Computer-based multi-platform informatization construction system and method

Abstract

The invention provides a multi-platform informatization construction system and method based on a computer, and solves the problems that existing multi-platform data synchronization is not timely, formats are not compatible and the like. The system comprises a real-time data capture module, and a data change event is generated in real time through database transaction log monitoring; the intelligent routing decision module optimizes a transmission path in combination with rules and machine learning; the data conversion adaptation module realizes automatic adaptation of a format and an interface by utilizing dynamic template matching; the conflict detection and resolution module guarantees data consistency based on version control and priority judgment; the data synchronization execution module transmits data according to an optimized path, and the monitoring management module monitors the whole process in real time. The method comprises the steps of data capture, routing decision, format conversion, synchronous execution and monitoring feedback. According to the scheme, the data synchronization efficiency and accuracy are improved, the multi-platform integration operation and maintenance cost is reduced, and the method is suitable for multi-scene informatization construction of enterprises, government affairs and the like.

Description

Technical Field

[0001] The present invention relates to computer informationization related fields, and in particular to a computer-based multi-platform information construction system and method. Background Art

[0002] With the rapid development of information technology, enterprises and organizations often build multiple business platforms with different functions to meet diverse business needs, such as sales management platforms, customer relationship management platforms, and financial management platforms. However, existing multi-platform information systems face serious data synchronization issues. For example, at a large enterprise, the sales management platform records customer order information, while the customer relationship management platform stores basic customer information and communication records. When sales personnel update a customer's order status on the sales management platform, the customer relationship management platform is unable to promptly access this information. As a result, customer service personnel still communicate with customers based on the old order status, resulting in a poor customer experience. This is because existing multi-platform information systems lack efficient data synchronization mechanisms. Data updates between platforms often rely on manual triggering or scheduled batch synchronization. Manual triggering is inefficient and prone to omissions, while scheduled batch synchronization can lead to data inconsistencies within time intervals. This makes it difficult to meet the needs of business scenarios with high real-time requirements, seriously affecting business collaboration efficiency and decision-making accuracy. Summary of the Invention

[0003] The purpose of the present invention is to provide a computer-based multi-platform information construction system and method to solve the problems raised in the above background technology.

[0004] To achieve the above objectives, the present invention provides the following technical solutions: a computer-based multi-platform information construction system, comprising a real-time data capture module, an intelligent routing decision module, a data conversion and adaptation module, a conflict detection and resolution module, a data synchronization execution module, and a monitoring and management module. The real-time data capture module is used to monitor data changes on each business platform in real time, and the real-time data capture module sets a data change monitoring point in the transaction log of each platform database. Once a data insert, update, or delete operation is detected, the relevant data change information is immediately captured, and a data change event containing key information such as the operation type, data object, and timestamp is generated.

[0005] The intelligent routing decision module uses a routing decision algorithm based on a combination of rules and machine learning based on the attributes of data change events and the characteristics of the target platform to pre-set basic rules for data flow. At the same time, it uses machine learning to analyze historical data synchronization paths and efficiency, dynamically optimizes routing decisions, and selects the optimal data transmission path.

[0006] The data conversion adaptation module adopts a dynamic template matching data conversion algorithm for data formats and interface specifications of different platforms, and constructs a rich data format template library. When receiving a data change event, the data conversion adaptation module automatically matches corresponding templates for data format conversion and interface adaptation according to data formats of a source platform and a target platform, so that data is accurately transmitted between different platforms.

[0007] The conflict detection and resolution module applies a conflict detection and resolution algorithm based on version control and priority determination to assign a version number to each data object. When multiple platforms operate on the same data object, the conflict detection and resolution algorithm automatically resolves data conflicts by comparing the version number and a preset operation priority, thereby ensuring data consistency.

[0008] The data synchronization execution module synchronizes data to a target platform according to the path determined by the intelligent routing decision module and the data processed by the data conversion adaptation module, and records the synchronization result.

[0009] The monitoring management module monitors the entire data synchronization process in real time, monitors and warns the running state, data synchronization progress, and abnormal conditions of each module, and is used for management and maintenance by an administrator.

[0010] Preferably, the intelligent routing decision module specifically implements the following logic:

[0011] Data change event receiving and analysis: The intelligent routing decision module receives a data change event from the real-time data capture module. The data change event includes, but is not limited to, operation type, data object, and timestamp key attribute information. These information is analyzed to extract key elements for routing decision, and the data change event is recorded as E={e1,e2,…,e n}, where e i represents the i-th attribute of the event E.

[0012] Basic rule matching: A basic rule library R={r1,r2,…,r m} of data flow direction is preset. Each rule r j is represented as a condition-result pair, that is, r j : IF C j THEN P j ; where C j is a condition expression composed of data change event attributes; P j is a corresponding transmission path decision result; the analyzed data change event attributes are matched with the rules in the rule library. If there is a rule that satisfies the condition C j , the corresponding transmission path P j is directly obtained, and routing decision is completed.

[0013] Machine learning prediction assisted decision making: if no matching rule is found in the basic rule base, a machine learning model is started to assist decision making; a path prediction model trained by historical data is used, the attributes of the data change event E are taken as the model input, and the input vector is F=[f1, f2, …, f n ], where f i corresponds to the attribute e i of the data change event E after feature engineering processing, the model outputs the probability distribution P=[p1, p2, … p k ] of each transmission path through internal calculation and weight parameters, where p s represents the probability of data transmission to the s-th path, and

[0014] Path optimization and adjustment: after determining the preliminary transmission path, the path is optimized in combination with the path selection, transmission time, and success rate data in the historical data synchronization process; the comprehensive evaluation index S of the current path is calculated, and the calculation formula is S=α×T+β×Q+γ×Z, where T is the average transmission time of the path in the historical data, Q is the transmission success rate, where Q is the number of successful transmissions / the total number of transmissions of the path in the past 24 hours, Z is the complexity of the path, α, β, γ are weight coefficients, which are set according to business requirements, and α+β+γ=1; if there is another path with a better comprehensive evaluation index, the transmission path is adjusted, and the optimal path is selected as the final data transmission path.

[0015] Route information generation and sending: after determining the final data transmission path, the intelligent routing decision module generates route information containing source platform information, target platform information, transmission path, etc., and sends the route information to the data conversion and adaptation module together with the data change event, so as to perform data format conversion and transmission subsequently.

[0016] Preferably, the path prediction model trained based on historical data has the following specific implementation steps:

[0017] Step 1, data preprocessing and feature engineering:

[0018] Features are extracted from historical data synchronization logs to build a sample set D={(X1, y1), (X2, y2), …, (X n , y n )}, where each input sample X i is a time series consisting of t time steps: X i =[x i1 , x i2 , …, x it ], and each time step feature vector x it contains:

[0019] Data change event attributes: operation type o t ; data object type d t ; timestamp feature ts t ; source platform ID st and target platform ID candidate set T t = {t t1 , t t2, …}; historical synchronization performance indicators: average delay lat t , success rate sr t and throughput thr t ; target variable y i is the best transmission path ID, where the best transmission path ID is selected from the candidate set T t ;

[0020] Step 2, improve the LSTM model architecture design:

[0021] Basic LSTM layer: process time series input, capture long-term dependencies: h t = LSTM(x t , h t―1 ), where h t is the hidden state, containing the comprehensive features of the data change event at this time step (such as operation type, data object, transmission delay, etc.), x t is the current input, and h t―1 is the hidden state at the previous time step;

[0022] Attention mechanism layer: weight the LSTM output, highlight key time steps: e t = v T tanh(Wh t +b), where a t is the attention weight, v, W, b are trainable parameters, where v is the "attention preference vector", the weight matrix W is the feature weighting of h t , and b is used to adjust the offset of the activation function to avoid all e t falling into the saturation region of the activation function; e t is the "attention score" calculated for each time step t;

[0023] Context vector:

[0024] Path probability output layer: p = softmax(W p c + b p ), where p is the probability distribution of each candidate path, W p , b p are output layer parameters;

[0025] Step 3, model training optimization

[0026] Loss function design: according to the characteristics of multi-platform data synchronization, the path success rate and delay index are integrated into the loss function, specifically, the loss function adopts category cross entropy: Where y ij is the one-hot encoding of the true path label of sample i, p ij is the predicted probability, N is the number of training samples, and K is the number of different categories of data flow paths, wherein the optimizer adopts Adam algorithm;

[0027] Step 4, training process with rule constraints:

[0028] Introduce rule consistency loss term: L rule = λ∑ i∈S ||p i ―s i || 2 , wherein S is the rule matching sample set, s i is the path probability distribution specified by the rule, and λ is the balance coefficient;

[0029] The total loss function is: L total =L+L rule ;

[0030] Step 5, prediction and decision process:

[0031] Input the current data change event feature sequence X current ; the path prediction model outputs the path probability distribution P = [p1, p2, … p k ]; select the path with the highest probability as the prediction result path pred = argmaxx j p j ; wherein x j is the feature value of the input vector.

[0032] Preferably, the data conversion adaptation module has the following working logic:

[0033] Data reception and information extraction: when the real-time data capture module monitors the data change of the business platform, and the intelligent routing decision module determines the transmission path, the data conversion adaptation module receives the data change event containing operation type, data object, and timestamp key information; at the same time, the source platform and target platform identifiers are extracted from the routing information to clarify the source and destination of the data, providing basic information for subsequent operations;

[0034] Template library matching: Based on the extracted source and target platform data format information, a search and match is performed in a pre-built data format template library. The template library stores data conversion templates for different platform combinations. Each template defines the mapping relationship between the source format and the target format. The best matching template is selected by calculating the feature similarity between the source data format and the template.

[0035] Data format conversion and interface adaptation: Utilize the matched template to convert the data format in the data change event. According to the mapping rules defined in the template, convert the source data structure, field name, and data type into a format recognized by the target platform. During the format conversion, perform interface adaptation processing on the data according to the interface specifications of the target platform, including but not limited to adjusting the data transmission protocol and request parameter format to ensure that the data can be smoothly connected to the interface of the target platform.

[0036] Processed data output and transmission: After completing data format conversion and interface adaptation, the processed data is sent to the data synchronization execution module; the data synchronization execution module synchronizes the adapted data to the target platform according to the path determined by the intelligent routing decision module, realizing accurate data transmission between multiple platforms and providing reliable data support for subsequent business processing.

[0037] Preferably, the specific implementation steps of the conflict detection and resolution module are as follows:

[0038] a. Version mark before data synchronization:

[0039] Version number assignment: When the real-time data capture module detects data changes, the conflict detection and resolution module generates a unique version number V for each data object in the format of node ID + timestamp + sequence number; metadata attachment: The version number is attached to the data change event as metadata and transmitted together with the data object;

[0040] b. Conflict pre-detection:

[0041] Version comparison: Before data synchronization is executed, the module queries the current version number V of the data object on the target platform current ; Conflict prediction: If the version number of the data to be synchronized is V new Less than or equal to V current , it is determined to be outdated data and synchronization is terminated; if V new >V current , then enter the synchronization process and continue to monitor conflicts;

[0042] c. Concurrent operation conflict detection:

[0043] Multi-source write monitoring: When multiple platforms simultaneously initiate change requests for the same data object, the module captures concurrent operations through a distributed lock mechanism; Version number collision detection: Compare the version numbers of each change request. If there are version numbers that are the same but the contents are different, it is determined to be a conflict;

[0044] d. Priority-based conflict resolution:

[0045] Operation type priority rules: Pre-set operation priority order. When a conflict occurs, the high-priority operation is executed first; Business rule priority: Define exclusive rules for specific data objects; Timestamp priority: When the operation types are the same, compare the timestamps in the version numbers, and prefer the latest operation;

[0046] e. Version merging and data integration:

[0047] Mergeable change processing: For non-mutually exclusive changes, the module automatically merges the change content to generate a new version number V merge ; Data integration algorithm: Use graph database technology to analyze the relationship between data to ensure logical consistency of the merged data;

[0048] f. Conflict log recording and analysis:

[0049] Log storage: Record detailed information of all conflict events, including conflict data, version number, and resolution method; Then, based on historical conflict data, optimize priority rules.

[0050] Preferably, a computer-based multi-platform information construction method includes the following steps:

[0051] Step S1, the real-time data capture module uses an event-driven data capture algorithm to monitor the transaction logs of each business platform database in real time. When a data change is detected, a data change event is generated and sent to the intelligent routing decision module;

[0052] Step S2, the intelligent routing decision module receives the data change event, determines the data transmission path based on a routing decision algorithm that combines rules and machine learning, and sends the data change event and routing information to the data conversion adaptation module:

[0053] Step S3, the data conversion adaptation module uses a dynamic template matching data conversion algorithm to perform format conversion and interface adaptation on the data based on the received source platform and target platform information, and then sends the processed data to the data synchronization execution module;

[0054] Step S4, the data synchronization execution module synchronizes data to the target platform according to the path determined by the intelligent routing decision module, and the conflict detection and resolution module detects and handles data conflicts by using a conflict detection and resolution algorithm based on version control and priority determination;

[0055] Step S5, the data synchronization execution module records the data synchronization result and feeds back the synchronization result to the monitoring management module, and the monitoring management module monitors and manages the entire data synchronization process in real time.

[0056] Compared with the prior art, the beneficial effects of the present application are: the intelligent routing decision module of the present application combines the rule base and the improved LSTM model to realize intelligent optimization of the transmission path, effectively solving the problems of chaotic multi-platform data transmission path and low efficiency; by comprehensively evaluating the indicators, the transmission time, success rate and complexity of the path are calculated in real time, and high-delay paths are automatically avoided;

[0057] The dynamic template matching algorithm of the data conversion adaptation module of the present application supports real-time conversion of several data formats such as XML, JSON and CSV, and the template library covers most of the common business fields of enterprises, solving the industry pain point of incompatible data formats of different manufacturer platforms; and for API interface specifications (such as RESTful and SOAP) of different platforms, the module automatically completes the request parameter format adjustment and protocol conversion (such as HTTP and TCP / IP), greatly reducing the system integration cost.

[0058] The conflict detection and resolution module realizes the unique identification of data changes in a distributed environment through the version number mechanism of "node ID + timestamp + sequence number". In the multi-platform concurrent operation scene, the consistency of key data such as financial data and inventory information is ensured; the preset operation type priority (such as delete > update > insert) and business rule priority (such as financial data priority) are combined with the latest timestamp judgment to realize intelligent adjudication of conflicts. For example, when the sales platform and the inventory platform update the price of a commodity at the same time, the system processes according to the "latest update time" priority principle, avoids price data confusion, and ensures the accuracy of the business process. BRIEF DESCRIPTION OF DRAWINGS

[0059] Fig. 1 It is a schematic diagram of the overall structure of the present application;

[0060] Fig. 2 It is a specific implementation flowchart of the intelligent routing decision module of the present application. DETAILED DESCRIPTION

[0061] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of the present invention.

[0062] Example 1

[0063] See also Figs. 1-2 The present invention provides a technical solution: a computer-based multi-platform information construction system, comprising a real-time data capture module, an intelligent routing decision module, a data conversion and adaptation module, a conflict detection and resolution module, a data synchronization execution module, and a monitoring and management module. The real-time data capture module is used to monitor data changes on each business platform in real time, and the real-time data capture module sets a data change monitoring point in the transaction log of each platform database. Once a data insert, update, or delete operation is detected, the relevant data change information is immediately captured, and a data change event containing key information such as the operation type, data object, and timestamp is generated.

[0064] The intelligent routing decision module uses a routing decision algorithm that combines rules and machine learning based on the attributes of data change events and the characteristics of the target platform. It pre-defines basic rules for data flow and analyzes historical data synchronization paths and efficiency through machine learning to dynamically optimize routing decisions and select the optimal data transmission path. The specific implementation logic is as follows:

[0065] Data change event reception and analysis: The intelligent routing decision module receives data change events from the real-time data capture module. Data change events include but are not limited to operation types (insert, update, delete), data objects (such as orders, customer information), and timestamp key attribute information. This information is analyzed to extract key elements for routing decisions. The data change event is recorded as E = {e1, e2, ..., e n}, where e i Represents the i-th attribute of event E, for example, e1 is the operation type, e2 is the data object, etc.

[0066] Basic rule matching: pre-set basic rule base R of data flow direction = {r1, r2, ..., r m}, each rule r j Represented as a condition-result pair, i.e. r j :IF C j THEN P j ; among them C j It is a conditional expression composed of data change event attributes; P jcorresponding transmission path decision result; the parsed data change event attributes are matched with the rules in the rule library, if there is a rule that meets the condition C j , the corresponding transmission path P j is directly obtained, and the routing decision is completed; for example, when e1 (operation type) is "insert" and e2 (data object) is "order", the condition of a certain rule is met; P j is the corresponding transmission path decision result, such as transmitting the data to the financial management platform. The parsed data change event attributes are matched with the rules in the rule library, if there is a rule that meets the condition C j , the corresponding transmission path P j is directly obtained, and the routing decision is completed.

[0067] Machine learning prediction assisted decision: if no matching rule is found in the basic rule library, a machine learning model is started for assisted decision; a path prediction model trained with historical data is used, the attributes of the data change event E are taken as the model input, and the input vector is F = [f1, f2, …, f n ], where f i corresponds to the value of the attribute e i of the data change event E after feature engineering processing; the model outputs the probability distribution P = [p1, p2, … p k ] of each transmission path through internal calculation and weight parameters, where p s represents the probability of data transmission to the s-th path, and

[0068] Path optimization and adjustment: after determining the preliminary transmission path, the path is optimized in combination with the path selection, transmission time, and success rate data in the historical data synchronization process; the comprehensive evaluation index S of the current path is calculated, and the calculation formula is S = α × T + β × Q + γ × Z, where T is the average transmission time of the path in the historical data, Q is the transmission success rate, Q is the number of successful transmissions / the total number of transmissions in the past 24 hours, Z is the complexity of the path (such as the number of intermediate nodes passed, etc.), α, β, γ are weight coefficients, which are set according to business requirements, and α + β + γ = 1; if there is another path with a better comprehensive evaluation index, the transmission path is adjusted, and the optimal path is selected as the final data transmission path.

[0069] Routing information generation and sending: after determining the final data transmission path, the intelligent routing decision module generates routing information containing source platform information, target platform information, transmission path, etc., and sends the routing information together with the data change event to the data conversion and adaptation module for subsequent data format conversion and transmission.

[0070] The specific implementation steps of the path prediction model trained based on historical data are as follows:

[0071] Step 1: Data preprocessing and feature engineering:

[0072] In multi-platform data synchronization scenarios, historical synchronization logs contain a wealth of decision-making basis. For example, when an order change event occurs on a sales platform, the model must consider not only the attributes of the event itself (such as the operation type and data object), but also time characteristics (such as 10 a.m. on weekdays is usually peak business hours) and historical synchronization performance (such as a path experiencing a significant increase in latency during peak hours). By extracting these multi-dimensional features and constructing a sample set, the model can learn complex patterns such as "order creation events should prioritize dedicated network paths on weekday mornings." Specifically, features are extracted from historical data synchronization logs to construct a sample set D = {(X1, y1), (X2, y2), ..., (X n ,y n )}, where each input sample X i Is a time series consisting of t time steps: X i =[x i1 ,x i2 ,…,x it ], where the time step t is dynamically adjusted according to the time interval of business data changes, for example, t = 5 minutes for high-frequency trading data and t = 1 hour for low-frequency data; the feature vector x of each time step is it Include:

[0073] Data change event attributes: Operation type o t ;Data object type d t ; Timestamp feature ts t ; Source platform IDst and target platform ID candidate set T t ={t t1 ,t t2 ,…}; Historical synchronization performance indicators: average delay lat t 、success rate sr t and throughput thr t Target variable y i is the best transmission path ID, where the best transmission path ID is selected from the candidate set T t Select from;

[0074] Step 2: Improve LSTM model architecture design:

[0075] Basic LSTM layer: processes time series input and captures long-term dependencies: h t =LSTM(x t ,h t―1 ), where h t is the hidden state, xt is the current input, h t―1 is the hidden state at the previous time step; for example, consecutive order update operations can cause the target platform cache to be invalidated, affecting subsequent synchronization performance. LSTM can capture this long-term dependency through memory cells, avoiding the short-term memory defect of traditional models.

[0076] Attention mechanism layer: weights the LSTM output, highlighting key time steps: e t = v T tanh(Wh t +b), where a t is the attention weight, v, W, and b are trainable parameters, where v is the "attention preference vector", the weight matrix W is to feature-weight h t , and b is used to adjust the offset of the activation function to avoid all e t falling into the saturation zone of the activation function; e t is the "attention score" calculated for each time step t; the importance of features at different time steps to routing decisions is different. For example, when synchronizing customer sensitive information, the weight of security indicators should be significantly higher than that of ordinary business data. The attention mechanism automatically calculates the weights, allowing the model to focus on key features, such as when processing payment data, focusing more on features related to encrypted channels.

[0077] Context vector:

[0078] Path probability output layer: p = softmax(W p c+b p ), where p is the probability distribution of each candidate path, W p , b p are output layer parameters; that is, the routing decision is converted into a probability distribution, quantifying uncertainty. For example, when there are multiple optional paths, the model not only gives the optimal path, but also provides the confidence of each path, providing a basis for subsequent rule fusion.

[0079] Step 3, model training and optimization

[0080] Loss function design: considering the characteristics of multi-platform data synchronization, the success rate and delay indicators of the path are integrated into the loss function. Specifically, the loss function uses categorical cross-entropy: where y ij is the one-hot encoding of the true path label for sample i, and p ijis the predicted probability, N is the number of training samples, K is the number of different categories of data flow paths, K is the dynamically scalable path category number, the number of model output layer neurons is automatically adjusted with K, and the model quickly adapts to new paths through transfer learning, wherein the optimizer adopts the Adam algorithm; In view of the characteristics of multi-platform data synchronization, the success rate and delay indicators of the path are integrated into the loss function;

[0081] Step 4, training process with rule constraints:

[0082] Introduce a rule consistency loss term: L rule =λ∑ i∈S ||p i ―s i || 2 , wherein S is a rule matching sample set, s i is a rule specified path probability distribution, specifically, s i is a one-hot vector with the jth dimension being 1, λ is a balance coefficient, and λ dynamically adjusts the weight of machine learning and rules; In the early stage of new platform access, the rule weight is higher; With the accumulation of historical data, the model weight gradually increases, realizing the smooth transition from "expert dominance" to "data-driven"; In some scenarios, expert experience or safety policy requires that certain types of data must be transmitted through a specific path. For example, financial data must pass through an encrypted channel. Through the rule consistency loss term, it is ensured that the model prediction result does not violate these hard constraints; The total loss function is: L total =L+L rule ;

[0083] Step 5, prediction and decision process:

[0084] Input the current data change event feature sequence X current ; the path prediction model outputs the path probability distribution P = [p1, p2, … p k ]; select the path with the highest probability as the prediction result path pred =argmaxx j p j ; wherein x j is the feature value of the input vector.

[0085] The data conversion adaptation module adopts a dynamic template matching data conversion algorithm to construct a rich data format template library. When receiving a data change event, it automatically matches the corresponding template for data format conversion and interface adaptation according to the data format of the source platform and the target platform, so that the data can be accurately transmitted between different platforms; The specific working logic is as follows:

[0086] Data reception and information extraction: When the real-time data capture module detects data changes in the business platform and the intelligent routing decision module determines the transmission path, the data conversion and adaptation module receives data change events containing operation type, data object, and timestamp key information. At the same time, the source platform and target platform identifiers are extracted from the routing information, clarifying the data source and destination, and providing basic information for subsequent operations;

[0087] Template library matching: Based on the extracted source platform and target platform data format information, search and match in the pre-constructed data format template library. The template library stores data conversion templates for different platform combinations, each template defines the mapping relationship between the source format and the target format. By calculating the feature similarity between the source data format and the template, the most matching template is selected. For example, if the source platform is a sales platform that stores order data in XML format, and the target platform is a stock management platform that uses JSON format, the module will find the corresponding template in the template library that converts XML to JSON and is suitable for order data.

[0088] Data format conversion and interface adaptation: Use the matched template to convert the data in the data change event. According to the mapping rules defined in the template, convert the structure, field name, and data type of the source data into the format recognized by the target platform. At the same time of format conversion, perform interface adaptation processing on the data according to the interface specifications of the target platform, including but not limited to adjusting the transmission protocol, request parameter format of the data, to ensure that the data can be smoothly connected to the interface of the target platform;

[0089] Processed data output and transmission: After completing data format conversion and interface adaptation, send the processed data to the data synchronization execution module. The data synchronization execution module synchronizes the adapted data to the target platform according to the path determined by the intelligent routing decision module, realizing accurate transmission of data between multiple platforms and providing reliable data support for subsequent business processing.

[0090] The conflict detection and resolution module uses a conflict detection and resolution algorithm based on version control and priority determination to assign a version number to each data object during data synchronization. When multiple platforms operate on the same data object, the conflict is automatically resolved by comparing the version number and the preset operation priority, ensuring data consistency. The specific implementation steps are as follows:

[0091] a. Version marking before data synchronization:

[0092] Version number allocation: When the real-time data capture module detects data changes, the conflict detection and resolution module generates a unique version number V for each data object in the format of node ID + timestamp + serial number; metadata attachment: The version number is attached to the data change event as metadata and transmitted together with the data object; for example, when the sales platform updates the order status, the order data will carry the latest version number.

[0093] b. Conflict pre-detection (before synchronization):

[0094] Version comparison: Before data synchronization is executed, the module queries the current version number V of the data object on the target platform current ; Conflict prediction: If the version number of the data to be synchronized is V new Less than or equal to V current , it is determined to be outdated data and synchronization is terminated; if V new >V current , then enter the synchronization process and continue to monitor conflicts;

[0095] c. Concurrent operation conflict detection (synchronization):

[0096] Multi-source write monitoring: When multiple platforms simultaneously initiate change requests for the same data object (such as customer information), the module captures concurrent operations through a distributed lock mechanism; version number collision detection: Compares the version numbers of each change request. If the version numbers are the same but the content is different, it is determined to be a conflict; for example, the inventory management platform and the sales platform simultaneously update the inventory of a certain product.

[0097] d. Priority-based conflict resolution:

[0098] Operation type priority rules: Preset operation priority order (e.g., delete > update > insert. For financial data, delete operations require manual review and have a lower priority than verified update operations, making the rules more flexible). When a conflict occurs, the higher-priority operation is executed first. For example, if a delete and update request are both received simultaneously, the delete operation takes precedence. Business rule priority: Define dedicated rules for specific data objects. For example, changes to financial data take precedence over changes to general business data to ensure the accuracy of cash flow. Timestamp priority: When operations of the same type are performed, the timestamp in the version number is compared, and the latest operation is prioritized.

[0099] e. Version merging and data integration:

[0100] Mergeable change processing: For non-mutually exclusive changes (such as one platform updating the customer phone number and another updating the email address), the module automatically merges the changes and generates a new version number V merge Data integration algorithm: Utilizes graph database technology to analyze relationships between data and ensure logical consistency of the merged data. For example, when merging orders and inventory changes, the inventory deduction logic is verified.

[0101] f. Conflict log recording and analysis:

[0102] Log storage: record detailed information of all conflict events, including conflict data, version number, resolution method; then optimize priority rules based on historical conflict data. For example, by analyzing that a certain type of order conflicts frequently, adjust its processing priority.

[0103] The data synchronization execution module is used to synchronize data to the target platform according to the path determined by the intelligent routing decision module and the data processed by the data conversion and adaptation module, and record the synchronization result;

[0104] The monitoring management module monitors the entire data synchronization process in real time, monitors and warns the running state, data synchronization progress and abnormal situation of each module, and is used for administrators to manage and maintain.

[0105] Embodiment 2

[0106] A computer-based multi-platform information construction method, comprising the following steps:

[0107] Step S1, the real-time data capture module uses an event-driven data capture algorithm to monitor the transaction log of each business platform database in real time, and when a data change is detected, generates a data change event and sends it to the intelligent routing decision module;

[0108] Step S2, the intelligent routing decision module receives the data change event, determines the transmission path of the data according to the routing decision algorithm based on the combination of rules and machine learning, and sends the data change event and routing information to the data conversion and adaptation module:

[0109] Step S3, the data conversion and adaptation module uses a dynamic template matching data conversion algorithm to perform format conversion and interface adaptation on the data according to the received source platform and target platform information, and then sends the processed data to the data synchronization execution module;

[0110] Step S4, the data synchronization execution module synchronizes the data to the target platform according to the path determined by the intelligent routing decision module, and the conflict detection and resolution module uses a conflict detection and resolution algorithm based on version control and priority determination to detect and handle data conflicts;

[0111] Step S5, the data synchronization execution module records the data synchronization result and feeds back the synchronization result to the monitoring management module, and the monitoring management module monitors and manages the entire data synchronization process in real time.

[0112] For example, deploy real-time data capture module, intelligent routing decision module, data conversion adaptation module, conflict detection and resolution module, data synchronization execution module and monitoring management module on enterprise data center server. According to the data flow and business demand of each platform, allocate server resources reasonably to ensure stable operation of the system. At the same time, configure data capture interface in the database of each business platform, so that the real-time data capture module can obtain data change information;

[0113] When the customer places an order on the sales platform, the sales platform database generates an order data insertion operation. The real-time data capture module immediately captures the data change event, generates a data change event containing the order number, product information, customer information, order time, etc., and sends it to the intelligent routing decision module; After the intelligent routing decision module receives the data change event, it determines to synchronize the order data to the inventory management platform and the financial management platform according to the pre-set rules and the optimized routing decision algorithm of machine learning, and generates the corresponding routing information and sends it to the data conversion adaptation module; According to the data format difference of sales platform and inventory management platform, financial management platform, data conversion adaptation module uses dynamic template matching data conversion algorithm to convert order data into formats that can be recognized by two target platforms. For example, convert the order data format of the sales platform into the inventory update data format required by the inventory management platform, and the order payment data format required by the financial management platform; The data synchronization execution module synchronizes the converted data to the inventory management platform and the financial management platform according to the path determined by the intelligent routing decision module. In the synchronization process, the conflict detection and resolution module monitors whether there is a data conflict in real time. Assuming that during the synchronization process, the inventory management platform simultaneously performs inventory counting operation to update the inventory data, and the conflict detection and resolution module automatically resolves the conflict by comparing the version number and the pre-set priority (the order data update priority is higher than the inventory counting data update), to ensure the consistency of the inventory data; The data synchronization execution module records the data synchronization result and feeds back the synchronization state information to the monitoring management module. The monitoring management module displays the data synchronization progress in real time, and if an abnormal situation occurs, such as data transmission failure, conflict cannot be solved, etc., it will send an early warning notice to the administrator for processing.

[0114] The application discloses a computer-based multi-platform informationization construction system and method, aiming to solve the problems of non-timely data synchronization, incompatible format, difficult conflict resolution, etc. in existing multi-platform informationization construction. The system includes real-time data capture module, intelligent routing decision module, data conversion adaptation module, conflict detection and resolution module, data synchronization execution module and monitoring management module.

[0115] The real-time data capture module generates data change events in real time by setting a listening point in the database transaction log; the intelligent routing decision module dynamically optimizes the transmission path using an algorithm based on the combination of rules and machine learning; the data conversion and adaptation module realizes the automatic adaptation of different platform data formats and interfaces with the help of a dynamic template matching algorithm; the conflict detection and resolution module guarantees data consistency through version control and priority determination; the data synchronization execution module transmits data according to the optimized path and records the results; and the monitoring management module monitors the whole process in real time.

[0116] The method includes the steps of data capture, routing decision, format conversion, synchronization execution and monitoring feedback. The system and method significantly improve the data synchronization efficiency and accuracy, reduce the multi-platform integration and operation and maintenance cost, and are suitable for information construction in multiple scenarios such as enterprises and government affairs.

[0117] Although embodiments of the present application have been shown and described, it is to be understood that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the present application, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A computer-based multi-platform information construction system, characterized in that: It includes a real-time data capture module, an intelligent routing decision module, a data conversion and adaptation module, a conflict detection and resolution module, a data synchronization execution module, and a monitoring and management module. The real-time data capture module is used to monitor data changes on each business platform in real time. The real-time data capture module sets a data change monitoring point in the transaction log of each platform database. Once a data insert, update, or delete operation is detected, the relevant data change information is immediately captured and a data change event containing key information such as the operation type, data object, and timestamp is generated. The intelligent routing decision module uses a routing decision algorithm based on a combination of rules and machine learning based on the attributes of data change events and the characteristics of the target platform to pre-set basic rules for data flow. At the same time, it uses machine learning to analyze historical data synchronization paths and efficiency, dynamically optimizes routing decisions, and selects the optimal data transmission path. The data conversion and adaptation module uses a data conversion algorithm with dynamic template matching to adapt to the data formats and interface specifications of different platforms, and builds a rich data format template library. When a data change event is received, it automatically matches the corresponding template to perform data format conversion and interface adaptation according to the data formats of the source and target platforms, so that data can be accurately transmitted between different platforms. During the data synchronization process, the conflict detection and resolution module uses a conflict detection and resolution algorithm based on version control and priority determination to assign a version number to each data object. When multiple platforms operate on the same data object, data conflicts are automatically resolved by comparing the version number and the preset operation priority to ensure data consistency. The data synchronization execution module is used to synchronize data to the target platform according to the path determined by the intelligent routing decision module and the data processed by the data conversion and adaptation module, and record the synchronization results; The monitoring management module monitors the entire data synchronization process in real time, monitors and warns of the operating status, data synchronization progress and abnormal situations of each module, and is used for management and maintenance by administrators.

2. A computer-based multi-platform information construction system according to claim 1, characterized in that: The specific implementation logic of the intelligent routing decision module is as follows: Data change event reception and analysis: The intelligent routing decision module receives data change events from the real-time data capture module. Data change events include but are not limited to key attribute information such as operation type, data object, and timestamp. This information is parsed to extract the key elements for routing decisions, and the data change event is recorded as E = {e1, e2, ..., e n }, where e i Represents the i-th attribute of event E; Basic rule matching: pre-set basic rule base R of data flow direction = {r1, r2, ..., r m }, each rule r j Represented as a condition-result pair, i.e. r j :IFC j THEN P j ; among them C j It is a conditional expression composed of data change event attributes; P j The corresponding transmission path decision result; match the parsed data change event attributes with the rules in the rule base, if there is one that meets the condition C j The corresponding transmission path P is directly obtained. j , complete the routing decision; Machine learning prediction and decision-making assistance: If no matching rules are found in the basic rule base, the path prediction model is started to assist in decision-making, and the output of the path prediction model needs to be cross-validated with the constraints in the rule base. Specifically, the path prediction model trained with historical data is used, and the attributes of the data change event E are used as the model input. The input vector is denoted as F = [f1, f2, ..., f n ], where f i Attribute e corresponding to data change event E i After the feature engineering processing, the model outputs the probability distribution P of each transmission path through internal calculation and weight parameters. k ], where p s represents the probability that data is transmitted to the sth path, and Path optimization and adjustment: After determining the preliminary transmission path, the path is optimized based on the path selection, transmission time, and success rate data from the historical data synchronization process. The comprehensive evaluation index S of the current path is calculated using the formula S = α × T + β × Q + γ × Z, where T is the average transmission time of the path in historical data, Q is the transmission success rate, where Q is the number of successful transmissions / total transmissions for the path in the past 24 hours, and Z is the complexity of the path. α, β, and γ are weight coefficients set according to business needs, with α + β + γ = 1. If another path has better comprehensive evaluation indicators, the transmission path is adjusted and the optimal path is selected as the final data transmission path. Routing information generation and sending: After determining the final data transmission path, the intelligent routing decision module generates routing information containing source platform information, target platform information, transmission path, etc., and sends the routing information together with the data change event to the data conversion adapter module for subsequent data format conversion and transmission.

3. The computer-based multi-platform information construction system according to claim 2, characterized in that: The specific implementation steps of the path prediction model trained based on historical data are as follows: Step 1: Data preprocessing and feature engineering: Extract features from historical data synchronization logs and construct a sample set D = {(X1, y1), (X2, y2), ..., (X n ,y n )}, where each input sample X i Is a time series consisting of t time steps: X i =[x i1 ,x i2 ,…,x it ], each time step feature vector x it Include: Data change event attributes: Operation type o t ;Data object type d t ; Timestamp feature ts t ; Source platform IDst and target platform ID candidate set T t ={t t1 ,t t2 ,…}; Historical synchronization performance indicators: average delay lat t 、success rate sr t and throughput thr t Target variable y i is the best transmission path ID, where the best transmission path ID is selected from the candidate set T t Select from; Step 2: Improve LSTM model architecture design: Basic LSTM layer: processes time series input and captures long-term dependencies: h t =LSTM(x t ,h t―1 ), where h t is the hidden state, x t is the current input, h t―1 It is the hidden state at the previous moment; Attention mechanism layer: weights the LSTM output to highlight key time steps: e t =v T tanh(Wh t +b), where α t is the attention weight, v, W, b are trainable parameters, where v is the "attention preference vector" and the weight matrix W is the weight matrix for h t Perform feature weighting, b is used to adjust the offset of the activation function to avoid all e t Falling into the saturation region of the activation function; e t To calculate the "attention score" for each time step t; Context vector: Path probability output layer: p = softmax(W p c+b p ), where p is the probability distribution of each candidate path, W p 、b p is the output layer parameter; Step 3: Model training optimization Loss function design: In view of the characteristics of multi-platform data synchronization, the path success rate and latency indicators are integrated into the loss function. Specifically, the loss function adopts category cross entropy: where y ij is the one-hot encoding of the true path label of sample i, p ij is the prediction probability, N is the number of training samples, K is the number of different categories of data flow paths, and the optimizer uses the Adam algorithm; Step 4: Training process of fusion rule constraints: Introduce rule consistency loss term: L rule =λ∑ i∈S ||p i ―s i || 2 , where S is the rule matching sample set, s i is the path probability distribution specified by the rule, λ is the balance coefficient; The total loss function is: L total =L+L rule ; Step 5: Forecasting and decision-making process: Input the current data change event feature sequence C current ; The path prediction model outputs the path probability distribution P = [p1, p2, ... p k ]; Select the path with the highest probability as the prediction result path pred =argmaxx j p j ; where x j is the eigenvalue of the input vector.

4. The computer-based multi-platform information construction system according to claim 1, characterized in that: The specific working logic of the data conversion adaptation module is as follows: Data reception and information extraction: When the real-time data capture module detects data changes on the business platform and the intelligent routing decision module determines the transmission path, the data conversion and adaptation module receives the data change event containing key information such as the operation type, data object, and timestamp. Simultaneously, it extracts the identifiers of the source and target platforms from the routing information to clarify the source and destination of the data, providing basic information for subsequent operations. Template library matching: Based on the extracted source and target platform data format information, a search and match is performed in a pre-built data format template library. The template library stores data conversion templates for different platform combinations. Each template defines the mapping relationship between the source format and the target format. The best matching template is selected by calculating the feature similarity between the source data format and the template. Data format conversion and interface adaptation: Use the matched template to convert the data format in the data change event; According to the mapping rules defined in the template, the structure, field name, and data type of the source data are converted into a format recognized by the target platform. During the format conversion, the data is adapted to the interface specifications of the target platform, including but not limited to adjusting the data transmission protocol and request parameter format to ensure that the data can be smoothly connected to the interface of the target platform. Processed data output and transmission: After completing data format conversion and interface adaptation, the processed data is sent to the data synchronization execution module; The data synchronization execution module synchronizes the adapted data to the target platform according to the path determined by the intelligent routing decision module, realizes the accurate transmission of data between multiple platforms, and provides reliable data support for subsequent business processing.

5. The computer-based multi-platform information construction system according to claim 1, characterized in that: The specific implementation steps of the conflict detection and resolution module are as follows: a. Version mark before data synchronization: Version number allocation: When the real-time data capture module detects data changes, the conflict detection and resolution module generates a unique version number V for each data object in the format of node ID + timestamp + sequence number; Metadata attachment: Attach the version number as metadata to the data change event and transmit it together with the data object; b. Conflict pre-detection: Version comparison: Before data synchronization is executed, the module queries the current version number V of the data object on the target platform current ; Conflict prediction: If the version number of the data to be synchronized is V new Less than or equal to V current , it is determined to be outdated data and synchronization is terminated; if V new >V current , then enter the synchronization process and continue to monitor conflicts; c. Concurrent operation conflict detection: Multi-source write monitoring: When multiple platforms initiate change requests for the same data object simultaneously, the module captures concurrent operations through a distributed lock mechanism. Version number collision detection: Compares the version numbers of each change request. If the version numbers are the same but the content is different, it is considered a conflict. d. Priority-based conflict resolution: Operation type priority rule: preset operation priority order. When a conflict occurs, the higher priority operation will be executed first. Business rule priority: define exclusive rules for specific data objects. Timestamp priority: when the operation type is the same, compare the timestamp in the version number and give priority to the latest operation. e. Version merging and data integration: Mergeable change processing: For non-mutually exclusive changes, the module automatically merges the changes and generates a new version number V merge Data integration algorithm: Utilize graph database technology to analyze the relationships between data and ensure the logical consistency of the merged data; f. Conflict log recording and analysis: Log storage: Records detailed information about all conflict events, including conflict data, version numbers, and resolution methods; then optimizes priority rules based on historical conflict data.

6. A computer-based multi-platform information construction method according to any one of claims 1 to 5, characterized in that: The following steps are involved: Step S1: The real-time data capture module uses an event-driven data capture algorithm to monitor the transaction logs of each business platform database in real time. When data changes are detected, a data change event is generated and sent to the intelligent routing decision module; Step S2: The intelligent routing decision module receives the data change event, determines the data transmission path based on a routing decision algorithm based on a combination of rules and machine learning, and sends the data change event and routing information to the data conversion adaptation module: Step S3: The data conversion and adaptation module uses the data conversion algorithm of dynamic template matching based on the received source platform and target platform information to perform format conversion and interface adaptation on the data, and then sends the processed data to the data synchronization execution module; Step S4: The data synchronization execution module synchronizes the data to the target platform according to the path determined by the intelligent routing decision module. The conflict detection and resolution module uses a conflict detection and resolution algorithm based on version control and priority determination to detect and resolve data conflicts. Step S5: The data synchronization execution module records the data synchronization results and feeds the synchronization results back to the monitoring and management module. The monitoring and management module monitors and manages the entire data synchronization process in real time.

Images