A business data visualization method and system for enterprise platform
By building a global semantic map and topological association network, the enterprise business logs are automatically processed, and the existing tools need to be developed and customized by themselves is solved, efficient integration and visualization of enterprise data is achieved, the threshold for IT investment and use is reduced, and the variety is met .
Patent Information
- Application Number
- CN202411026991.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-30
- Publication Date
- 2024-11-12
- Estimated Expiration
- 2044-07-30
AI Technical Summary
The existing general data visualization tools require secondary development and customization by enterprises, which increase the threshold for IT investment and use, making it difficult to meet the needs of specific industries and business scenarios. Especially for small and medium-sized enterprises, there is a problem of low data integration and analysis efficiency. .
By obtaining historical business logs based on the enterprise platform, building a global semantic map, performing multi-dimensional anomaly detection and correction, mining business interaction lines, building topological association networks, performing attribute space mapping and dynamic topological evolution simulation, and finally performing three-dimensional visualization to realize business Automated integration and visual presentation of data.
It realizes seamless connection between the internal business system of the enterprise, automatically completes data extraction and integration, reduces IT investment and labor costs, and provides flexible custom analysis functions to meet the strategic planning and operational control needs of different enterprises.
Smart Images

Figure CN118551040B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of data visualization, and in particular to a business data visualization method and system for an enterprise platform. Background Art
[0002] In today's ever-changing business environment, enterprises need to obtain and analyze their own business data in a timely manner to support strategic decision-making and improve operational efficiency. However, enterprises usually accumulate a large amount of business data, including sales, procurement, finance, production and other fields. These data are often scattered in different information systems, with different data formats and structures, which brings challenges to enterprise data integration and analysis. Traditional data analysis methods rely on tools such as Excel for manual processing and report generation, which have problems such as low analysis efficiency and poor visualization. With the development of big data and artificial intelligence technology, many business data visualization solutions have appeared on the market, which can display complex data in intuitive ways such as charts and dashboards. Help enterprise managers better understand the current business situation and discover potential problems. However, existing general-purpose data visualization tools often require enterprises to conduct secondary development and customization to meet the needs of specific industries and business scenarios. This not only increases the company's IT investment, but also requires the participation of personnel with certain data analysis and visualization development experience. There is a high threshold for use for small and medium-sized enterprises. Therefore, there is an urgent need for a business data visualization method tailored to enterprise needs, which can seamlessly connect to various business systems within the enterprise, automatically complete data extraction, integration and visualization, reduce enterprise IT investment and labor costs, and at the same time have flexible custom analysis functions to meet the actual needs of different enterprises in strategic planning, operation control and other aspects. Summary of the invention
[0003] In order to solve the above technical problems, the present invention proposes a business data visualization method and system for an enterprise platform to solve at least one of the above technical problems.
[0004] To achieve the above object, the present invention provides a business data visualization method for an enterprise platform, comprising the following steps:
[0005] Step S1: Obtain historical business logs based on the enterprise platform; construct and process the global semantic graph of the historical business logs to obtain a semantic feature knowledge graph;
[0006] Step S2: Perform multi-dimensional anomaly detection on historical business logs based on the semantic feature knowledge graph to generate outlier difference values; perform outlier correction processing on the historical business logs based on the outlier difference values to obtain outlier corrected business logs;
[0007] Step S3: Perform business interaction mining on the outlier-corrected business logs to obtain business interaction lines; perform topological association fitting on the business interaction lines to construct a business topological association network;
[0008] Step S4: correcting the service log based on the outliers to obtain the service attributes of each interaction line; using the service attributes of each interaction line to perform attribute space mapping on the service topology association network to obtain an attribute mapping topology network;
[0009] Step S5: simulate the dynamic topology evolution of the attribute mapping topology network to obtain the dynamic topology evolution network; analyze the path evolution correlation strength change of the dynamic topology evolution network to obtain the topology correlation strength change data;
[0010] Step S6: Perform three-dimensional visualization processing on the dynamic topology evolution network based on the topology association strength change data, and construct a three-dimensional visualization business evolution model.
[0011] The present invention obtains rich business data by acquiring historical business logs based on an enterprise platform, constructs a global semantic graph for the historical business logs, extracts key information from the business data, and forms a semantic feature knowledge graph. The semantic feature knowledge graph contains semantic associations and feature information of the business data, so as to better understand and analyze the business data. Multi-dimensional anomaly detection is performed on the historical business logs based on the semantic feature knowledge graph, anomalies that are inconsistent with normal business behaviors are identified, and outlier difference values are generated. These values represent the degree of deviation of each data point from normal business behaviors, which are used to quantify the degree of anomalies. Outlier correction is performed on the historical business logs to improve the accuracy and reliability of the data by correcting the outliers. Business interaction mining is performed on the outlier-corrected business logs to discover interactive behaviors in the business data and obtain business interaction lines. These lines represent the interactive relationships between different entities in the business data. Topological association fitting is performed on the business interaction lines to construct a business topological association network to describe the topological structure and association relationships between entities in the business data. The business logs are corrected based on the outliers to obtain the business attributes of each interaction line. The characteristics and attribute information related to the interaction lines are extracted from the business data, and the business attributes of each interaction line are used to perform attribute space mapping on the business topology association network. The business attributes are mapped to the nodes and edges in the topology network to obtain the attribute mapping topology network. This network combines the attribute information of the business data with the topological structure, providing a richer form of business data representation. The attribute mapping topological network is subjected to dynamic topological evolution simulation to simulate the evolution of the topological structure in the business data over time to obtain the dynamic topological evolution network. This network reflects the dynamic changes in the topological relationship in the business data. The dynamic topological evolution network is subjected to path evolution association strength change analysis to quantify the topological association strength and degree of change to obtain topological association strength change data. Based on the topological association strength change data, the dynamic topological evolution network is subjected to three-dimensional visualization processing to display the business data in a visualized form, and a three-dimensional visualization business evolution model is constructed. The evolution process of the business data is presented to the user through visualization technology. The three-dimensional visualization business evolution model helps users to more intuitively understand and analyze the evolution trend and association relationship of business data.
[0012] Preferably, step S1 comprises the following steps:
[0013] Step S11: Obtaining historical business logs based on the enterprise platform;
[0014] Step S12: Perform log data structuring processing on the historical business log to obtain structured fields;
[0015] Step S13: performing global context semantic analysis on the structured field, thereby generating global context semantic features;
[0016] Step S14: Perform deep learning modeling on the global context semantic features and construct a semantic feature knowledge graph.
[0017] The present invention obtains a large amount of original business data by obtaining historical business logs based on an enterprise platform. The historical business logs record past business activities and events and contain rich information and details. The log data is structured and processed to convert the original text logs into structured fields. The structured fields contain key information extracted from the logs, such as timestamps, user IDs, operation types, associated entities, etc. The generation of structured fields makes business data easier to process and analyze, and provides basic data for subsequent steps. Global context semantic analysis is performed on the structured fields, and global context semantic features are extracted by processing the associations and context information between the fields. The global context semantic features contain the semantic associations and context information between the fields, and can better capture the semantic meanings and relationships of the business data. Deep learning modeling is performed on the global context semantic features, and a semantic feature knowledge graph is constructed using a deep learning algorithm. The semantic feature knowledge graph models and represents the semantic features and associations of the business data to form a structured graph. The semantic feature knowledge graph provides a higher level of semantic understanding and expression capabilities, and can more comprehensively describe and analyze the semantic features of the business data.
[0018] Preferably, the specific steps of step S13 are:
[0019] Perform time series analysis on historical business logs to generate business log sequences;
[0020] Calculate the null value rate of the structured field to obtain the field null value rate;
[0021] Perform frequency distribution analysis of null value fields on business log sequences based on field null value rates to generate frequency distribution data of null value fields;
[0022] Based on the frequency distribution data of null value fields, semantic feature analysis is performed on the business log sequence to obtain log semantic features;
[0023] Calculate the conditional probability transfer of adjacent semantic features on the log semantic features to generate a conditional probability transfer data matrix;
[0024] Perform semantic feature association analysis on the conditional probability transfer data matrix to obtain the dependency relationship between neighboring semantic features;
[0025] The global context semantics mining is performed on the business log sequence according to the dependency relationship between the semantic features of neighbors to generate the global context semantic features.
[0026] The present invention generates a business log sequence by performing time series analysis on historical business logs. The log records are arranged in chronological order. The business log sequence reflects the time flow and evolution of business activities, provides time series analysis capabilities for business processes, calculates the null value rate of structured fields, and determines the data missing situation in each field. The field null value rate represents the proportion of null values (missing values) in the field, which helps to understand the integrity and availability of the data. Based on the field null value rate, the business log sequence is subjected to a frequency distribution analysis of null values to obtain the distribution of null values in different fields. The frequency distribution data of null values fields shows the frequency and distribution of null values in each field, which helps to discover patterns and trends of data missing. Based on the frequency distribution data of null values fields, the business log sequence is subjected to semantic feature analysis to obtain the semantic features of the logs. The log semantic features reflect the potential semantics in the business data. Information helps understand the intrinsic meaning and correlation of business data, calculates the conditional probability transfer of adjacent semantic features of log semantic features, and generates a conditional probability transfer data matrix. The conditional probability transfer data matrix shows the conditional probability relationship between adjacent semantic features, and reveals the transfer mode and dependency relationship between semantic features. Semantic feature association analysis is performed on the conditional probability transfer data matrix to find the dependency relationship between adjacent semantic features. The dependency relationship between adjacent semantic features reveals the interaction and correlation degree between semantic features in business data, which helps understand the semantic structure of business data. Based on the dependency relationship between adjacent semantic features, global context semantic mining is performed on business log sequences to generate global context semantic features. Global context semantic features integrate the dependency relationship between different semantic features in business data, and provide a more comprehensive and integrated semantic understanding capability.
[0027] Preferably, the specific steps of step S2 are:
[0028] Step S21: Perform multi-dimensional anomaly detection on historical business logs based on the semantic feature knowledge graph to obtain abnormal outliers;
[0029] Step S22: quantifying the main difference of the abnormal outliers to generate outlier difference values;
[0030] Step S23: Analyze the main data distribution characteristics of the historical business logs to obtain the main data distribution characteristics;
[0031] Step S24: calculating an abnormality correction threshold for the outlier difference value based on the main data distribution characteristics to obtain an abnormality correction range;
[0032] Step S25: Perform outlier correction processing on the historical business log based on the abnormal correction data to obtain an outlier corrected business log.
[0033] The present invention performs multi-dimensional anomaly detection on historical business logs based on semantic feature knowledge graphs, detects abnormal outliers in multiple semantic feature dimensions, and detects abnormal outliers. Abnormal outliers represent data records that are significantly different from normal business behaviors, including errors, abnormal activities or potential problems. The main differences of abnormal outliers are quantified to generate outlier difference values. The outlier difference values reflect the degree of difference between abnormal outliers and normal data, and help measure the degree of abnormality. The main data distribution characteristics of historical business logs are analyzed to obtain the distribution of main data in various dimensions. The main data distribution characteristics describe the data distribution patterns of business data in different dimensions, and provide a benchmark understanding of normal data behavior. Based on the main data distribution characteristics, an abnormal correction threshold is calculated for the outlier difference value to determine the range of abnormal correction. The abnormal correction threshold specifies the range in which the outlier difference value is considered to be abnormal under a specific main data distribution, and is used to determine which outliers need to be corrected. Based on the abnormal correction data, outlier correction processing is performed on the historical business log to obtain an outlier-corrected business log. The outlier-corrected business log corrects the original outlier data to make it closer to the normal data distribution, thereby improving the accuracy and reliability of the data.
[0034] Preferably, the specific steps of step S21 are:
[0035] Perform entity recognition on historical business logs to generate business log entities;
[0036] Perform entity relationship analysis on business log entities based on semantic feature knowledge graph to obtain log entity association data;
[0037] Perform interaction frequency statistics on business log entities and generate log entity interaction frequency values;
[0038] Perform entity association deviation analysis on log entity association data based on the log entity interaction frequency value to generate entity association deviation data;
[0039] Extracting multi-dimensional features from the business log entity to obtain multi-dimensional business feature data; the multi-dimensional business feature data includes time, location and quantity;
[0040] Based on the entity association deviation data, abnormal distribution of multi-dimensional business feature data is identified to obtain abnormal distribution data of business features;
[0041] Calculate the confidence level of abnormality for abnormal distribution data of business features to obtain the confidence score of dimensional business features;
[0042] The confidence score of each dimensional business feature is compared based on the preset log abnormal data threshold. When the preset log abnormal data threshold is less than or equal to the dimensional business feature confidence score, it is determined to be an abnormal business feature;
[0043] According to the abnormal business characteristics, outliers are located in the historical business logs to obtain abnormal outliers.
[0044] The present invention performs entity recognition on historical business logs to extract business log entities with practical significance from the logs. Business log entities are people, places, objects or other entities related to the business, and provide a specific description and identification of business data. Based on the semantic feature knowledge graph, entity relationship analysis is performed on the business log entities to identify the association relationship between entities. The log entity association data reflects the relationship network between business log entities, which helps to understand the interaction and dependency relationship between entities. The interaction frequency statistics are performed on the business log entities to calculate the interaction frequency value between each entity. The log entity interaction frequency value provides a measure of the degree of interaction between entities, which helps to identify entity relationships with frequent or rare interactions. Based on the log entity interaction frequency value, entity association deviation analysis is performed on the log entity association data to calculate the deviation degree of entity association. The entity association deviation data describes the degree of difference between the entity association situation and the expected association, and abnormal entity association patterns are discovered. Multi-dimensional feature extraction is performed on the business log entities, including multiple dimensions such as time, location and quantity. The multi-dimensional business feature data provides The description of business log entities in different dimensions helps to understand the various characteristics of business data. Based on entity association deviation data, the abnormal distribution of multi-dimensional business feature data is identified, and the abnormal distribution patterns of business feature data in different dimensions are discovered. The abnormal distribution data of business features reveals the abnormal situation of business data in different dimensions, helping to detect data distribution that does not conform to the normal pattern. The abnormal degree confidence of business feature abnormal distribution data is calculated to quantify the abnormal degree of business features in each dimension. The dimension business feature confidence score reflects the magnitude of the abnormal degree of business features and is used to evaluate the severity of the abnormality. Based on the preset log abnormal data threshold, the confidence score of each dimension business feature is compared to determine whether it is an abnormal business feature. When the preset log abnormal data threshold is less than or equal to the dimension business feature confidence score, the business feature of the dimension is considered to be abnormal. According to the abnormal business features, the outliers of the historical business logs are located to determine the abnormal outliers. The abnormal outliers represent data records that are significantly different from normal business behaviors, identifying potential problems and abnormal activities.
[0045] Preferably, the specific steps of step S3 are:
[0046] Step S31: performing business interaction mining on the outlier correction business log to obtain business interaction lines;
[0047] Step S32: Quantitatively calculate the business activity frequency and flow of the business interaction line to generate business line direct correlation strength data;
[0048] Step S33: performing potential embedding association analysis on the business interaction lines to obtain a business line indirect association matrix;
[0049] Step S34: Perform topological correlation fitting on the outlier corrected business logs based on the business line direct correlation strength data and the business line indirect correlation matrix to construct a business topological correlation network.
[0050] The present invention discovers business interaction behaviors in business logs by performing business interaction mining on outlier-corrected business logs. Business interaction lines describe the direct interaction relationships between different entities in business logs, provide a visual representation of interaction patterns in business data, perform quantitative calculations on business activity frequencies and flows on business interaction lines, calculate the frequencies and flows of business activities, and business line direct correlation strength data reflects the degree of direct correlation between business activities, helping to understand the importance and attention between different activities in business data. Business interaction lines are subjected to potential embedded correlation analysis to reveal potential correlation patterns between business interaction lines. A business line indirect correlation matrix provides the degree of indirect correlation between business interaction lines, helping to understand the potential correlation between different interaction lines in business data. Based on the business line direct correlation strength data and the business line indirect correlation matrix, topological correlation fitting is performed on outlier-corrected business logs to construct a business topological correlation network, which presents the correlation relationships between different business lines in business data, as well as the degrees of direct and indirect correlation between them.
[0051] Preferably, the specific steps of step S4 are:
[0052] Step S41: Calculate key business indicators for the outlier correction business log to obtain business indicator values;
[0053] Step S42: performing business attribute analysis on the business interaction lines based on the business indicator values to obtain business attributes of each interaction line;
[0054] Step S43: Utilize the service attributes of each interaction line to perform attribute space mapping on the service topology association network to obtain an attribute mapping topology network.
[0055] The present invention calculates key business indicators for outlier corrected business logs and extracts important business indicators from business data. The business indicator values reflect the key performance, efficiency or other important metrics in the business data, and provide basic data for subsequent analysis and visualization. Based on the business indicator values, business attribute analysis is performed on business interaction lines to understand the specific business attributes of each interaction line. The business attributes describe the characteristics and properties of the business interaction lines in the business data, such as the type, function, and importance of the interaction lines. The business attributes of each interaction line are used to perform attribute space mapping on the business topology association network, and the business attributes are mapped to the topology association network. The attribute mapping topology network shows the association relationship between different business lines in the business data, and presents the spatial distribution of business attributes on the topology network.
[0056] Preferably, the specific steps of step S5 are:
[0057] Step S51: performing a time-series interaction evolution analysis on the service interaction line to obtain a time-series evolution rule of the service line;
[0058] Step S52: based on the time-series evolution law of the service line, a dynamic topology evolution simulation is performed on the attribute mapping topology network, thereby obtaining a dynamic topology evolution network;
[0059] Step S53: Identify the service transmission trace of the dynamic topology evolving network to obtain the dynamic service transmission trace;
[0060] Step S54: performing path evolution trend analysis on the dynamic service transmission trajectory to generate service trajectory evolution trend data;
[0061] Step S55: analyzing the correlation strength change of the business trajectory evolution trend data, thereby obtaining topological correlation strength change data.
[0062] The present invention performs a time-series interaction evolution analysis on the business interaction line, observes the changes of the business line over time, and reveals the time-series evolution law of the business line. The time-series interaction evolution analysis can help users understand the activity intensity, change trend, and existing periodic or trend laws of the business line in different time periods. Based on the time-series evolution law of the business line, the attribute mapping topology network is dynamically simulated to generate a dynamic topology evolution network. The dynamic topology evolution simulation applies the time-series evolution law of the business line to the attribute mapping topology network, simulates the dynamic changes of the business line over time, and thus generates a dynamic topology evolution network. The business conduction trajectory of the dynamic topology evolution network is identified to identify The transmission trajectory of the business in the topological network. The business transmission trajectory identification reveals the propagation path and transmission mode of the business in the topological network, helping users understand the transmission process of the business in the network, and performing path evolution trend analysis on the dynamic business transmission trajectory to generate business trajectory evolution trend data. The business trajectory evolution trend data describes the changing trend of the business transmission path over time, helping users understand the evolution and change pattern of the business path. The business trajectory evolution trend data is analyzed for correlation strength changes to obtain topological correlation strength change data. The correlation strength change data reflects the change and evolution of the correlation degree between business paths, helping users understand the dynamic changes of business correlation relationships.
[0063] Preferably, the specific steps of step S6 are:
[0064] Step S61: predicting the evolution of business interactions in a dynamic topology evolution network based on the topology association strength change data, and building a business evolution prediction network;
[0065] Step S62: Perform multi-dimensional spatial projection on the business evolution prediction network to generate a business evolution prediction space model;
[0066] Step S63: Perform three-dimensional visualization processing on the business evolution prediction space model to construct a three-dimensional visualized business evolution model.
[0067] The present invention predicts the evolution of business interactions in a dynamic topology evolution network based on data on changes in the strength of topology associations, constructs a business evolution prediction network, predicts the evolution trend and interaction pattern of the business in the topology network by analyzing changes in the strength of topology associations, and thus constructs a business evolution prediction network. The business evolution prediction network is projected into a multidimensional space to generate a business evolution prediction space model. The multidimensional space projection maps the business evolution prediction network into a space of a lower dimension to better understand and analyze the relationships and patterns of business evolution. The business evolution prediction space model is visualized in three dimensions to construct a three-dimensional visualized business evolution model. By visualizing the business evolution prediction space model as a three-dimensional model, a user can intuitively observe and explore the structure, trend and correlation of business evolution.
[0068] In this specification, a business data visualization system for an enterprise platform is provided, which is used to execute the business data visualization method for an enterprise platform as described above, including:
[0069] The semantic graph module is used to obtain historical business logs based on the enterprise platform; the global semantic graph is constructed and processed for the historical business logs to obtain a semantic feature knowledge graph;
[0070] An outlier correction module is used to perform multi-dimensional anomaly detection on historical business logs based on the semantic feature knowledge graph to generate outlier difference values; perform outlier correction processing on historical business logs based on the outlier difference values to obtain outlier-corrected business logs;
[0071] The topological association module is used to perform business interaction mining on the outlier correction business logs to obtain business interaction lines; perform topological association fitting on the business interaction lines to build a business topological association network;
[0072] An attribute space mapping module is used to correct the business log based on the outlier to obtain the business attributes of each interaction line; the business attributes of each interaction line are used to perform attribute space mapping on the business topology association network to obtain an attribute mapping topology network;
[0073] The topology evolution module is used to simulate the dynamic topology evolution of the attribute mapping topology network, thereby obtaining the dynamic topology evolution network; analyze the change of path evolution correlation strength of the dynamic topology evolution network, thereby obtaining the topology correlation strength change data;
[0074] The three-dimensional visualization module is used to perform three-dimensional visualization processing on the dynamic topology evolution network based on the data of the strength change of topology associations, and to build a three-dimensional visualization business evolution model.
[0075] The present invention obtains historical business logs based on an enterprise platform, and obtains a semantic feature knowledge graph through global semantic graph construction and processing. The semantic feature knowledge graph contains semantic relations and feature information in the historical business logs, and can more comprehensively describe the intrinsic structure and semantic association of business data. Based on the semantic feature knowledge graph, multi-dimensional anomaly detection is performed on the historical business logs to generate outlier difference values. Outlier correction processing is performed on the historical business logs through the outlier difference values to obtain outlier-corrected business logs. The outlier correction module can discover and correct abnormal data in the historical business logs to improve the accuracy and reliability of the data. Business interaction mining is performed on the outlier-corrected business logs to obtain business interaction lines. A business topology association network is constructed through topological association fitting. The topology association module can discover the interaction relationship between business data and construct a business topology association network to help understand the overall structure and association of business data. Attribute extraction is performed on the outlier-corrected business logs to obtain The business attributes of each interactive line are used to perform attribute space mapping of the business topology association network to obtain the attribute mapping topology network. The attribute space mapping module can integrate the attribute information of the business data into the topology association network to provide richer data expression and analysis capabilities. The dynamic topology evolution simulation of the attribute mapping topology network is performed to obtain a dynamic topology evolution network. The topology association strength change data is obtained through the analysis of the path evolution association strength change. The topology evolution module can simulate the dynamic evolution process of business data and reveal the evolution trend and correlation change of business data by analyzing the topology association strength change data. Based on the topology association strength change data, the dynamic topology evolution network is visualized in three dimensions to construct a three-dimensional visualization business evolution model, allowing users to intuitively observe and understand the evolution process and topological structure of business data. The three-dimensional visualization module can provide an intuitive visualization interface to help users better understand and analyze the evolution of business data. BRIEF DESCRIPTION OF THE DRAWINGS
[0076] Figure 1 A schematic diagram of the steps of a business data visualization method for an enterprise platform according to the present invention;
[0077] Figure 2 Detailed implementation flow chart of step S1;
[0078] Figure 3 Detailed implementation flow chart of step S2;
[0079] Figure 4 Detailed implementation flow chart of step S3. DETAILED DESCRIPTION
[0080] It should be understood that the specific embodiments described herein are only used to explain the present invention, and are not used to limit the present invention.
[0081] The example of this application provides a business data visualization method and system for an enterprise platform. The execution subject of the business data visualization method and system for an enterprise platform includes but is not limited to: mechanical equipment, data processing platform, cloud server node, network upload device, etc. equipped with the system can be regarded as the general computing node of this application, and the data processing platform includes but is not limited to: at least one of an audio image management system, an information management system, and a cloud data management system.
[0082] See also Figures 1 to 4 The present invention provides a business data visualization method for an enterprise platform, and the business data visualization method for an enterprise platform comprises the following steps:
[0083] Step S1: Obtain historical business logs based on the enterprise platform; construct and process the global semantic graph of the historical business logs to obtain a semantic feature knowledge graph;
[0084] Step S2: Perform multi-dimensional anomaly detection on historical business logs based on the semantic feature knowledge graph to generate outlier difference values; perform outlier correction processing on the historical business logs based on the outlier difference values to obtain outlier corrected business logs;
[0085] Step S3: Perform business interaction mining on the outlier-corrected business logs to obtain business interaction lines; perform topological association fitting on the business interaction lines to construct a business topological association network;
[0086] Step S4: correcting the service log based on the outliers to obtain the service attributes of each interaction line; using the service attributes of each interaction line to perform attribute space mapping on the service topology association network to obtain an attribute mapping topology network;
[0087] Step S5: simulate the dynamic topology evolution of the attribute mapping topology network to obtain the dynamic topology evolution network; analyze the path evolution correlation strength change of the dynamic topology evolution network to obtain the topology correlation strength change data;
[0088] Step S6: Perform three-dimensional visualization processing on the dynamic topology evolution network based on the topology association strength change data, and construct a three-dimensional visualization business evolution model.
[0089] The present invention obtains rich business data by acquiring historical business logs based on an enterprise platform, constructs a global semantic graph for the historical business logs, extracts key information from the business data, and forms a semantic feature knowledge graph. The semantic feature knowledge graph contains semantic associations and feature information of the business data, so as to better understand and analyze the business data. Multi-dimensional anomaly detection is performed on the historical business logs based on the semantic feature knowledge graph, anomalies that are inconsistent with normal business behaviors are identified, and outlier difference values are generated. These values represent the degree of deviation of each data point from normal business behaviors, which are used to quantify the degree of anomalies. Outlier correction is performed on the historical business logs to improve the accuracy and reliability of the data by correcting the outliers. Business interaction mining is performed on the outlier-corrected business logs to discover interactive behaviors in the business data and obtain business interaction lines. These lines represent the interactive relationships between different entities in the business data. Topological association fitting is performed on the business interaction lines to construct a business topological association network to describe the topological structure and association relationships between entities in the business data. The business logs are corrected based on the outliers to obtain the business attributes of each interaction line. The characteristics and attribute information related to the interaction lines are extracted from the business data, and the business attributes of each interaction line are used to perform attribute space mapping on the business topology association network. The business attributes are mapped to the nodes and edges in the topology network to obtain the attribute mapping topology network. This network combines the attribute information of the business data with the topological structure, providing a richer form of business data representation. The attribute mapping topological network is subjected to dynamic topological evolution simulation to simulate the evolution of the topological structure in the business data over time to obtain the dynamic topological evolution network. This network reflects the dynamic changes in the topological relationship in the business data. The dynamic topological evolution network is subjected to path evolution association strength change analysis to quantify the topological association strength and degree of change to obtain topological association strength change data. Based on the topological association strength change data, the dynamic topological evolution network is subjected to three-dimensional visualization processing to display the business data in a visualized form, and a three-dimensional visualization business evolution model is constructed. The evolution process of the business data is presented to the user through visualization technology. The three-dimensional visualization business evolution model helps users to more intuitively understand and analyze the evolution trend and association relationship of business data.
[0090] In the embodiment of the present invention, refer to Figure 1 , is a schematic flow chart of the steps of a business data visualization method for an enterprise platform of the present invention. In this example, the steps of the business data visualization method for an enterprise platform include:
[0091] Step S1: Obtain historical business logs based on the enterprise platform; construct and process the global semantic graph of the historical business logs to obtain a semantic feature knowledge graph;
[0092] In this embodiment, after obtaining authorization from the enterprise platform, historical business log data is obtained from the enterprise platform. These log data include operation logs of the enterprise's internal system, user interaction records, business process records, etc. The historical business logs are preprocessed, including data cleaning, deduplication and format conversion, etc. Natural language processing (NLP) technology is used to perform semantic analysis and entity recognition on the log text, and keywords, topics and entity information are extracted. Based on the extracted semantic information, a semantic graph is constructed. The semantic graph is a graph structure in which nodes represent entities or keywords, and edges represent relationships between entities or semantic associations between keywords. When constructing the semantic graph, a graph database or graph analysis tool is used to store and process graph structure data. According to the constructed semantic graph, semantic feature knowledge is extracted, including entity attributes, relationship information and the importance of keywords.
[0093] Step S2: Perform multi-dimensional anomaly detection on historical business logs based on the semantic feature knowledge graph to generate outlier difference values; perform outlier correction processing on the historical business logs based on the outlier difference values to obtain outlier corrected business logs;
[0094] In this embodiment, multi-dimensional anomaly detection is performed on historical business logs, and the multiple dimensions include time dimension, business dimension, user dimension, etc. The outlier difference value is calculated by comparing the difference between the current log data and the historical data. The outlier difference value indicates the degree of deviation between the current data and the historical data. According to the generated outlier difference value, the historical business log is corrected for outliers. Different correction strategies are adopted for outliers. A common strategy is to regard outliers as outliers and delete or mark them. Another strategy is to adjust or interpolate data according to the difference value of the outliers to make it closer to the data within the normal range. The corrected business log is called an outlier-corrected business log, in which the outliers have been processed and the data is more accurate and reliable.
[0095] Step S3: Perform business interaction mining on the outlier-corrected business logs to obtain business interaction lines; perform topological association fitting on the business interaction lines to construct a business topological association network;
[0096] In this embodiment, business logs are corrected for outliers to mine and analyze business interactions. Business interactions include system operations, user behaviors, business processes, etc. Business interaction lines are identified and extracted based on timestamps and operation records in business logs. Business interaction lines are sequences of related operations or events, representing the execution path of business processes. Based on the extracted business interaction lines, topological association fitting is performed on the interaction lines. Topological association represents the relationship between business interaction lines, such as sequence, dependency, etc. A business topological association network is constructed based on the execution sequence and association rules of the business interaction lines. The business topological association network is a graph structure, in which nodes represent business interaction lines and edges represent the relationship between interaction lines. Business analysis and decision support are performed using the constructed business topological association network. The business topological association network is analyzed through graph analysis algorithms, such as community discovery and graph centrality analysis, to discover key business processes, bottlenecks or abnormal situations. Based on the analysis results, support is provided for business decisions, such as optimizing business processes, improving user experience, or adjusting system configuration.
[0097] Step S4: correcting the service log based on the outliers to obtain the service attributes of each interaction line; using the service attributes of each interaction line to perform attribute space mapping on the service topology association network to obtain an attribute mapping topology network;
[0098] In this embodiment, the outlier correction service log is analyzed to extract the service attributes of each interaction line, including the type of interaction line, execution time, participant role, operation type, etc.
[0099] Based on the information and domain knowledge recorded in the business log, the business attributes of each interaction line are identified and extracted. The business attributes of each interaction line are mapped into the attribute space to construct an attribute mapping topology network. The attribute mapping topology network is a graph structure based on the attribute space, in which nodes still represent interaction lines and edges represent the relationship between interaction lines.
[0100] The attribute space is a multidimensional space, in which each dimension represents a business attribute. The business attributes of the interaction line are mapped to the corresponding dimensions in the attribute space.
[0101] Step S5: simulate the dynamic topology evolution of the attribute mapping topology network to obtain the dynamic topology evolution network; analyze the path evolution correlation strength change of the dynamic topology evolution network to obtain the topology correlation strength change data;
[0102] In this embodiment, a dynamic topology evolution simulation is performed based on the attribute mapping topology network. The dynamic topology evolution simulation simulates the evolution process of the business interaction line. According to the relationship and attribute changes of the interaction line, a dynamic topology evolution network is constructed. In the simulation process, the new addition, deletion, attribute changes of the interaction line, etc. are considered, so that the topology network can reflect the dynamic changes of the business. Based on the dynamic topology evolution network, the path evolution correlation strength change analysis is performed. The path evolution correlation strength change analysis is to analyze the changes of the path in the topology network, including the formation, disappearance, and strength changes of the path. By calculating the weight of the path, degree centrality and other indicators, the correlation strength of the path is evaluated. At the same time, the change trend of the path is analyzed to identify the key path and the pattern of correlation strength change.
[0103] Step S6: Perform three-dimensional visualization processing on the dynamic topology evolution network based on the topology association strength change data, and construct a three-dimensional visualization business evolution model.
[0104] In this embodiment, a suitable node representation method is selected, such as a sphere, a cube, etc., to represent an interaction line or path, and a connection line or a pipe is used to represent the association relationship between the interaction lines or paths. According to the change of the association strength, the thickness or color of the connection line is adjusted, and an animation or interactive function is designed so that the evolution process of the business topology is presented in a time series manner. The size or height of the node is set according to the magnitude of the association strength. Nodes with higher association strength are represented as larger or higher shapes, while nodes with lower association strength are represented as smaller or lower shapes. For connection representation, the thickness or color of the connection line is adjusted according to the change of the association strength. Connections with higher association strength are represented as Thicker or brighter lines, while connections with lower association strength are represented as thinner or lighter lines. A three-dimensional visual business evolution model is constructed. According to the results of design and data mapping, the corresponding three-dimensional visualization tools or libraries are used to construct a three-dimensional visual business evolution model. The dynamic topology evolution network and its association strength change data are imported into the tool, and the corresponding parameters and interaction methods are set for three-dimensional visualization demonstration and analysis. The dynamic evolution process of the business topology is displayed through the three-dimensional visual business evolution model. Interactive operations such as rotation and zooming are performed to observe and analyze the topological changes of the business, and the evolution process is presented through animation, timeline or interactive controls.
[0105] In this embodiment, refer to Figure 2 , is a flowchart of detailed implementation steps of step S1. In this embodiment, the detailed implementation steps of step S1 include:
[0106] Step S11: Obtaining historical business logs based on the enterprise platform;
[0107] Step S12: Perform log data structuring processing on the historical business log to obtain structured fields;
[0108] Step S13: performing global context semantic analysis on the structured field, thereby generating global context semantic features;
[0109] Step S14: Perform deep learning modeling on the global context semantic features and construct a semantic feature knowledge graph.
[0110] In this embodiment, the enterprise platform from which historical business logs need to be obtained is determined, such as the enterprise's information system, application, network equipment, etc. For the selected platform, the relevant department or person in charge is communicated to obtain the storage location, format and acquisition method of the historical business log. According to the acquisition method, appropriate tools or methods are used to obtain historical business log data from the enterprise platform. Identity authentication or permission control is required to ensure that the required historical log data is obtained. The obtained historical business logs are preprocessed and cleaned to remove invalid or redundant data to ensure data accuracy and consistency. According to business requirements and the characteristics of log data, a suitable log data structure is designed to convert log data into structured fields. The corresponding log processing tools or programming languages are used to parse and extract key information from the log data and map it to predefined structured fields. Regular expressions, word segmentation technology and other methods are used to extract information and match fields. Global context semantic analysis is performed on structured fields to understand the association and context semantics between fields. Natural language processing (NLP) technology is used. Such as word vector models (such as Word2Vec, GloVe), semantic role labeling, entity recognition, etc., to process and analyze the text information in the structured fields. Based on NLP technology and semantic analysis methods, the structured fields are converted into global context semantic features. These features include word meaning similarity, semantic roles, entity relationships, etc., which are used to express the semantic associations between fields. Based on the global context semantic features, a deep learning model is designed for modeling. Using graph neural networks, knowledge graph representation learning and other technologies, the global context semantic features are converted into a representation that can be calculated and analyzed. According to specific modeling needs, appropriate deep learning algorithms and model architectures are selected, such as graph convolutional networks (GCN), attention mechanisms (Attention), graph attention networks (GAT), etc., to construct semantic feature knowledge graphs. Based on the selected model, training and optimization are performed to capture the relationship between global context semantic features and generate a knowledge graph with semantic information. The knowledge graph is represented in the form of nodes and edges, where nodes represent semantic features and edges represent the association between features.
[0111] In this embodiment, the specific steps of step S13 are:
[0112] Perform time series analysis on historical business logs to generate business log sequences;
[0113] Calculate the null value rate of the structured field to obtain the field null value rate;
[0114] Perform frequency distribution analysis of null value fields on business log sequences based on field null value rates to generate frequency distribution data of null value fields;
[0115] Based on the frequency distribution data of null value fields, semantic feature analysis is performed on the business log sequence to obtain log semantic features;
[0116] Calculate the conditional probability transfer of adjacent semantic features on the log semantic features to generate a conditional probability transfer data matrix;
[0117] Perform semantic feature association analysis on the conditional probability transfer data matrix to obtain the dependency relationship between neighboring semantic features;
[0118] The global context semantics mining is performed on the business log sequence according to the dependency relationship between the semantic features of neighbors to generate the global context semantic features.
[0119] In this embodiment, historical business log data is loaded, and the logs are sorted according to timestamps or other time information to ensure that they are organized in chronological order. According to business needs and the structure of log data, key fields that need to be extracted are determined, such as timestamps, event types, user IDs, etc. The log data is read line by line, key fields are extracted, and they are organized into business log sequences in chronological order. For structured fields, the number of null values in each field is counted, that is, the number of missing values in the field, and the null value rate of each field is calculated, that is, the number of null values in the field is divided by the total number of samples or the number of valid samples, to obtain the null value rate of each field as a reference indicator for subsequent analysis. For each Field, count the frequency distribution of null values in the business log sequence, that is, the ratio of the number of null values in each field to the total number of samples or the number of valid samples. Draw a histogram or other visual chart based on the null value frequency distribution data to show the null value situation and distribution of each field. Use the null value field frequency distribution data to perform semantic feature analysis on the business log sequence. According to the null value rate of the field and the distribution pattern of the null value, the fields are divided into high null value rate fields, low null value rate fields or other categories, and each field is assigned a corresponding semantic feature label. According to the analysis results, each log in the business log sequence is assigned a corresponding semantic feature to form a log. Log semantic feature sequence, based on the log semantic feature sequence, calculate the conditional probability transfer between adjacent semantic features, that is, calculate the probability of the next feature appearing when the previous feature is known, count the transfer frequency between each semantic feature and its successor feature, and normalize it to a probability value, store the transfer probability as a data matrix, where the row represents the current feature, the column represents the successor feature, and the matrix element represents the transfer probability from the current feature to the successor feature. Based on the conditional probability transfer data matrix, perform semantic feature association analysis to reveal the dependency relationship between adjacent semantic features, and use clustering analysis, association rule mining, network analysis and other methods to find features. The correlation patterns and dependencies between them are obtained according to the analysis results, such as the correlation and conditional probability between feature A and feature B. Based on the dependencies between neighboring semantic features, global context semantic mining is performed on the business log sequence to discover a wider range of semantic associations and context information between features. The dependencies between neighboring features are used to expand or supplement features. For example, related features are combined into new features according to dependencies, or missing feature values are filled to obtain more complete context information. Based on the results of global context semantic mining, global context semantic features are generated for subsequent analysis and application.
[0120] In this embodiment, refer to Figure 3 , is a flowchart of detailed implementation steps of step S2. In this embodiment, the detailed implementation steps of step S2 include:
[0121] Step S21: Perform multi-dimensional anomaly detection on historical business logs based on the semantic feature knowledge graph to obtain abnormal outliers;
[0122] Step S22: quantifying the main difference of the abnormal outliers to generate outlier difference values;
[0123] Step S23: Analyze the main data distribution characteristics of the historical business logs to obtain the main data distribution characteristics;
[0124] Step S24: calculating an abnormality correction threshold for the outlier difference value based on the main data distribution characteristics to obtain an abnormality correction range;
[0125] Step S25: Perform outlier correction processing on the historical business log based on the abnormal correction data to obtain an outlier corrected business log.
[0126] In this embodiment, a semantic feature knowledge graph is established, in which nodes represent semantic features and edges represent relationships between features. Domain knowledge or data analysis methods are used to construct a knowledge graph, and the knowledge graph is used for multi-dimensional anomaly detection. A graph-structured anomaly detection algorithm, such as a graph-based outlier detection algorithm or a graph-based abnormal subgraph discovery algorithm, is used to analyze the performance of historical business log data on the knowledge graph, identify abnormal outliers, i.e., data points with abnormal relationships with other nodes or significantly different feature values. For abnormal outliers, they are compared with normal data, the difference between the outliers and the main body of normal data is quantified, and the difference values between the outliers and normal data in various dimensions are calculated. For example, Euclidean distance, Manhattan distance or other distance measurement methods are used, and the difference value is used as the difference measurement of the outliers to represent the degree of difference between the outliers and normal data. The main data distribution characteristics of the normal data in the historical business logs are analyzed. Analysis, considering the distribution of each feature, mean, variance, skewness, kurtosis and other statistical characteristics, using histograms, probability density plots, box plots and other visualization methods to explore the distribution characteristics of the data, and perform numerical calculations and statistical analysis, using the main data distribution characteristics to calculate the anomaly correction threshold of the abnormal outlier difference value, according to the normal distribution assumption or other distribution models, calculate the threshold range of the outlier difference value, such as based on the mean and standard deviation method, determine the anomaly correction range, mark the outliers outside the range as abnormal data that need to be corrected, use the anomaly correction range to correct the outliers in the historical business logs, correct the outliers outside the anomaly correction range to a reasonable value, use interpolation, replace with the mean or median and other methods to correct, the corrected business log data is the outlier-corrected business log, these data points have been corrected for outliers while maintaining the overall distribution characteristics of the log.
[0127] In this embodiment, the specific steps of step S21 are:
[0128] Perform entity recognition on historical business logs to generate business log entities;
[0129] Perform entity relationship analysis on business log entities based on semantic feature knowledge graph to obtain log entity association data;
[0130] Perform interaction frequency statistics on business log entities and generate log entity interaction frequency values;
[0131] Perform entity association deviation analysis on log entity association data based on the log entity interaction frequency value to generate entity association deviation data;
[0132] Extracting multi-dimensional features from the business log entity to obtain multi-dimensional business feature data; the multi-dimensional business feature data includes time, location and quantity;
[0133] Based on the entity association deviation data, abnormal distribution of multi-dimensional business feature data is identified to obtain abnormal distribution data of business features;
[0134] Calculate the confidence level of abnormality for abnormal distribution data of business features to obtain the confidence score of dimensional business features;
[0135] The confidence score of each dimensional business feature is compared based on the preset log abnormal data threshold. When the preset log abnormal data threshold is less than or equal to the dimensional business feature confidence score, it is determined to be an abnormal business feature;
[0136] According to the abnormal business characteristics, outliers are located in the historical business logs to obtain abnormal outliers.
[0137] In this embodiment, natural language processing (NLP) technology is used to perform text parsing and entity recognition on historical business logs, and entity recognition algorithms, such as named entity recognition (NER) algorithms, are applied to extract entities in business logs, such as names, places, organizations, etc. The recognized and extracted entities will be used as basic data for subsequent steps. The semantic feature knowledge graph is used to associate business log entities and their attributes with nodes and edges in the graph, and the association relationships between business log entities, such as contextual relationships, dependencies, similarities, etc., are analyzed. Graph algorithms are used to reason and analyze entity relationships, generate log entity association data, and record association information between business log entities for subsequent analysis and statistics of business logs. The interaction frequency between entities, that is, the number of times they appear in the same context in historical logs. For each pair of entities, their interaction frequency values are calculated to indicate the closeness of their interaction. The interaction frequency values of log entities are analyzed to compare whether there is a deviation in the interaction frequency between different entities. Indicators such as the interaction frequency difference and relative frequency deviation between entities are calculated to measure the degree of deviation of entity associations. Entity association deviation data is generated and the deviation information of entity associations is recorded for subsequent analysis and anomaly detection. Multidimensional features are extracted from business log entities, including time, location, and quantity. Time features include timestamps, dates, time periods, etc., which are used to analyze the time distribution and trends of business activities. Location features include geographic coordinates. , region, and location relationships, etc., are used to analyze the spatial distribution and correlation of business activities. Quantitative characteristics include business quantity and frequency, etc., which are used to analyze the quantitative distribution and regularity of business activities. Entity association deviation data is used to analyze the distribution of multi-dimensional business feature data. Statistical methods, data mining algorithms, or machine learning models are used to identify abnormal distribution patterns in multi-dimensional business feature data, generate business feature abnormal distribution data, record abnormal distribution patterns and abnormality levels for subsequent analysis and anomaly detection, calculate the abnormality level of business feature abnormal distribution data, use statistical methods, outlier detection algorithms, etc., calculate the confidence score of each dimensional business feature according to the abnormality level, and indicate the abnormality level of the business feature of that dimension. Degree, set a pre-defined log abnormal data threshold, which is used to determine whether the abnormal degree of the dimensional business feature exceeds the threshold, and compare the confidence score of each dimensional business feature with the preset threshold. If the confidence score of the dimensional business feature is greater than the threshold, it means that the dimensional business feature is abnormal and needs to be processed later. According to the business features determined to be abnormal, the historical business logs are screened and filtered, and only the log data related to the abnormal business features are retained. The outlier detection algorithm, such as based on cluster analysis, statistical methods or machine learning models, is applied to the retained log data to locate the outliers, and finally the abnormal outliers are obtained, that is, the abnormal data points or abnormal events related to the abnormal business features in the historical business logs.
[0138] In this embodiment, refer to Figure 4, is a flowchart of detailed implementation steps of step S3. In this embodiment, the detailed implementation steps of step S3 include:
[0139] Step S31: performing business interaction mining on the outlier correction business log to obtain business interaction lines;
[0140] Step S32: Quantitatively calculate the business activity frequency and flow of the business interaction line to generate business line direct correlation strength data;
[0141] Step S33: performing potential embedding association analysis on the business interaction lines to obtain a business line indirect association matrix;
[0142] Step S34: Perform topological correlation fitting on the outlier corrected business logs based on the business line direct correlation strength data and the business line indirect correlation matrix to construct a business topological correlation network.
[0143] In this embodiment, business interaction mining is performed on the corrected outlier business log data. By analyzing the interaction behaviors in the business logs, such as request-response, call relationships, etc., the interaction relationships between businesses are identified. According to the interaction behaviors and associated information in the business logs, business interaction lines are constructed to represent the direct interaction relationships between businesses. The business interaction lines are analyzed, the frequency and flow of business activities are counted, the frequency of each business interaction line is calculated to represent the frequency of interaction between businesses, the flow of each business interaction line is calculated to represent the amount of data or resource consumption of interaction between businesses, the business line direct association strength data is generated, and the direct association strength between businesses is recorded for subsequent analysis and topological association fitting. , use potential embedding models (such as node embedding, graph embedding, etc.) to perform association analysis on business interaction lines, convert business interaction lines into vector representations, and learn the potential associations between business lines through the embedding model. Based on the learned embedding vectors, construct a business line indirect association matrix to represent the degree of indirect association between business lines. Comprehensively utilize the business line direct association intensity data and the business line indirect association matrix to perform topological association fitting. Based on the direct association intensity of the business lines, construct direct connection edges between business lines. Based on the indirect association matrix of the business lines, construct indirect connection edges between business lines. Combine direct connection edges and indirect connection edges to construct a business topological association network to represent the association relationship between business lines.
[0144] In this embodiment, step S4 includes the following steps:
[0145] Step S41: Calculate key business indicators for the outlier correction business log to obtain business indicator values;
[0146] Step S42: performing business attribute analysis on the business interaction lines based on the business indicator values to obtain business attributes of each interaction line;
[0147] Step S43: Utilize the service attributes of each interaction line to perform attribute space mapping on the service topology association network to obtain an attribute mapping topology network.
[0148] In this embodiment, key business indicators are calculated for the corrected outlier business log data, and the key business indicators that need to be calculated are determined according to business needs and goals, such as response time, throughput, error rate, etc. For each business indicator, the business log data is calculated and counted accordingly to obtain the corresponding business indicator value, and business attribute analysis is performed on the business interaction lines. According to the business indicator values, the business interaction lines are classified or marked, for example, the interaction lines are divided into different business attributes such as high load and low response, the characteristics and behaviors of the business interaction lines are analyzed, and the specific business attributes of each interaction line are identified. The business topology association network is mapped in attribute space, and the business attributes of each interaction line are mapped to the nodes of the topological network, which are represented by the color, size, shape, etc. of the nodes. According to the similarity or correlation of the business attributes, the layout and connection method between the nodes in the topological network are adjusted to make the nodes with similar attributes closer or connected.
[0149] In this embodiment, the specific steps of step S5 are:
[0150] Step S51: performing a time-series interaction evolution analysis on the service interaction line to obtain a time-series evolution rule of the service line;
[0151] Step S52: based on the time-series evolution law of the service line, a dynamic topology evolution simulation is performed on the attribute mapping topology network, thereby obtaining a dynamic topology evolution network;
[0152] Step S53: Identify the service transmission trace of the dynamic topology evolving network to obtain the dynamic service transmission trace;
[0153] Step S54: performing path evolution trend analysis on the dynamic service transmission trajectory to generate service trajectory evolution trend data;
[0154] Step S55: analyzing the correlation strength change of the business trajectory evolution trend data, thereby obtaining topological correlation strength change data.
[0155] In this embodiment, a time series analysis is performed on the business interaction line to observe the evolution process and mode of the business line. Based on the time series data, the change trend, periodicity and trend of the business interaction line are analyzed. Through statistical and data mining methods, the time series evolution law of the business line, such as peak period, trough period, etc., is obtained. The attribute mapping topology network is dynamically evolved. According to the time series evolution law, the attribute mapping and connection relationship of the nodes in the topology network are adjusted to simulate the topological changes of the business line in different time periods. The simulation process is iterated to generate a series of topological network snapshots to represent the dynamic topological evolution process of the business line. According to the snapshot sequence of the dynamic topological evolution network, the business transmission trajectory is identified and the topology is analyzed. Changes and connection relationships between nodes in the network, identify the path and transmission direction of business conduction, determine the business attributes corresponding to the conduction trajectory based on the node attribute mapping information, analyze the path evolution trend of the dynamic business conduction trajectory, observe changes in the business conduction path, identify the path growth, shortening, branching and other evolution trends, generate business trajectory evolution trend data, record changes in path evolution trends, analyze changes in the strength of the association between business trajectories based on the business trajectory evolution trend data, compare the evolution trends of different business trajectories, identify the pattern of enhancement, weakening or change of association strength, generate topological association strength change data, and record changes in the strength of the association between business trajectories.
[0156] In this embodiment, step S6 includes the following steps:
[0157] Step S61: predicting the evolution of business interactions in a dynamic topology evolution network based on the topology association strength change data, and building a business evolution prediction network;
[0158] Step S62: Perform multi-dimensional spatial projection on the business evolution prediction network to generate a business evolution prediction space model;
[0159] Step S63: Perform three-dimensional visualization processing on the business evolution prediction space model to construct a three-dimensional visualized business evolution model.
[0160] In this embodiment, business interaction evolution is predicted for a dynamic topology evolution network, patterns and trends of changes in topology association strength are analyzed, and the strength and changes of associations between future business interaction lines are predicted. Based on the prediction results, a business evolution prediction network is constructed to represent the associations and changes between future business lines. Multidimensional space projection is performed on the business evolution prediction network. According to the changes in the strength of associations between nodes in the business evolution prediction network, the nodes are mapped to coordinate points in the multidimensional space. Through the multidimensional space projection, a business evolution prediction space model is generated to represent the spatial relationship and evolution trend between the nodes. The business evolution prediction space model is three-dimensionally visualized, and the coordinate points obtained by the multidimensional space projection are mapped to three-dimensional space to represent the positions of the nodes in the three-dimensional space. According to the changes in the strength of associations between the nodes and the spatial positions, a three-dimensional visualized business evolution model is constructed to display the evolution trends and associations of the business lines in a visualized manner.
[0161] In this embodiment, a business data visualization system for an enterprise platform is provided, which is used to execute the business data visualization method for an enterprise platform as described above, including:
[0162] The semantic graph module is used to obtain historical business logs based on the enterprise platform; the global semantic graph is constructed and processed for the historical business logs to obtain a semantic feature knowledge graph;
[0163] An outlier correction module is used to perform multi-dimensional anomaly detection on historical business logs based on the semantic feature knowledge graph to generate outlier difference values; perform outlier correction processing on historical business logs based on the outlier difference values to obtain outlier-corrected business logs;
[0164] The topological association module is used to perform business interaction mining on the outlier correction business logs to obtain business interaction lines; perform topological association fitting on the business interaction lines to build a business topological association network;
[0165] An attribute space mapping module is used to correct the business log based on the outlier to obtain the business attributes of each interaction line; the business attributes of each interaction line are used to perform attribute space mapping on the business topology association network to obtain an attribute mapping topology network;
[0166] The topology evolution module is used to simulate the dynamic topology evolution of the attribute mapping topology network, thereby obtaining the dynamic topology evolution network; analyze the change of path evolution correlation strength of the dynamic topology evolution network, thereby obtaining the topology correlation strength change data;
[0167] The three-dimensional visualization module is used to perform three-dimensional visualization processing on the dynamic topology evolution network based on the data of the strength change of topology associations, and to build a three-dimensional visualization business evolution model.
[0168] The present invention obtains historical business logs based on an enterprise platform, and obtains a semantic feature knowledge graph through global semantic graph construction and processing. The semantic feature knowledge graph contains semantic relations and feature information in the historical business logs, and can more comprehensively describe the intrinsic structure and semantic association of business data. Based on the semantic feature knowledge graph, multi-dimensional anomaly detection is performed on the historical business logs to generate outlier difference values. Outlier correction processing is performed on the historical business logs through the outlier difference values to obtain outlier-corrected business logs. The outlier correction module can discover and correct abnormal data in the historical business logs to improve the accuracy and reliability of the data. Business interaction mining is performed on the outlier-corrected business logs to obtain business interaction lines. A business topology association network is constructed through topological association fitting. The topology association module can discover the interaction relationship between business data and construct a business topology association network to help understand the overall structure and association of business data. Attribute extraction is performed on the outlier-corrected business logs to obtain The business attributes of each interactive line are used to perform attribute space mapping of the business topology association network to obtain the attribute mapping topology network. The attribute space mapping module can integrate the attribute information of the business data into the topology association network to provide richer data expression and analysis capabilities. The dynamic topology evolution simulation of the attribute mapping topology network is performed to obtain a dynamic topology evolution network. The topology association strength change data is obtained through the analysis of the path evolution association strength change. The topology evolution module can simulate the dynamic evolution process of business data and reveal the evolution trend and correlation change of business data by analyzing the topology association strength change data. Based on the topology association strength change data, the dynamic topology evolution network is visualized in three dimensions to construct a three-dimensional visualization business evolution model, allowing users to intuitively observe and understand the evolution process and topological structure of business data. The three-dimensional visualization module can provide an intuitive visualization interface to help users better understand and analyze the evolution of business data.
[0169] Therefore, the embodiments should be regarded as illustrative and non-restrictive from all points, and the scope of the present invention is limited by the appended claims rather than the above description, and it is therefore intended that all changes falling within the meaning and range of equivalent elements of the application documents are included in the present invention.
[0170] The above is only a specific embodiment of the present invention, so that those skilled in the art can understand or implement the present invention. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but should conform to the widest scope consistent with the principles and novel features invented herein.
Claims
1. A business data visualization method for an enterprise platform, characterized in that: The following steps are involved: Step S1: Obtain historical business logs based on the enterprise platform; construct and process the global semantic graph of the historical business logs to obtain a semantic feature knowledge graph; The specific steps of step S1 are: Step S11: Obtaining historical business logs based on the enterprise platform; Step S12: Perform log data structuring processing on the historical business log to obtain structured fields; Step S13: performing global context semantic analysis on the structured field, thereby generating global context semantic features; Step S14: Perform deep learning modeling on the global context semantic features to construct a semantic feature knowledge graph; Step S2: Perform multi-dimensional anomaly detection on historical business logs based on the semantic feature knowledge graph to generate outlier difference values; perform outlier correction processing on the historical business logs based on the outlier difference values to obtain outlier corrected business logs; The specific steps of step S2 are: Step S21: Perform multi-dimensional anomaly detection on historical business logs based on the semantic feature knowledge graph to obtain abnormal outliers; Step S22: quantifying the main difference of the abnormal outliers to generate outlier difference values; Step S23: Analyze the main data distribution characteristics of the historical business logs to obtain the main data distribution characteristics; Step S24: calculating an abnormality correction threshold for the outlier difference value based on the main data distribution characteristics to obtain an abnormality correction range; Step S25: performing outlier correction processing on the historical business log based on the abnormal correction data to obtain an outlier corrected business log; The specific steps of step S21 are: Perform entity recognition on historical business logs to generate business log entities; Perform entity relationship analysis on business log entities based on semantic feature knowledge graph to obtain log entity association data; Perform interaction frequency statistics on business log entities and generate log entity interaction frequency values; Perform entity association deviation analysis on log entity association data based on the log entity interaction frequency value to generate entity association deviation data; Extracting multi-dimensional features from the business log entity to obtain multi-dimensional business feature data; the multi-dimensional business feature data includes time, location and quantity; Based on the entity association deviation data, abnormal distribution of multi-dimensional business feature data is identified to obtain abnormal distribution data of business features; Calculate the confidence level of abnormality for abnormal distribution data of business features to obtain the confidence score of dimensional business features; The confidence score of each dimensional business feature is compared based on the preset log abnormal data threshold. When the preset log abnormal data threshold is less than or equal to the dimensional business feature confidence score, it is determined to be an abnormal business feature; According to the abnormal business characteristics, outliers are located in the historical business logs to obtain abnormal outliers; Step S3: Perform business interaction mining on the outlier-corrected business logs to obtain business interaction lines; perform topological association fitting on the business interaction lines to construct a business topological association network; The specific steps of step S3 are: Step S31: performing business interaction mining on the outlier correction business log to obtain business interaction lines; Step S32: Quantitatively calculate the business activity frequency and flow of the business interaction line to generate business line direct correlation strength data; Step S33: performing potential embedding association analysis on the business interaction lines to obtain a business line indirect association matrix; Step S34: performing topological correlation fitting on the outlier corrected business logs based on the business line direct correlation strength data and the business line indirect correlation matrix to construct a business topological correlation network; Step S4: correcting the service log based on the outliers to obtain the service attributes of each interaction line; using the service attributes of each interaction line to perform attribute space mapping on the service topology association network to obtain an attribute mapping topology network; The specific steps of step S4 are: Step S41: Calculate key business indicators for the outlier correction business log to obtain business indicator values; Step S42: performing business attribute analysis on the business interaction lines based on the business indicator values to obtain business attributes of each interaction line; Step S43: using the service attributes of each interaction line to perform attribute space mapping on the service topology association network to obtain an attribute mapping topology network; Step S5: simulate the dynamic topology evolution of the attribute mapping topology network to obtain the dynamic topology evolution network; analyze the path evolution correlation strength change of the dynamic topology evolution network to obtain the topology correlation strength change data; The specific steps of step S5 are: Step S51: performing a time-series interaction evolution analysis on the service interaction line to obtain a time-series evolution rule of the service line; Step S52: based on the time-series evolution law of the service line, a dynamic topology evolution simulation is performed on the attribute mapping topology network, thereby obtaining a dynamic topology evolution network; Step S53: Identify the service transmission trace of the dynamic topology evolving network to obtain the dynamic service transmission trace; Step S54: performing path evolution trend analysis on the dynamic service transmission trajectory to generate service trajectory evolution trend data; Step S55: analyzing the correlation strength change of the business trajectory evolution trend data, thereby obtaining topological correlation strength change data; Step S6: Perform three-dimensional visualization processing on the dynamic topology evolution network based on the topology association strength change data, and construct a three-dimensional visualization business evolution model; The specific steps of step S6 are: Step S61: predicting the evolution of business interactions in a dynamic topology evolution network based on the topology association strength change data, and building a business evolution prediction network; Step S62: Perform multi-dimensional spatial projection on the business evolution prediction network to generate a business evolution prediction space model; Step S63: Perform three-dimensional visualization processing on the business evolution prediction space model to construct a three-dimensional visualized business evolution model.
2. The business data visualization method for an enterprise platform according to claim 1, characterized in that: The specific steps of step S13 are: Perform time series analysis on historical business logs to generate business log sequences; Calculate the null value rate of the structured field to obtain the field null value rate; Perform frequency distribution analysis of null value fields on business log sequences based on field null value rates to generate frequency distribution data of null value fields; Based on the frequency distribution data of null value fields, semantic feature analysis is performed on the business log sequence to obtain log semantic features; Calculate the conditional probability transfer of adjacent semantic features on the log semantic features to generate a conditional probability transfer data matrix; Perform semantic feature association analysis on the conditional probability transfer data matrix to obtain the dependency relationship between neighboring semantic features; The global context semantics mining is performed on the business log sequence according to the dependency relationship between the semantic features of neighbors to generate the global context semantic features.
3. A business data visualization system for an enterprise platform, characterized in that: The method for performing business data visualization for an enterprise platform according to claim 1 comprises: The semantic graph module is used to obtain historical business logs based on the enterprise platform; the global semantic graph is constructed and processed for the historical business logs to obtain a semantic feature knowledge graph; An outlier correction module is used to perform multi-dimensional anomaly detection on historical business logs based on the semantic feature knowledge graph to generate outlier difference values; perform outlier correction processing on historical business logs based on the outlier difference values to obtain outlier-corrected business logs; The topological association module is used to perform business interaction mining on the outlier correction business logs to obtain business interaction lines; perform topological association fitting on the business interaction lines to build a business topological association network; An attribute space mapping module is used to correct the business log based on the outlier to obtain the business attributes of each interaction line; the business attributes of each interaction line are used to perform attribute space mapping on the business topology association network to obtain an attribute mapping topology network; The topology evolution module is used to simulate the dynamic topology evolution of the attribute mapping topology network, thereby obtaining the dynamic topology evolution network; analyze the change of path evolution correlation strength of the dynamic topology evolution network, thereby obtaining the topology correlation strength change data; The three-dimensional visualization module is used to perform three-dimensional visualization processing on the dynamic topology evolution network based on the data of the strength change of topology associations, and to build a three-dimensional visualization business evolution model.
Citation Information
Patent Citations
Knowledge base construction method and system based on cloud service
CN114168747A
Customer data management method and system based on big data
CN117575011A
Project declaration and enterprise information interaction method and system based on artificial intelligence
CN118195533A
Multidimensional structured data visualization method and apparatus, text visualization method and apparatus, method and apparatus for visualizing and graphically navigating the world wide web, method and apparatus for visualizing hierarchies
US20040090472A1
Cited By
Data intelligent visualization method and system for enterprise management
CN122570598A
A data intelligent visualization method and system for enterprise management
CN122570598B
