the system immediately reduces the weight of that edge and triggers a reassessment process.

[0021] The introduction of virtual collaborator nodes provides multi-perspective checks and balances: Several virtual collaborator nodes are pre-configured in the network. These are not real users, but logical entities endowed with the mindset of specific financial roles. When the data processed by the dynamic integration engine involves high-risk areas, or when the conclusions drawn from different integration paths differ significantly, the relevant virtual collaborator nodes are activated. They simulate their corresponding role's perspective, automatically generating a brief comparative analysis report or a clear risk warning, which is pushed to the real users using the system in the form of "virtual opinions" as additional reference for decision-making, achieving intelligent risk warning and multi-faceted review.

[0022] Dynamic generation and interactive display of narrative visualization This step aims to transform cold, aggregated data into logical, insightful, and interactive "data stories." The system, based on the currently logged-in user's role and permissions, their business focus, and the current data set output by the integrated knowledge network, activates the narrative visualization generator. Its workflow is as follows: Storyline Detection and Triggering: The generator has a built-in storyline detection unit that continuously monitors a set of preset key financial indicators. It calculates in real-time the deviation of these indicators' current values ​​from their historical baselines, budget targets, or industry benchmarks. When the deviation of any indicator exceeds a preset threshold based on business importance, the system automatically determines that a "story" has begun and triggers the root cause analysis narrative chain generation task for that indicator.

[0023] Link Tracing and Root Cause Locator: Once the story is triggered, the generator immediately interacts with the evolving and integrated knowledge network. It traces back along the higher-weighted edges connected to the anomaly node in the network, automatically locating the upstream related nodes that caused the anomaly and recording the complete link path. Visual Narrative Dynamic Arrangement: The visual narrative arrangement unit takes over. Based on the traced causal chain and link path, it intelligently selects chart types from the visualization component library, arranges the order of appearance and focal positions of these charts, and controls the narrative pace. Explanatory Narrative Text Generation: While arranging the charts, the system drives a text generator. Based on a pre-set financial analysis narrative template, this generator fills the template with key nodes, relationships, and data confidence levels from the tracing path, generating a coherent and natural explanatory narrative text. Generation of Interactive Data Narrative Sequence: Finally, the system merges the arranged chart sequence with the narrative text in terms of timeline and logic, forming an interactive data narrative sequence. The user interface no longer presents a static dashboard, but a dynamic analysis report that can be "played" or "stepped through." Users can click on highlighted words in the narrative text to directly access the corresponding chart; they can also perform interactive operations such as drill-down and filtering on the chart panel, while the narrative text updates dynamically. The entire sequence provides complete analytical guidance from problem discovery to root cause identification.

[0024] Intelligent collaborative interaction, feedback learning, and solidification of the audit chain This step handles multi-user collaboration scenarios, ensuring all operations are traceable and auditable, while forming a learning loop. On the collaborative interface, multiple users can simultaneously view or manipulate the same data narrative sequence. When any user initiates a collaborative operation command, the system executes as follows: Command execution and real-time synchronization: The system immediately executes the command in the background, updating the affected integrated dataset. Due to the correlation between data, this update is automatically propagated to relevant data nodes through the integrated knowledge network. The visualization view is then refreshed accordingly, with all online users seeing synchronized updates to the data and charts.

[0025] Feedback learning and network evolution: The system encapsulates the instruction and its complete context into a structured training sample. This sample is fed back into an evolvable, integrated knowledge network. For example, if the user frequently manually corrects the results generated by an automatically integrated rule, the network will gradually decrease the weight of the edge corresponding to that rule; if the user consistently confirms a certain association, its weight will be increased. This achieves an intelligent closed loop of continuous learning from human-computer interaction.

[0026] Generating an immutable collaborative decision-making block: Simultaneously, the system generates a collaborative decision-making block with a digital signature. The digital signature is generated by cryptographically hashing the block content using the operator's identity key, ensuring the integrity of the block and the operator's non-repudiation. This block content is stored in a structured manner and must include: the hash value of the operation instruction, a list of all affected data gene identifiers, a comparison of key data snapshots before and after the operation, a list of all user identifiers participating in this collaborative discussion or decision, and a precise timestamp.

[0027] Audit Chain Construction and Traceability: The system links newly generated collaborative decision-making blocks to all previous blocks, forming a chronologically extending audit chain that can only be appended to, not modified. This chain can be technically implemented using hash pointer linked lists, blockchain, or other tamper-proof data structures. Any subsequent audit needs can be precisely located on this audit chain using data identifiers as clues, pinpointing all operation blocks affecting the data item. This allows for a complete reproduction of its detailed history from data collection to its current state, including every state change, who modified it, and under what context, meeting the highest level of financial compliance and audit traceability requirements.

[0028] Feedback-based learning loop (used to continuously optimize the best steps of the system) To further enhance the system's support value for actual business decisions and enable it to learn from business results, this method preferably includes an automatically running background effect feedback learning loop. This loop operates independently of daily operations at a low frequency, and its specific process is as follows: Decision Snapshot Encapsulation: When the system detects that a user has made an action marked as a "critical decision" based on the provided data and analysis view, the system automatically creates a decision data package. This data package encapsulates a complete snapshot of information at the moment of the decision. Actual Business Result Collection: After a preset evaluation period, the system automatically or manually imports quantitative indicators of the actual business results corresponding to the decision from relevant business operation systems via an interface. Attribution Analysis and Contribution Evaluation: The effect feedback learning module initiates an attribution analysis model. This model takes the various data characteristics and network weight states in the decision data package as input, and the actual business result indicators as the target, analyzing which data factors and which integrated logical relationships have a significant impact on the final business results, and quantifying the degree of their impact.

[0029] System Self-Optimization: Based on the results of attribution analysis, the system automatically generates optimization instructions. For example, if the analysis shows that a certain association contributes highly to the prediction result, but the weight of that association in the current integrated knowledge network is low, the system will suggest or automatically increase the weight of that association edge. Conversely, for association rules with extremely low contribution or even a counterproductive effect, it will suggest reducing their weight or initiating a re-evaluation. Simultaneously, if data from certain data sources is found to be unreliable in multiple decision analyses, the system will automatically lower the base score of "source authority" in the quality label for that data source. Through this cycle, the system no longer optimizes solely based on internal operations, but instead prioritizes "generating positive business results" as its ultimate goal, driving the entire data processing and integration logic to continuously evolve in the direction most valuable to the business.

[0030] Financial data collaborative integration and visualization processing system based on structured data Corresponding to the above method embodiments, the system embodiments of the present invention are implemented in the form of software modules and deployed on a server cluster or cloud platform. Each module exchanges data and collaborates through defined application programming interfaces or message queues, collectively forming a complete system.

[0031] The system includes: Data Gene Management Module: This module is the system's data entry point. It includes a connector pool, a data acquisition task scheduler, a data cache, and a gene identifier generator. Its functions include executing data acquisition and gene identifier attachment steps, establishing and managing connections with all external heterogeneous data sources, triggering acquisition tasks according to plans or events, receiving raw data streams, generating and binding unique data gene identifiers for each data item in real time, and then pushing the identifiable raw data to downstream modules.

[0032] Quality Assessment and Standardization Module: This module acts as the gatekeeper of data quality. It incorporates a data cleaning rule engine, an anomaly detection model server, a financial data meta-model library, and a quality scoring calculator. Its function is to perform data cleaning, standardization, and multi-dimensional quality label generation steps. It receives labeled raw data from upstream sources, sequentially performs rule cleaning, model anomaly detection and isolation, then encapsulates it into standard data units according to the meta-model, and calculates and generates multi-dimensional quality scoring labels by integrating various information. Finally, it outputs clean, standardized data units with quality labels.

[0033] Evolvable Integral Knowledge Network Module: This module is the brain and core knowledge base of the system. Physically, it is typically implemented based on a graph database, including graph storage services, dynamic weight adjustment logic processors, virtual collaborator simulators, and network snapshot services. Its function is to execute dynamic weighted integration steps based on the evolving integral knowledge network. It stores and maintains the continuously evolving integral knowledge network. It receives standard data units, incorporating them into the network as new nodes or updating existing nodes; it responds to queries from the integration engine, providing data associations and computational paths; it dynamically adjusts edge weights according to preset mechanisms; and it manages the logic of virtual collaborator nodes, generating reference opinions under specific conditions. This module is the concentrated embodiment of the system's intelligence.

[0034] Narrative Visualization Engine: This module is the core of the system's presentation and interaction. It includes a storyline monitor, a visualization component library, a narrative template library, a text generation service, and a sequence synthesizer. Its function is to execute the dynamic generation and interactive display steps of narrative visualization. It obtains data from the integration network module, discovers anomalies and story clues through the storyline detection unit, calls components and templates through the visual narrative arrangement unit, dynamically arranges charts and generates narrative text, and finally synthesizes and outputs an interactive data narrative sequence to the display front end.

[0035] The Intelligent Collaboration and Audit Chain Module serves as the hub for system collaboration and trust assurance. It comprises a collaboration session manager, instruction parser, audit block builder, digital signature service, and audit chain storage service. Its functions include executing intelligent collaborative interactions, feedback learning, and audit chain solidification steps; managing multi-user collaborative sessions; receiving and parsing user collaborative instructions; coordinating data updates and view synchronization; constructing digitally signed collaborative decision blocks; and securely appending them to the immutable audit chain. Simultaneously, it is responsible for sending the context information of collaborative operations as feedback signals back to the evolvable and integrable knowledge network module.

[0036] Effect Feedback Learning Module: This module is the optimizer for achieving a closed loop of business value in the system. It includes a decision snapshot capturer, a business result collection interface, an attribution analysis model server, and an optimization suggestion generator. Its function is to execute the effect feedback learning loop steps. It periodically captures decision data packets of key decisions, collects quantitative indicators of actual business results in later stages, runs the attribution analysis model to evaluate the contribution of data and rules, and generates optimization suggestions for knowledge network weights and data quality labels based on this, driving the system to self-improve.

[0037] In summary, this invention provides an innovative method and system for collaborative integration and visualization of financial data based on structured data, aiming to solve the core problems of rigid data integration rules, lack of in-depth interpretability in visualization, untraceable collaborative processes, and missing feedback loops in existing technologies. This method ensures traceability by attaching "genetic identifiers" to data and constructs an "evolvable integration knowledge network" that can dynamically adjust weights based on usage feedback and business results, achieving adaptive optimization of the integration logic. The system further introduces a "narrative visualization generator," automatically transforming data anomalies into coherent analytical stories and interaction sequences, significantly enhancing decision-making insight. All collaborative operations are recorded in an immutable "audit chain," ensuring reliable traceability throughout the process. Finally, through an "effect feedback learning loop," the system feeds back the actual effectiveness of business decisions to the data processing logic, forming a complete intelligent closed loop from data to business value. This invention achieves a fundamental transformation in data processing, from static to dynamic, from isolated to collaborative, and from presentation to narrative. Although embodiments of the present invention have been shown and described, these specific embodiments are merely explanations of the invention and are not intended to limit it. The specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. After reading this specification, those skilled in the art may make modifications, substitutions, and variations to the embodiments as needed without departing from the principles and spirit of the invention, but such modifications, substitutions, and variations are protected by patent law as long as they are within the scope of the claims of the present invention.

Claims

1. A method for collaborative integration and visualization of financial data based on structured data, characterized in that, Includes the following steps: S1: Collect structured raw financial data from multiple heterogeneous financial data sources and attach a data gene identifier containing the source, timestamp and version to each data item; S2: Clean and standardize the original financial data to generate standard data units with multi-dimensional quality scoring labels; S3: Input the standard data unit into the dynamic integration engine for processing; wherein, the operation of the dynamic integration engine is based on an evolvable integration knowledge network, which uses data entities as nodes and financial logical relationships and historical collaborative consensus between entities as edges. The weight of the edges is adaptively strengthened or weakened according to the frequency of use of the association, the success rate of conflict resolution, and the cross-user verification results. S4: Based on the integrated data set output by the integrated knowledge network and the current user role, call the narrative visualization generator; Based on a preset financial analysis framework and real-time detected data anomaly patterns or key inflection points, the generator dynamically arranges the types, order, and focus of visualization charts and generates explanatory narrative text, which together form an interactive data narrative sequence. S5: Display the data narrative sequence on the collaborative interface and receive user collaborative instructions; in response to the instructions, the system not only updates the data and view, but also feeds the instructions, their context, and execution results back to the integrated knowledge network as a new training sample to adjust the weights of the relevant relationships, and generates a collaborative decision block with a digital signature, which is recorded in the local audit chain.

2. The method according to claim 1, characterized in that, In step S3, the adaptability of the evolvable integrated knowledge network is specifically reflected in: When a certain relationship is frequently referenced by multiple users in collaborative operations within a preset period and does not trigger subsequent corrections, its weight is automatically increased. When data integrated based on a certain relationship is flagged by the user through a preset challenge mechanism or is subsequently disproven by imported external audit results, the downgrade of the relationship weight and the re-evaluation process of the association rule are triggered. The integrated knowledge network contains virtual collaborator nodes, which are pre-trained to simulate the perspective of specific financial roles. When high-risk or high-uncertainty situations arise during the integration process, alternative integration solutions or risk warnings are automatically generated and pushed to relevant users as reference views.

3. The method according to claim 1, characterized in that, In step S4, the operation of the narrative visualization generator includes: Monitor the deviation of key financial indicators from historical baselines, budgets, or industry benchmarks. When the deviation exceeds a threshold, automatically trigger the generation of a root cause analysis narrative chain. The root cause analysis narrative chain automatically locates and visualizes abnormal changes in upstream related data items by associating them with the integrated knowledge network, and presents the association path and confidence level in the form of narrative text.

4. The method according to claim 1, characterized in that, In step S5, the records of the collaborative decision-making block include: The content of the collaborative decision block includes at least the operation instruction hash, the set of affected data gene identifiers, a comparison of data snapshots before and after the operation, and the user identifiers and timestamps of the users participating in the collaboration. The audit chain is appended to the collaborative decision-making block and allows the complete evolution history and decision context of any data item to be traced through the data gene identifier.

5. The method according to claim 1, characterized in that, The process after step S5 also includes: S6: The system background runs an effect feedback learning loop, periodically collecting quantitative indicators of actual business results for key business decisions supported by the integrated data; The quantitative indicators of actual business results are correlated with the state of the data at the time of decision-making and the weight configuration of the integrated knowledge network at that time; Based on the analysis results, global optimization suggestions are generated for the weights of relevant relationships in the integrated knowledge network, or corrections are triggered for the quality score labels of specific data sources.

6. The method according to claim 5, characterized in that, In step S6, the effect feedback learning loop specifically includes: Establish a decision data package, which encapsulates the data set used at a specific business decision moment, the state snapshot of the integrated knowledge network, and the corresponding visual narrative sequence; After the predetermined evaluation period, obtain the actual financial performance indicators of the decision from the business system; By using attribution analysis models, the impact of each data item and its integration relationship in the decision data package on the final outcome is evaluated, and the integrated knowledge network is then trained through reinforcement learning.

7. A financial data collaborative integration and visualization processing system based on structured data, characterized in that, The system for implementing the method as described in any one of claims 1 to 6 comprises: The data gene management module is used to collect data from heterogeneous data sources and attach data gene identifiers; The quality assessment and standardization module is used to clean data and generate standard data units with quality score labels. An evolvable and integrable knowledge network module is used to store and dynamically adjust an evolvable and integrable knowledge network composed of data entities and financial logical relationships, and to drive the data integration process. A narrative visualization engine for dynamically generating data narrative sequences that include visual charts and narrative text based on integrated results and anomaly detection. The intelligent collaboration and audit chain module is used to manage collaborative interactions, process user collaborative instructions, record collaborative decision blocks to the audit chain, and send collaborative feedback back to the evolvable integrated knowledge network module.

8. The system according to claim 7, characterized in that, The system also includes: The effect feedback learning module is used to collect actual effect data of business decisions, correlate the data status and integrated knowledge network status at the time of decision-making through attribution analysis, and form optimization feedback on the output of the evolvable integrated knowledge network module and the quality assessment and standardization module.

9. The system according to claim 7, characterized in that, The evolvable and integrable knowledge network module has at least one virtual collaborator node pre-configured, and the virtual collaborator node is configured as follows: Risk signals during the monitoring data integration process; When a risk signal reaches a threshold, a comparative analysis report or risk questioning view is generated based on its pre-trained perspective, and inserted into the current user's collaborative interface through the intelligent collaboration and audit chain module.

10. The system according to claim 7, characterized in that, The narrative visualization engine includes: The storyline detection unit is used to continuously analyze and integrate the data to identify significant patterns or turning points that conform to the preset narrative template. The visual narrative orchestration unit selects and sorts charts from a library of visualization components based on the detected storylines, and drives a text generator to produce coherent explanatory narratives, merging charts and text into a data narrative sequence that can be played step by step or interacted with.

Images