Data flow dashboard generation method and device, computer equipment and storage medium

Abstract

The application relates to a data flow dashboard generation method and device, computer equipment, a storage medium and a computer program product. The method is applied to the fields of cloud computing and intelligent transportation. The method comprises the following steps: acquiring a domain model graph of a business system and different data source entities; determining relationship data between various domain entities in the domain model graph according to a first representation vector of the domain model graph; determining data source entity relationships, data flow relationships and influence factors between the data source entities based on the relationship data and the data source entities; and generating a visual data flow dashboard corresponding to the business system based on the data source entity relationships, the data flow relationships and the influence factors. The method can effectively improve the quality of a data flow dashboard and generate a high-quality data flow dashboard.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to a method, apparatus, computer device, storage medium, and computer program product for generating a data flow dashboard. Background Technology

[0002] With the development of computer and internet technologies, data dashboards have become a crucial method in visual management, leading to widespread attention on data dashboard design. Data dashboards can intuitively reflect changes in business operations, helping decision-makers adjust and make informed decisions.

[0003] However, in current methods of generating data flow dashboards, developers typically need to configure the data source entities. In actual business development, the data dashboard is specified based on product or business requirements. Developers manually specify two or more dimensions based on their experience and then create data dashboards for different dimensions. Because the data source entities need to be manually configured, the relationships inherent in the data source entities themselves are ignored. This may result in redundant logic, high data collection overhead, and other issues, which can easily lead to poor quality of the final generated data flow dashboard. Summary of the Invention

[0004] Therefore, it is necessary to provide a method, apparatus, computer equipment, computer-readable storage medium, and computer program product for generating data flow dashboards that can effectively improve the quality of data flow dashboards, addressing the aforementioned technical problems.

[0005] Firstly, this application provides a method for generating a data flow dashboard. The method includes: acquiring a domain model diagram of a business system and different data source entities; determining relationship data between domain entities in the domain model diagram based on a first representation vector of the domain model diagram; determining data source entity relationships, data flow relationships, and influencing factors among the data source entities based on the relationship data and each of the data source entities; and generating a visual data flow dashboard corresponding to the business system based on the data source entity relationships, the data flow relationships, and the influencing factors.

[0006] Secondly, this application also provides a data flow dashboard generation apparatus. The apparatus includes: an acquisition module for acquiring a domain model graph of a business system and different data source entities; a determination module for determining relationship data between domain entities in the domain model graph based on a first representation vector of the domain model graph; determining data source entity relationships, data flow relationships, and influencing factors between the data source entities based on the relationship data and each of the data source entities; and a generation module for generating a visual data flow dashboard corresponding to the business system based on the data source entity relationships, the data flow relationships, and the influencing factors.

[0007] In one embodiment, the apparatus further includes: a processing module, configured to perform vectorization processing on the domain model graph to obtain a first representation vector of the domain model graph; and a determining module, configured to determine, based on the first representation vector, relationship data between a target domain entity and combinations of other domain entities in the domain model graph; or, based on the first representation vector, determine relationship data between various domain entities in the domain model graph.

[0008] In one embodiment, the apparatus further includes a selection module and an aggregation module. The acquisition module is further configured to acquire raw data corresponding to different data source entities; the selection module is configured to select target data from the raw data; the aggregation module is configured to aggregate the data source entity relationships, data flow relationships, and influencing factors between the various data source entities with the target data to obtain first aggregated data; the generation module is further configured to generate a visual data flow dashboard corresponding to the business system based on the first aggregated data.

[0009] In one embodiment, the acquisition module is further configured to acquire the original data corresponding to different data source entities; the aggregation module is further configured to aggregate the data source entity relationships, data flow relationships and influencing factors between the data source entities with the original data to obtain second aggregated data; the generation module is further configured to generate a visual data flow dashboard corresponding to the business system based on the second aggregated data.

[0010] In one embodiment, the generation module is further configured to use the aggregated data as configuration data for the platform backend, so that the platform frontend generates a visual data flow dashboard corresponding to the business system based on the configuration data; wherein, the aggregated data is the first aggregated data or the second aggregated data; the platform backend and the platform frontend are respectively the backend and frontend of the data flow dashboard platform.

[0011] In one embodiment, the apparatus further includes: a display module for displaying the data stream dashboard through the platform front end; the data stream dashboard includes a visual data chart and a download button; the processing module is further configured to download the visual data chart in response to a trigger operation of the download button; or, export the data stream dashboard.

[0012] In one embodiment, the processing module is further configured to process the relational data and the identification information of the data source entities using a deep neural network model to obtain the data source entity relationships and influence factors between each of the data source entities; the determining module is further configured to determine the data flow relationships between each of the data source entities based on the data source entity relationships between each of the data source entities, the influence factors, and the original data corresponding to different data source entities.

[0013] In one embodiment, the acquisition module is further configured to acquire the initial influence factors of each of the data source entities; the processing module is further configured to process the relational data, each of the data source entities, and the initial influence factors of each of the data source entities through the deep neural network model to obtain the data source entity relationships and influence factors between each of the data source entities.

[0014] In one embodiment, the selection module is further configured to select a target impact factor that meets the threshold condition from the impact factors; the acquisition module is further configured to acquire the data source entity relationship corresponding to the target impact factor, and use the data source entity relationship as the target data source entity relationship.

[0015] In one embodiment, the apparatus further includes a configuration module and a training module. The acquisition module is further configured to acquire entity samples from a data source and convert the entity samples into vector data; the aggregation module is further configured to aggregate the vector data according to different dimensions to obtain aggregated vector data; the configuration module is configured to configure the relationships between different aggregated vector data to obtain relationship data samples; and the training module is configured to train the deep neural network model based on the aggregated vector data and the relationship data samples.

[0016] In one embodiment, the acquisition module is further configured to acquire domain model graph samples of different business systems; the determination module is further configured to determine entity relationship data between domain entities in the domain model graph sample based on the second representation vector of the domain model graph sample; and the configuration module is further configured to configure the relationship between different aggregate vector data based on the entity relationship data to obtain a relationship data sample.

[0017] Thirdly, this application also provides a computer device. The computer device includes a memory and a processor. The memory stores a computer program, and the processor, when executing the computer program, performs the following steps: acquiring a domain model diagram of a business system and different data source entities; determining relationship data between domain entities in the domain model diagram based on a first representation vector of the domain model diagram; determining data source entity relationships, data flow relationships, and influencing factors between the data source entities based on the relationship data and each of the data source entities; and generating a visual data flow dashboard corresponding to the business system based on the data source entity relationships, the data flow relationships, and the influencing factors.

[0018] Fourthly, this application also provides a computer-readable storage medium. The computer-readable storage medium stores a computer program thereon, which, when executed by a processor, performs the following steps: acquiring a domain model graph of a business system and different data source entities; determining relationship data between domain entities in the domain model graph based on a first representation vector of the domain model graph; determining data source entity relationships, data flow relationships, and influencing factors between the data source entities based on the relationship data and each of the data source entities; and generating a visual data flow dashboard corresponding to the business system based on the data source entity relationships, the data flow relationships, and the influencing factors.

[0019] Fifthly, this application also provides a computer program product. The computer program product includes a computer program that, when executed by a processor, performs the following steps: acquiring a domain model graph of a business system and different data source entities; determining relationship data between domain entities in the domain model graph based on a first representation vector of the domain model graph; determining data source entity relationships, data flow relationships, and influencing factors between the data source entities based on the relationship data and each of the data source entities; and generating a visual data flow dashboard corresponding to the business system based on the data source entity relationships, the data flow relationships, and the influencing factors.

[0020] The aforementioned data flow dashboard generation method, apparatus, computer equipment, storage medium, and computer program product acquire the domain model diagram of the business system and different data source entities; determine the relationship data between domain entities in the domain model diagram based on the first representation vector of the domain model diagram; determine the data source entity relationships, data flow relationships, and influencing factors between the data source entities based on the relationship data and each data source entity; and generate a visualized data flow dashboard corresponding to the business system based on the data source entity relationships, data flow relationships, and influencing factors. Since the relationship data between domain entities is determined based on the domain model diagram of the business system, higher-quality data source entity relationships and data flow relationships can be determined based on the relationship data between domain entities and each data source entity. This expresses the influence of multiple data source entities implicit in multi-layered relationships and obtains more accurate influencing factors. This allows for the generation of higher-quality data flow dashboards based on data source entity relationships, data flow relationships, and influencing factors, reducing labor costs, minimizing redundant data, and effectively improving the quality of the data flow dashboard. Attached Figure Description

[0021] Figure 1 This is an application environment diagram of the data flow dashboard generation method in one embodiment;

[0022] Figure 2This is a flowchart illustrating a method for generating a data flow dashboard in one embodiment;

[0023] Figure 3 This is a schematic diagram of a domain model in one embodiment;

[0024] Figure 4 This is a schematic diagram of another domain model in one embodiment;

[0025] Figure 5 This is a schematic diagram illustrating how a deep neural network model processes input data in the input layer in one embodiment.

[0026] Figure 6 This is a flowchart illustrating the steps of configuring the relationships between different aggregated vector data in one embodiment to obtain relational data samples.

[0027] Figure 7 This is a schematic diagram of the overall architecture of a high-quality data flow dashboard method based on deep learning and domain models in one embodiment;

[0028] Figure 8 This is a domain model diagram of the data source entity of the palm-swiping system in one embodiment;

[0029] Figure 9 This is a schematic diagram of a multi-layer neural network structure for a subway traffic business scenario in one embodiment;

[0030] Figure 10 This is a schematic diagram of deep learning parameter tuning processing in one embodiment;

[0031] Figure 11 This is a structural block diagram of a data flow dashboard generation device in one embodiment;

[0032] Figure 12 This is an internal structural diagram of a computer device in one embodiment. Detailed Implementation

[0033] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.

[0034] The data stream dashboard generation method provided in this application embodiment can be applied to, for example, Figure 1In the application environment shown, terminal 102 communicates with server 104 via a network. A data storage system can store the data that server 104 needs to process. The data storage system can be integrated onto server 104 or placed on the cloud or other servers. Server 104 can obtain the domain model diagram of the business system and different entity data sources locally; server 104 can also obtain the domain model diagram of the business system and different entity data sources from terminal 102. Further, server 104 determines the relationship data between domain entities in the domain model diagram based on the first representation vector of the domain model diagram, and based on the relationship data and each data source entity, determines the data source entity relationships, data flow relationships, and influencing factors between each data source entity. Based on the data source entity relationships, data flow relationships, and influencing factors, server 104 generates a visual data flow dashboard corresponding to the business system.

[0035] The terminal 102 can be, but is not limited to, various desktop computers, laptops, smartphones, tablets, IoT devices, and portable wearable devices. IoT devices can include smart speakers, smart TVs, smart air conditioners, smart in-vehicle devices, etc. Portable wearable devices can include smartwatches, smart bracelets, head-mounted devices, etc.

[0036] Server 104 can be implemented using a standalone server or a server cluster consisting of multiple servers. It is understood that server 104 provided in this embodiment can also be a service node in a blockchain system. The service nodes in this blockchain system form a peer-to-peer (P2P) network. The P2P protocol is an application layer protocol running on top of the Transmission Control Protocol (TCP).

[0037] Cloud technology refers to a managed technology that unifies a series of resources such as hardware, software, and networks within a wide area network or local area network to achieve data computing, storage, processing, and sharing.

[0038] Cloud technology is a collective term for network technologies, information technologies, integration technologies, management platform technologies, and application technologies applied to the cloud computing business model. It can form resource pools, providing flexible and convenient on-demand access. Cloud computing technology will become a crucial support. Backend services of technical network systems require substantial computing and storage resources, such as video websites, image websites, and many portal websites. With the rapid development and application of the internet industry, every item may have its own identification mark in the future, requiring transmission to backend systems for logical processing. Data at different levels will be processed separately, and various industry data will all require robust system support, which can only be achieved through cloud computing.

[0039] Big data refers to data sets that cannot be captured, managed, and processed within a certain timeframe using conventional software tools. It represents massive, rapidly growing, and diverse information assets that require new processing models to achieve stronger decision-making, insightful discovery, and process optimization capabilities. With the advent of the cloud era, big data has attracted increasing attention. Big data requires specialized technologies to effectively process large amounts of data within a tolerable timeframe. Technologies suitable for big data include massively parallel processing databases, data mining, distributed file systems, distributed databases, cloud computing platforms, the internet, and scalable storage systems.

[0040] Artificial intelligence (AI) is the theory, methods, technology, and application systems that use digital computers or machines controlled by digital computers to simulate, extend, and expand human intelligence, perceive the environment, acquire knowledge, and use that knowledge to achieve optimal results. In other words, AI is a comprehensive technology within computer science that attempts to understand the essence of intelligence and produce a new kind of intelligent machine that can react in a way similar to human intelligence. AI studies the design principles and implementation methods of various intelligent machines, enabling them to possess the functions of perception, reasoning, and decision-making.

[0041] Artificial intelligence (AI) is a comprehensive discipline encompassing a wide range of fields, including both hardware and software technologies. Fundamental AI technologies generally include sensors, dedicated AI chips, cloud computing, distributed storage, big data processing, operating / interactive systems, and mechatronics. AI software technologies primarily include computer vision, speech processing, natural language processing, as well as machine learning / deep learning, autonomous driving, and intelligent transportation.

[0042] Computer vision (CV) is a science that studies how to enable machines to "see." More specifically, it refers to machine vision, which uses cameras and computers to replace human eyes in recognizing and measuring targets, and then performs image processing to create images more suitable for human observation or transmission to instruments. As a scientific discipline, computer vision studies related theories and technologies, attempting to build artificial intelligence systems capable of extracting information from images or multidimensional data. Computer vision technologies typically include image processing, image recognition, image semantic understanding, image retrieval, OCR, video processing, video semantic understanding, video content / behavior recognition, 3D object reconstruction, 3D technology, virtual reality, augmented reality, simultaneous localization and mapping (SLAM), autonomous driving, intelligent transportation, and common biometric recognition technologies such as facial recognition and fingerprint recognition.

[0043] Machine Learning (ML) is a multidisciplinary field involving probability theory, statistics, approximation theory, convex analysis, and algorithm complexity theory. It specifically studies how computers can simulate or implement human learning behavior to acquire new knowledge or skills and reorganize existing knowledge structures to continuously improve their performance. Machine learning is the core of artificial intelligence and the fundamental way to endow computers with intelligence; its applications span all areas of artificial intelligence. Machine learning and deep learning typically include techniques such as artificial neural networks, belief networks, reinforcement learning, transfer learning, inductive learning, and instructional learning.

[0044] Deep learning is a branch of machine learning that attempts to perform high-level abstractions of data using multiple processing layers containing complex structures or consisting of multiple nonlinear transformations. Deep learning is an algorithm within machine learning based on representation learning of data. Several deep learning frameworks, such as convolutional neural networks, deep belief networks, and recurrent neural networks, have been applied in fields such as computer vision, speech recognition, natural language processing, audio recognition, and bioinformatics, achieving excellent results.

[0045] In one embodiment, such as Figure 2 As shown, a method for generating a data flow dashboard is provided, which can be applied to... Figure 1 Taking the server in the example, the following steps are included:

[0046] Step 202: Obtain the domain model diagram of the business system and different data source entities.

[0047] In this context, a business system refers to the business processes required for an enterprise to achieve its positioning, the roles played by various partners, and the methods and content of cooperation and transactions among stakeholders. Business systems can include different types of systems, meaning different business systems correspond to different business functions. For example, a business system could include one for online shopping or one for subway transportation scenarios.

[0048] A domain model is a visual representation of conceptual classes or real-world objects within a domain. It is also known as a conceptual model, domain object model, or analysis object model. It focuses on analyzing the problem domain itself, uncovering important business domain concepts, and establishing relationships between these concepts. For example, domain models include class diagrams and sequence diagrams.

[0049] Domain model diagrams refer to UML diagrams. UML stands for Unified Modeling Language, also known as the standard modeling language. It is a language used for visual modeling of software-intensive systems. The definition of UML includes two elements: UML semantics and UML notation. Unified Modeling Language (UML) is a modeling language, and models are mostly represented in the form of diagrams.

[0050] An entity represents a discrete object. An entity can be (roughly) thought of as a noun, such as a computer, an employee, a song, or a mathematical theorem. A relation describes how two or more entities are related to each other.

[0051] A data source entity refers to an object that generates data in a business context. In this application, the data source entity in the business system can include multiple different data source entities. For example, a car generates data such as car price and car quality. In this case, the entity and the data source entity are both cars, and the car price and car quality are data generated by the car entity.

[0052] Specifically, the server can obtain domain model diagrams corresponding to different business systems from the local database. At the same time, the server can also directly obtain multiple data source entities from a certain business system from the local database.

[0053] For example, let's take a subway transportation scenario as an example. A palm-scanning system is a system that uses palmprint authentication to obtain the corresponding permissions to perform various tasks. This type of system can serve a large number of business systems. For the large amount of continuously generated data in the subway transportation scenario, a data flow dashboard needs to be established for real-time management at the business control level. The subway transportation scenario's business can use the palm-scanning system to monitor the data flow of passengers traveling at different times. During business development, developers need to monitor a large amount of data flow, including but not limited to data returned from the subway gate, such as information on currently entering passengers, entry speed, and whether passengers choose palm scanning or subway cards. The user device receives returned data, such as palmprint recognition status, recognition speed, and whether there are any false positives. Therefore, the server can obtain the domain model diagram corresponding to the palm-scanning system in the subway transportation scenario. Simultaneously, the server can also obtain different data source entities in the palm-scanning system within the subway transportation scenario, such as passengers and subway gates.

[0054] Step 204: Determine the relationship data between domain entities in the domain model graph based on the first representation vector of the domain model graph.

[0055] The first representation vector refers to the vector used to describe the domain model diagram. Since the domain model diagram is compiled by developers for each business system, the first representation vectors obtained from different domain model diagrams may also be different.

[0056] Domain entities refer to the entities described in the domain model diagram. Domain class entities are a type of entity in the domain model, typically including entity classes, control classes, and boundary classes, which bear domain logic. In this application, domain entities mainly represent data source entities. For example, if a domain model diagram A contains Entity1 and Entity2, then Entity1 and Entity2 are both domain entities. The relationship data between domain entities is used to reflect the relationship between Entity1 and Entity2. For example, the entity relationship triplet extracted from domain model diagram A is {Entity1, Entity2, generalization}.

[0057] Relational data refers to data used to express the relationships between entities in a domain. For example, relational data can be {Entity1, Entity2, generalization}, which reflects that the relationship between Entity1 and Entity2 is generalization.

[0058] Specifically, after the server obtains the domain model diagram of the business system and different entity data sources, it can determine the relationship data between various domain entities in the domain model diagram based on the first representation vector of the diagram. In other words, the server can transform the domain model diagram into a computable representation vector and determine the relationships between various domain entities based on this vector. For example, ... Figure 3 The image shown is a schematic diagram of a domain model. Domain models are typically expressed using UML diagrams. UML diagrams can be transformed into vectors describing the graph using rules; that is, a UML diagram is a tool for describing entity relationships. Servers can then use diagrams such as... Figure 3 The domain model graph shown is transformed into entity relation triples: {Entity1, Entity2, generalization}. For example... Figure 4 The image shown is a schematic diagram of another domain model. The server can access data from sources such as... Figure 4 The domain model diagram shown extracts entity relationship triples as follows: {Entity1, Entity2, Entity3, one-way association}, {Entity2, Entity3, variable}, {Entity1, Entity2, member variable}, {Entity1, Entity3, member variable}. Alternatively, the server can further extract from... Figure 4 The relationship between entities and entity combinations in the domain model diagram shown is: {Entity1, {Entity2, Entity3}, unidirectional association}.

[0059] Step 206: Based on the relational data and each data source entity, determine the data source entity relationships, data flow relationships, and influencing factors among the data source entities.

[0060] Among them, the data source entity relationship refers to the association relationship between data source entities. For example, the association relationship includes, but is not limited to, one-way association, generalization, combination, derivation and aggregation.

[0061] A data stream is an ordered sequence of bytes with a start and an end, including input and output streams. A data stream is defined as a sequence of data read once in a prescribed order. The data stream relationship in this application refers to a formal description of the data propagation path within the system, i.e., a formal description of the data stream. For example, the data stream relationship in this application can be represented in vector form, and updates to the components of the vector appear as a stream of binary tuples.

[0062] The influence factor is a parameter used to reflect the relationship between entities in a data source. The larger the value of the influence factor, the greater its weight. For example, assuming that the five relationship bits between TableInfo and RouteKey are: one-way relationship, generalization relationship, combination relationship, derivation relationship, and aggregation relationship, from (TableInfo, RouteKey, 0.9, 0, 0, 0, 0, 1, 2, 0), we know that the influence factors corresponding to the five relationship bits are (0.9, 0, 0, 0, 0). Assuming that the preset threshold is 0.9, since the influence factor of the one-way relationship is 0.9, which meets the preset threshold, we can determine that the relationship between tableInfo and routeKey is a one-way relationship.

[0063] Specifically, after determining the relationship data between domain entities in the domain model graph based on the first representation vector, the server can input this relationship data and various data source entities into a pre-trained deep neural network model. After processing by the deep neural network model, the output includes the data source entity relationships and influencing factors between the data source entities. Furthermore, the server can obtain the original data corresponding to different data source entities in the business system and, based on the data source entity relationships, influencing factors, and original data corresponding to different data source entities output by the deep neural network model, determine the data flow relationships between different data source entities. In other words, the server can use the relevant domain entity relationships extracted from the domain model graph as parameter tuning factors for the deep neural network model, enabling the model to discover the optimal relationships between data source entities.

[0064] It can be understood that the pre-trained deep neural network model in this application is essentially a classifier with a nonlinear decision boundary. The input data is the product of the input vector and the vector weights. The output data is obtained by calculating the loss value in the forward direction and adjusting the gradient parameters through the backpropagation method. That is, the pre-trained deep neural network model in this application can obtain the optimal model parameters through pre-training with labeled data.

[0065] For example, suppose the input data contains 2 entity bits, 5 relation bits, 2 quantity bits, and 1 additional bit. The 5 relation bits are: one-way association, generalization, composition, derivation, and aggregation. The server can process the following data:

[0066] (Repository,TableInfo,1,0,0,2,1,1,1,1)

[0067] (TableInfo,RouteKey,0,0,0,2,1,1,2,1)

[0068] (Repository,RouteKey,0,0,0,2,1,1,5,1)

[0069] (TableInfo,Field,1,0,0,2,1,1,7,1)

[0070] (Repository,Field,1,0,1,2,1,1,9,1)

[0071] The input data is fed into a pre-trained deep neural network model. After processing by the deep neural network model, the output data is obtained as follows:

[0072] (Repository,TableInfo,0.9,0,0.1,0,1.2,1,1,1)

[0073] (TableInfo,RouteKey,0.2,0,0,0,0,1,2,1)

[0074] (Repository,RouteKey,0.3,0,0,0,0.4,1,5,1)

[0075] (TableInfo,Field,1.3,0,0.5,0.2,0,1,7,1)

[0076] (Repository,Field,1.1,0.1,0.6,2.3,1.7,1,9,1)

[0077] Its actual function is:

[0078] (Repository,TableInfo,0.9,0,0.1,0,1.2,1,1,0): The recommended relationship between Repository and tableInfo is a unidirectional association, with a one-to-one quantitative relationship;

[0079] (TableInfo,RouteKey,0.9,0,0,0,0,1,2,0): The association between tableInfo and routeKey is unidirectional, with a 1-to-2 relationship.

[0080] (Repository,RouteKey,0.3,0,0,0,0.4,1,5,0): The relationship between Repository and routeKey is unrelated. It is assumed that the threshold at each position is not met, so no generation is performed.

[0081] (TableInfo,Field,0.3,0,0.1,0.9,0,1,7,0): The association between tableInfo and field is aggregation;

[0082] (Repository,Field,0,9.1,0.6,0.3,0,1,9,0): The relationship between Repository and field is a one-way association.

[0083] Step 208: Based on the data source entity relationships, data flow relationships, and influencing factors, generate a visual data flow dashboard corresponding to the business system.

[0084] Among them, a data flow dashboard refers to a data monitoring platform that displays data changes in real time. For example, the display interface of the data flow dashboard contains information such as the entities of the monitored data sources, the monitored events, and the impact between the monitored data sources. The data flow dashboard plays an important role in the collection and control of information on business operations.

[0085] A visual data flow dashboard refers to displaying monitored data in a visual way, such as through visual charts.

[0086] Specifically, after determining the data source entity relationships, data flow relationships, and influencing factors among the data source entities based on relational data and each data source entity, the server can generate a visual data flow dashboard corresponding to the business system based on the data source entity relationships, data flow relationships, and influencing factors. That is, the data source entity relationships in this application are used to guide the generation of the data dashboard. Since there is a one-to-one correspondence between the domain entities in the domain model diagram and the data source entities, it is necessary to obtain good domain entity relationships in the domain model diagram in order to obtain more accurate data source entity relationships.

[0087] In this application, a deep neural network model is trained by selecting entity relationships from a good domain model diagram as a training set. This allows the pre-trained deep neural network model to determine theoretically better domain entity relationships in the domain model when a new business system's UML diagram is subsequently passed in. These relationships are then converted into data source entity relationships, resulting in more accurate and strongly correlated relationships between data source entities. This can serve as a guide for building data flow dashboards.

[0088] For example, let's take a swipe-based payment system used in a subway transportation scenario as an example. The server can obtain the domain model graph A and multiple data source entities of the swipe-based payment system used in the subway transportation scenario, and determine the relationship data between the domain entities in the domain model graph A based on the first representation vector of the domain model graph A. Furthermore, the server can use the relationship data between the domain entities and the data source entities obtained above as input layer data, input into a pre-trained deep neural network model, and after processing by the deep neural network model, output the data source entity relationships and influencing factors between the data source entities.

[0089] Assuming the data source entity relationships and influencing factors output by the deep neural network model are (T1, S1, 0.9, 0, 0, 0, 0, 1, 2, 0), representing a unidirectional association between T1 and S1 with a 1-to-2 quantitative relationship, where T1 represents the clearance speed and S1 represents the recognition speed, the server can obtain the raw data corresponding to data source entities T1 and S1. For example, the raw data for data source entity T1 at time t might be 1 minute, and the raw data for data source entity S1 at time t might be 2 seconds. Furthermore, based on the obtained raw data for data source entities T1 and S1, the data source entity relationships and influencing factors output by the deep neural network model, the server can determine the data flow relationships between each data source entity. The server can then aggregate the data flow relationships between each data source entity, the raw data for data source entities T1 and S1, and the data source entity relationships and influencing factors output by the deep neural network model to obtain aggregated data. Based on this aggregated data, a visualized data flow dashboard corresponding to the swipe-based system used in subway transportation scenarios can be generated.

[0090] In this embodiment, the domain model diagram of the business system and different data source entities are obtained; the relationship data between each domain entity in the domain model diagram is determined according to the first representation vector of the domain model diagram; based on the relationship data and each data source entity, the data source entity relationships, data flow relationships, and influencing factors between each data source entity are determined; and based on the data source entity relationships, data flow relationships, and influencing factors, a visual data flow dashboard corresponding to the business system is generated. Since the relationship data between domain entities is determined based on the domain model diagram of the business system, higher quality data source entity relationships and data flow relationships can be determined based on the relationship data between domain entities and each data source entity. This expresses the influence of multiple data source entities implicit in multi-level relationships and obtains more accurate influencing factors. This allows for the generation of higher quality data flow dashboards based on data source entity relationships, data flow relationships, and influencing factors, reducing labor costs, reducing redundant data and time overhead for intermediate data collection, and effectively improving the quality and efficiency of data flow dashboard generation.

[0091] In one embodiment, the step of determining the relationship data between domain entities in the domain model graph based on the first representation vector of the domain model graph includes:

[0092] Vectorize the domain model graph to obtain the first representation vector of the domain model graph;

[0093] Based on the first representation vector, determine the relationship data between the target domain entity and combinations of other domain entities in the domain model graph; or...

[0094] Based on the first representation vector, determine the relationship data between entities in each domain in the domain model graph.

[0095] In this context, the target domain entity refers to one or more domain entities selected from a pool of domain entities, while the combination of other domain entities refers to the combination of other domain entities in the domain model diagram besides the target domain entity. For example, if a business system's domain model diagram contains three domain entities: Entity1, Entity2, and Entity3, and Entity1 is selected as the target domain entity, then the combination of Entity2 and Entity3 can be considered as the combination of other domain entities, i.e., the combination of other domain entities is {Entity2, Entity3}.

[0096] Specifically, after obtaining the domain model graph of the business system, the server can vectorize the domain model graph using the domain model module to obtain a first representation vector. Based on the first representation vector, the server determines the relationship data between the target domain entity and other domain entities in the domain model graph; or, the server determines the relationship data between various domain entities in the domain model graph based on the first representation vector. In other words, the server in this application can use the domain model module to convert the domain model graph into a computable vector form of data to parse out the relationship data used to express the relationships between entities.

[0097] It is understood that the domain model module in this application can be used to parse the domain model diagram of a business system. A domain model diagram can describe a wide range of content. The domain model diagram used in this embodiment only retains the relationships between data source entities; that is, this embodiment only involves the data source entities and does not consider entities that are not data source entities. For example, cars generate data, but car manufacturers do not. There is a corresponding relationship between cars and car manufacturers, and car manufacturers are also represented in the domain model diagram. However, when the server extracts the data source entity relationships from the domain model diagram, it does not consider car manufacturers. This rule can be predefined by the user.

[0098] For example, such as Figure 3 The image shown is a schematic diagram of a domain model. The server can handle, for example... Figure 3The domain model graph shown is vectorized to obtain the first representation vector of the domain model graph. Based on the first representation vector, the entity relationship triple in the domain model graph is determined to be: {Entity1, Entity2, generalization}. That is, the server determines that the relationship between Entity1 and Entity2 is generalization.

[0099] like Figure 4 The image shown is a schematic diagram of another domain model. The server can handle, for example... Figure 4 The domain model graph shown is vectorized to obtain its first representation vector. Based on this first representation vector, the relationship data between the target domain entity and other domain entities in the domain model graph is determined as: {Entity1, {Entity2, Entity3}, unidirectional association}. That is, the server determines that the relationship between the target domain entity Entity1 and the combination of other domain entities {Entity2, Entity3} is a unidirectional association. Therefore, by leveraging past experience with data source entities, such as the domain model graph of the business system and manually labeled known influence parameters between data source entities, higher-quality data source entity relationships and data flow relationships can be obtained. These relationships express the influence of multiple entities implicit in multi-layered relationships, providing more accurate relational data for subsequently generating higher-quality data flow dashboards.

[0100] In one embodiment, the method further includes:

[0101] Obtain the original data corresponding to different data source entities;

[0102] Select the target data from the original data, and aggregate the data source entity relationships, data flow relationships, and influencing factors between the data source entities with the target data to obtain the first aggregated data;

[0103] The step of generating a visual data flow dashboard corresponding to the business system based on the data source entity relationship, the data flow relationship, and the influencing factors includes:

[0104] Based on the first aggregated data, a visual data flow dashboard corresponding to the business system is generated.

[0105] Here, raw data refers to the specific data value corresponding to the data source entity. For example, if the data source entity is Entity2, the raw data corresponding to Entity2 is 2s.

[0106] The first aggregated data refers to the data obtained by aggregating the relationships between data source entities and the original data.

[0107] Specifically, based on the relationship data between domain entities in the domain model graph and the data source entities, the server determines the data source entity relationships, data flow relationships, and influencing factors among the data source entities. The server can then obtain the original data corresponding to different data source entities, select a portion of the original data as target data, and aggregate the data source entity relationships, data flow relationships, and influencing factors among the data source entities with the target data to obtain the first aggregated data. The first aggregated data contains both the relationships between data source entities and the original business data corresponding to the data source entities. Therefore, the server can generate a visual data flow dashboard corresponding to the business system based on the first aggregated data.

[0108] For example, suppose the server, based on the relationship data between domain entities in the domain model graph and the entities of each data source, determines the data source entity relationships and influencing factors as: (Repository, Field, 0, 9.1, 0.6, 0.3, 0, 1, 9, 0), representing a unidirectional association between Repository and field. Then the server can obtain the original data corresponding to the data source entities Repository and field respectively. Assume the server obtains the original data corresponding to the data source entities Repository and field as {A1, A...}. 2. Given the original data {A1, A2, A3, B1, B2, B3, B4}, the server can select the target data {A1, A2, B1, B2} from the original data {A3, B1, B2, B3, B4} according to preset rules. It then aggregates the data source entity relationships, data flow relationships, and influencing factors between each data source entity with the target data {A1, A2, B1, B2} to obtain aggregated data S. Aggregated data S contains data source entity relationships, data flow relationships, influencing factors, and the original business data. Therefore, the server can generate a visual data flow dashboard corresponding to the business system based on aggregated data S. This enables the provision of higher-quality data flow dashboards for business development, reduces labor costs, minimizes redundant data, improves the maintainability of data flow dashboards, and more effectively empowers business exploration.

[0109] In one embodiment, the method further includes:

[0110] Obtain the original data corresponding to different data source entities;

[0111] The data source entity relationships, data flow relationships, and influencing factors between each data source entity are aggregated with the original data to obtain the second aggregated data;

[0112] The step of generating a visual data flow dashboard corresponding to the business system based on the data source entity relationship, the data flow relationship, and the influencing factors includes:

[0113] Based on the second aggregated data, a visual data flow dashboard corresponding to the business system is generated.

[0114] The second aggregated data is used to distinguish it from the second aggregated data, representing another aggregated data.

[0115] Specifically, based on the relationship data between domain entities in the domain model graph and the data source entities, the server determines the data source entity relationships, data flow relationships, and influencing factors among the data source entities. The server can then obtain the original data corresponding to different data source entities and aggregate the data source entity relationships, data flow relationships, and influencing factors among the data source entities with the original data to obtain the second aggregated data. The second aggregated data contains both the relationships between data source entities and the original business data corresponding to the data source entities. Therefore, the server can generate a visual data flow dashboard corresponding to the business system based on the second aggregated data.

[0116] For example, suppose the server determines the data source entity relationships and influencing factors between data source entities based on the relationship data between domain entities in the domain model graph and the data source entities: (Repository, Field, 0, 9.1, 0.6, 0.3, 0, 1, 9, 0), representing a unidirectional association between Repository and field. The server can then obtain the original data corresponding to the data source entities Repository and field respectively. Suppose the server obtains the original data corresponding to the data source entities Repository and field as {A1, A2, A3, B1, B2, B3, B4}. The server can aggregate the data source entity relationships, data flow relationships, and influencing factors between the data source entities with the original data {A1, A2, A3, B1, B2, B3, B4} to obtain aggregated data S1. Aggregated data S1 contains the data source entity relationships, data flow relationships, influencing factors, and original business data. Therefore, the server can generate a visual data flow dashboard corresponding to the business system based on aggregated data S1. This enables the provision of higher-quality data flow dashboards for business development, reduces labor costs, minimizes redundant data, improves the maintainability of data flow dashboards, and more effectively empowers business exploration.

[0117] In one embodiment, the step of generating a visual data flow dashboard corresponding to the business system based on data source entity relationships, data flow relationships, and influencing factors includes:

[0118] The aggregated data is used as the configuration data for the platform backend, so that the platform frontend can generate a visual data flow dashboard corresponding to the business system based on the configuration data; wherein, the aggregated data is either the first aggregated data or the second aggregated data; the platform backend and the platform frontend are the backend and frontend of the data flow dashboard platform, respectively.

[0119] Aggregated data refers to data obtained by aggregating the relationships between data source entities and the original data. For example, the aggregated data in the embodiments of this application may include the aforementioned first aggregated data and second aggregated data.

[0120] Specifically, based on relational data and the entities of each data source, after determining the data source entity relationships, data flow relationships, and influencing factors among the data source entities, the server can obtain the original data corresponding to different data source entities, select the target data from the original data, and aggregate the data source entity relationships, data flow relationships, and influencing factors among the data source entities with the target data to obtain the first aggregated data. Furthermore, the server can use the first aggregated data as configuration data for the platform backend, enabling the platform frontend to generate a visual data flow dashboard corresponding to the business system based on the configuration data; where the platform backend and platform frontend are the backend and frontend of the data flow dashboard platform, respectively.

[0121] In addition, after the server obtains the original data corresponding to different data source entities, the server can also aggregate the data source entity relationships, data flow relationships and influencing factors between the data source entities with the original data to obtain the second aggregated data. The server can use the second aggregated data as the configuration data of the platform backend so that the platform frontend can generate a visual data flow dashboard corresponding to the business system based on the configuration data.

[0122] In this embodiment, by aggregating the data source entity relationships, data flow relationships, and influencing factors between various data source entities with different original data, different aggregated data can be obtained. Therefore, the different aggregated data can be used as the configuration data of the platform backend, so that the platform frontend can generate a visual data flow dashboard that meets business needs or product requirements based on the configuration data, so that the data flow dashboard can express the data flow situation that is more in line with business needs or product requirements.

[0123] In one embodiment, the method further includes:

[0124] The platform front-end displays a data flow dashboard; the data flow dashboard includes visual data charts and a download button.

[0125] In response to the download button, download the visualized data chart; or export the data stream dashboard.

[0126] Among them, the platform front-end refers to the front-end of the data flow dashboard platform. The front-end in this application is mainly used to generate visual data curves and provide functions such as downloading Excel reports based on the configuration data generated by the back-end.

[0127] Specifically, when the server, i.e., the data flow dashboard platform, generates a visual data flow dashboard corresponding to the business system based on data source entity relationships, data flow relationships, and influencing factors, the server displays the data flow dashboard through the platform's front end. The data flow dashboard displayed on the interface includes visual data charts and a download button. Furthermore, in response to the user's trigger action of the download button, the server downloads the visual data charts currently displayed on the interface; or, it exports the data flow dashboard currently displayed on the interface. The user's trigger action can include various actions, such as clicks, swipes, long presses, and shakes.

[0128] For example, the server displays a data flow dashboard through the platform's front end. Let's assume the data flow dashboard displayed on the interface contains line charts representing the visual data. Figure 1 and sector Figure 2 When developers need to download a specific data visualization displayed in the current data flow dashboard, they can double-click the line. Figure 1 If the icon is selected, the server responds to the developer's double-click action by downloading the polyline currently displayed on the screen. Figure 1 This enables the provision of higher-quality data flow dashboards for business development, reduces labor costs, minimizes redundant data, improves the maintainability of data flow dashboards, and empowers developers to more effectively explore business needs.

[0129] In one embodiment, the steps of determining the data source entity relationships, data flow relationships, and influencing factors among the data source entities based on relational data and each data source entity include:

[0130] By processing the identification information of relational data and data source entities through a deep neural network model, the data source entity relationships and influencing factors between each data source entity are obtained.

[0131] Based on the relationships between data source entities, influencing factors, and the original data corresponding to different data source entities, the data flow relationships between data source entities are determined.

[0132] The identification information refers to the information used to uniquely identify the data source entity. The identification information can be custom-defined. For example, the name of the data source entity can be used as the identification information of the data source entity. Assuming the name of the data source entity is Repository, then the identification information of the data source entity is Repository.

[0133] Specifically, when the server determines the data source entity relationships, data flow relationships, and influencing factors between data source entities based on relational data and each data source entity, the server can process the relational data and the identification information of the data source entities through a pre-trained deep neural network model to obtain the data source entity relationships and influencing factors between each data source entity. Furthermore, the server can obtain the original data corresponding to different data source entities and determine the data flow relationships between each data source entity based on the data source entity relationships, influencing factors, and the original data corresponding to different data source entities.

[0134] For example, suppose the relationship data between domain entities and the data source entities in the domain model graph are: (Repository, key-value table information, direct association, 1, 1, derived). The server inputs the above relationship data and the identifier information of the data source entities into a pre-trained deep neural network model, i.e., the input data is: (Repository, TableInfo, 1, 0, 0, 2, 1, 1, 1, 1). After processing by the deep neural network model, the output is the data source entity relationship and influence factor between the data source entities: (Repository, TableInfo, 0.9, 0, 0.1, 0, 1.2, 1, 1, 1). This output data shows that the relationship between Repository and tableInfo is a unidirectional association, a one-to-one quantitative relationship.

[0135] Furthermore, the server can retrieve the raw data corresponding to the Repository and tableInfo. This raw data is displayed in JSON format, and typically requires fields such as the data source entity name, time, and business fields. Based on the data source entity relationships between the Repository and tableInfo, the influencing factors, and the corresponding raw data, the server determines the data flow relationships between the Repository and tableInfo. This allows for the generation of higher-quality data source entity relationships and data flow relationships by leveraging past experience with data source entities—such as the domain model diagram of the business system and manually labeled known influence parameters between data source entities. These relationships express the influence of multiple entities implicit in multi-layered relationships and obtain more accurate influence parameters.

[0136] In one embodiment, the step of processing relational data and identifier information of data source entities using a deep neural network model to obtain the data source entity relationships and influencing factors among various data source entities includes:

[0137] Obtain the initial impact factor for each data source entity;

[0138] By processing relational data, entities from various data sources, and their initial influence factors using a deep neural network model, the relationships and influence factors between entities from various data sources are obtained.

[0139] The initial impact factor refers to the initial weight parameter set for different data source entities. For example, the initial impact factor can be set to a fixed value or an empirical value.

[0140] Specifically, the server processes the relational data and the identification information of the data source entities through a pre-trained deep neural network module to obtain the data source entity relationships and influencing factors between each data source entity. The pre-trained deep neural network module in this embodiment is a typical neural network structure for performing deep learning, which can be divided into an input layer, multiple hidden layers, and an output layer. Internal details include neurons, activation functions, etc. The pre-trained deep neural network module is used to mine the relationships between data source entities and the influencing factors between data streams from the data streams of the data source entities, and uses the mined relationships and influencing factors as parameters configured in the backend of the data dashboard low-code generation module. The input data of the data source entity neurons is a vector converted from business data, and an initial weight, i.e., an initial influencing factor, can be pre-set for different input vectors. To adjust the influence of the weights on the final data of the neurons, the entire pre-trained deep neural network module is tuned based on the backpropagation technique of deep learning. For example, the composition formula of the neurons used in the swipe palm business system is: multi-dimensional swipe palm system data vector + weight + activation function ReLU + data hidden relation vector output to the hidden layer. The activation function ReLU can be a linearly rectified function with leakage, i.e., Parametric ReLU, where the parameter α is a learnable variable.

[0141] For example, the server can convert the data source entities Repository and tableInfo into vector data, obtaining data source entity vectors X1 and X2; furthermore, the server can obtain the initial influence factors w for the data source entities Repository and tableInfo, respectively. i1 and w i2 The server will store the relationship data between entities in the domain model graph, the data source entity vectors X1 and X2, and the initial influence factor w. i1 and w i2 As input layer data, the data is fed into the pre-trained deep neural network module. The pre-trained deep neural network module then processes the relationship data between entities in the domain model graph, the data source entity vectors X1 and X2, and the initial influence factor w. i1 and w i2After processing, the output yields the data source entity relationship between the data source entities Repository and tableInfo, and the influence factors wo1 and wo2. It can be understood that in this embodiment, the final output influence factors wo1 and wo2 are obtained by using a deep neural network model to process the initial input influence factor w. i1 and w i2 After adjustments, a more accurate impact factor was obtained.

[0142] In this embodiment, by adjusting the initial input influence factor through a deep neural network model, a more accurate influence factor can be obtained, enabling the data flow dashboard low-code generation platform to generate high-quality data flow dashboards based on data source entity relationships, data flow relationships, and a more accurate influence factor.

[0143] In one embodiment, the method further includes:

[0144] Select target impact factors that meet the threshold conditions from among the impact factors;

[0145] Obtain the data source entity relationship corresponding to the target impact factor, and use the data source entity relationship as the target data source entity relationship.

[0146] The target impact factor refers to the impact factor that meets the threshold condition selected from multiple candidate impact factors. For example, if the candidate impact factors include (0.9,0,0,0,0.3), then the candidate impact factor 0.9 that meets the threshold condition can be selected as the target impact factor.

[0147] Specifically, the server can obtain the initial impact factors of each data source entity, and use the relationship data between domain entities extracted from the domain model graph, the data source entities, and their initial impact factors as input data into a pre-trained deep neural network model. After processing by the deep neural network model, the server obtains the data source entity relationships and impact factors between each data source entity. Furthermore, the server can select target impact factors that meet a threshold condition from among the various impact factors. For example, assuming the threshold condition is: impact factor greater than or equal to 0.9, the server can select impact factors greater than or equal to 0.9 as target impact factors, obtain the data source entity relationships corresponding to the target impact factors, and use these data source entity relationships as target data source entity relationships.

[0148] For example, assuming the threshold condition is set to an impact factor greater than or equal to 0.9, the deep neural network model outputs multiple candidate impact factors as follows: (Repository, TableInfo, 0.9, 0, 0.1, 0, 1.2, 1, 1, 1). The first three digits represent the impact factor of the relationship position. Since 0.9 meets the threshold condition, the server can select 0.9 as the target impact factor from the multiple candidate impact factors (0.9, 0, 0.1), and obtain the data source entity relationship corresponding to the target impact factor 0.9 as a one-way association. This one-way association is then used as the data source entity relationship between Repository and TableInfo. This allows for a more direct and accurate determination of the relationship between data source entities based on the impact factor, providing higher-quality data flow dashboards for business development, reducing labor costs, minimizing redundant data and time overhead in intermediate data collection, effectively improving the generation efficiency of data flow dashboards, and more effectively empowering business exploration.

[0149] In one embodiment, the data source entity relationships and influence factors are obtained by processing the relationship data and each data source entity using a deep neural network model; the method further includes:

[0150] Obtain entity samples from the data source and convert them into vector data;

[0151] The vector data is aggregated according to different dimensions to obtain aggregated vector data.

[0152] Configure the relationships between different aggregated vector data to obtain a sample of relational data.

[0153] A deep neural network model is trained based on aggregated vector data and relational data samples.

[0154] Vector data refers to data in vector form. For example, if the data source entity sample is in JSON format, the data source entity sample will be converted into vector form.

[0155] Specifically, when training the initial deep neural network model, the server can acquire entity samples from the data source and convert them into vector data. Since each entity class from the data source can contain entities of different dimensions, the server can aggregate the vector data of different dimensions corresponding to each entity from the data source to obtain aggregated vector data. Furthermore, the server can configure the relationships and initial influencing factors between different aggregated vector data, that is, the server can configure the relationships and initial weights between different entity classes from the data source in the training set to obtain relational data samples. The server trains the deep neural network model based on the aggregated vector data and relational data samples.

[0156] For example, let's take the palm-swiping system used in subway transportation as an example. Figure 5 The diagram illustrates how a deep neural network model processes input data at its input layer. During the initial training of the deep neural network model, the server can obtain entity samples from multiple data sources within the swipe-to-pay business system and convert these samples into vector data. The resulting vector data includes: X 11 X 12 X 21 X 3i …X ji Furthermore, the server can aggregate the vector data according to different dimensions to obtain aggregated vector data, such as X. 11 and X 12 If the vector data is of different dimensions for a certain entity class, then the vector data X is divided according to the entity dimension. 11 and X 12 Data aggregation yields aggregated data, denoted as X1. In a subway transportation system using a palm-scanning system, X1 represents the palm-scanning recognition speed, and X2 represents the passenger gate passage speed. After obtaining the aggregated vector data, the server can configure relationships and initial influence factors between different aggregated vector data. This means the server can configure relationships and initial weights between different data source entity classes in the training set. For example, the server can configure a strong correlation between passenger gate passage speed X2 and palm-scanning recognition speed X1, with the correlation coefficient calculated based on past experience. The server can set the initial influence factor for these two vectors to 0.3. This initial influence factor corresponds to the implicit relationships between multiple data source entities to be mined in the future. Thus, the server can obtain relational data samples of {X2, X1, 0.3}. Assuming the server obtains aggregated vector data of {X1, X2, X3…X},... j The server can aggregate vector data {X1,X2,X3…X}. j Given a data sample {X2, X1, 0.3} and relational data, a deep neural network model is trained. For example... Figure 5 The input data for the deep neural network model shown is the vector data {X1, X2, X3…X} from the swipe-to-pay business system. j}

[0157] It is understood that in this embodiment of the application, the server can also perform data aggregation on vector data according to other different dimensions, such as time, scene, etc.

[0158] In this embodiment, the deep neural network model is pre-trained, specifically using a good UML graph entity relationship training set. This enables the deep neural network model to extract the relationship features of domain entities in the business system domain model when applied, and to extract relationships from the input data source entities, thereby obtaining relatively higher quality data source entity relationships. In other words, the deep neural network model can directly mine more direct and accurate entity relationships, enabling the generation of higher quality data flow dashboards based on more accurate entity relationships.

[0159] In one embodiment, such as Figure 6 As shown, the steps for configuring the relationships between different aggregated vector data to obtain relational data samples include:

[0160] Step 602: Obtain domain model diagram samples from different business systems;

[0161] Step 604: Determine the entity relationship data between each domain entity in the domain model graph sample based on the second representation vector of the domain model graph sample;

[0162] Step 606: Configure the relationships between different aggregated vector data based on entity relationship data to obtain relationship data samples.

[0163] The second representation vector refers to the vector used to describe the domain model graph sample. Since the domain model graph samples are compiled by developers for each business system, the second representation vectors obtained from different domain model graph samples may also be different. It can be understood that the domain model graph sample in this application can be a domain model graph corresponding to the target business scenario, selected based on that scenario.

[0164] Specifically, during the training of the initial deep neural network model, the server aggregates the vector data of different dimensions corresponding to each data source entity. After obtaining the aggregated vector data, the server can acquire domain model graph samples from different business systems and determine the entity relationship data between various domain entities in the domain model graph samples based on the second representation vector of the domain model graph samples. Furthermore, the server can configure the relationships between different aggregated vector data based on the entity relationship data between various domain entities in the domain model graph samples to obtain relationship data samples. That is, the server can configure the relationships and initial weights between different data source entity classes in the training set to obtain relationship data samples, so that the server can subsequently train the initial deep neural network model based on the relationship data samples to obtain a trained deep neural network model.

[0165] For example, let's take the swipe-based system used in a subway transportation scenario as an example. When training the initial deep neural network model, the server can obtain a domain model graph sample of the swipe-based system used in the subway transportation scenario. Based on the second representation vector of the domain model graph sample, it determines the entity relationship data between entities in each domain within the domain model graph sample. Assume that the server determines the entity relationship data between entities in each domain within the domain model graph sample based on the second representation vector of the domain model graph sample, including: {X2, X1, one-way association}, where X1 represents the swipe recognition speed and X2 represents the passenger gate passage speed. Based on the entity relationship data {X2, X1, one-way association} between entities in each domain within the domain model graph sample, the server can configure the relationship and initial weights between the aggregate vector data X1 and X2 to obtain a relationship data sample. This allows the server to subsequently train the initial deep neural network model based on the relationship data sample, resulting in a trained deep neural network model. Therefore, by pre-training the deep neural network model, specifically by using a good UML graph entity relationship training set, the deep neural network model can be used to extract the relationship features of domain entities in the business system domain model when applied. This allows for the extraction of relationships from the input data source entities, resulting in higher-quality data source entity relationships. In other words, the deep neural network model can directly mine more direct and accurate entity relationships, enabling the generation of higher-quality data flow dashboards based on these more accurate entity relationships.

[0166] This application also provides an application scenario in which the above-described data flow dashboard generation method is applied. Specifically, the data flow dashboard generation method is applied in this scenario as follows:

[0167] For large amounts of continuously generated data, when real-time management of data flow dashboards is required at the business control level, the aforementioned data flow dashboard generation method can be used. After staff log into the data flow dashboard platform, the platform automatically obtains the domain model diagram of the target business system and different data source entities. Based on the first representation vector of the domain model diagram, it determines the relationship data between the domain entities. Furthermore, based on the relationship data and the data source entities, the platform determines the data source entity relationships, data flow relationships, and influencing factors between the data source entities. Based on these relationships, the platform generates a visual data flow dashboard corresponding to the target business system. The data flow dashboard platform can be implemented using a dedicated server or a server cluster consisting of multiple servers.

[0168] The method provided in this application can be applied to various business development scenarios, as well as applications that require real-time monitoring of relevant data streams. The following description uses a subway transportation business scenario serving a mobile payment system as an example to illustrate the data stream dashboard generation method provided in this application.

[0169] In the traditional approach, data flow dashboards are displays of relationships between a few data source entities as defined by business requirements, and developers create them according to these requirements. With the development of low-code platforms, developers typically need to manually configure the corresponding data source entities in the configuration system, while the business logic can automatically generate the data flow dashboard by organizing it on the front end.

[0170] The disadvantages of the aforementioned traditional methods include:

[0171] (1) Data flow dashboards require manual configuration of data source entities;

[0172] (2) It ignores the relationships inherent in the data source entities themselves, as well as the impact of the domain model of the business system on the data flow.

[0173] (3) There may be redundancy, expired data, etc., which may lead to poor quality of the final data flow dashboard.

[0174] In actual business development, data dashboards are defined based on product or business requirements. Traditionally, this requires manually specifying two or more dimensions and then creating dashboards for different dimensions. For example, the relationship between price and car sales can be manually defined, based on experience. Developers want to understand the impact of overall car prices on car sales. Similarly, the relationship between car parts and car sales can be manually defined in other data dashboards. However, the price of car parts is also related to the overall car price. Therefore, these two types of dashboards express repetitive and redundant logic. Traditional methods easily overlook the inherent relationships between data entities and the impact of the business system's domain model on data flow. In real-world business scenarios, there are numerous entity concepts, and developers aim to find relationships between them. In dashboards composed of different entity concepts, the content of one dashboard might be expressed by another.

[0175] For example, if a person has multiple identities, and each identity corresponds to only one ability, then in reality, only a dashboard showing the correspondence between people and abilities is needed. The correspondence between people and identities, and between identities and abilities, is actually redundant in terms of recording identity data. That is, it is assumed that identity entities do not need to exist. The best entity relationship to be found is the direct correspondence between people and abilities.

[0176] Data expiration primarily refers to an excessive number of hierarchical relationships between entities, making it impossible to capture the true relationships between data entities in a timely manner. For example, in traditional methods, the relationships between A and B, B and C, and C and D are collected. However, only the relationship between A and D needs to be collected. There is no need for the time overhead of intermediate data collection. In other words, traditional methods require manual configuration of data source entities, ignoring the relationships inherent in the data source entities themselves. This may result in redundant logic and significant data collection time overhead.

[0177] Therefore, to address the aforementioned issues, this application provides a high-quality data flow dashboard generation method based on deep learning and domain models. This method, through deep learning, first pre-trains a large amount of data flow from a mobile phone system's subway transportation business scenario. It constructs neurons from multiple aspects of the data flow, such as data volume, quality, and time, for each data source entity, mining the relationships between multiple data source entities and the mutual influence of multiple data flows. Simultaneously, it obtains a high-quality domain model graph of the mobile phone system's subway transportation business scenario, manually edited during business system development, and extracts relevant entity relationships from this domain model graph as parameter tuning factors for the deep learning model, thereby helping to discover the optimal data source entity relationships and data flow relationships. Furthermore, the method provided in this application can be combined with low-code data flow dashboard configuration methods to generate corresponding high-quality data flow dashboards. The essence of the method provided in this application is that it can directly mine more direct and accurate entity relationships. Pre-drawing a good domain model graph enables the extraction of more accurate entity relationships, and using accurate entity relationships to construct a data dashboard can better express the relationships between entities.

[0178] The technical solutions provided in this application can solve the following problems:

[0179] (1) For multiple known data source entities that need to find related relationships, a deep learning model trained on a large amount of data can extract the relational features of domain entities in the domain model of the business system, extract the relationships of the input data source entities, and obtain relatively higher quality data source entity relationships.

[0180] (2) It can explore and discover the relationships between entities in the data source;

[0181] (3) Combining the data source entity relationships generated by the deep learning model with the low-code Kanban generation method can produce higher quality data flow Kanban.

[0182] In this application, the data source relationship is used to guide the generation of data dashboards. The object that generates data in the business can be called the data source entity. For example, a car generates data such as car price and car quality. In this case, both the entity and the data source entity are "car". The car price and car quality are data generated by the car entity, which may be attributes of the entity. For example, a car itself has a price, but its quality needs to be calculated.

[0183] Since there is a one-to-one correspondence between domain entities and data source entities expressed in UML diagrams, obtaining good data source entity relationships is equivalent to obtaining good UML diagram entity relationships. Therefore, this application can use a good UML diagram entity relationship training set to train a deep learning model. When applying the trained deep learning model, if a new project's UML diagram is input, the model can extract the entity relationships from the new project's UML diagram that are theoretically better in the domain model context and convert them into data source entity relationships. This guides the relationships between data sources to be more direct and strongly correlated, serving as a guide for building data flow dashboards. For example, the method provided in this application can discover that the best entity relationship is the direct correspondence between people and abilities, allowing for the direct construction of a data flow dashboard of people-ability relationships without the time overhead of intermediate data collection.

[0184] In this application, deep learning refers to a machine learning method implemented using a pre-trained deep neural network model. The model is generally a pre-trained deep neural network model, consisting of an input layer, hidden layers, and an output layer (activation function). Each layer consists of multiple neurons, obtained from the pre-trained model and parameters, and can uncover the connections between multiple neurons.

[0185] Pre-trained deep neural network models are essentially classifiers with nonlinear decision boundaries. The input data is the product of the input vector and the vector weights. The output data is obtained by calculating the loss value in the forward direction and adjusting the gradient parameters through backpropagation. In other words, the optimal model parameters can be obtained by pre-training with labeled data.

[0186] Domain model: A model used in domain-driven design to represent concepts within a domain, often including class diagrams, sequence diagrams, etc.

[0187] Domain-level entities: A type of entity in the domain model, typically including entity classes, control classes, and boundary classes, which undertake domain logic. In this embodiment, they mainly represent data source entities.

[0188] Data flow dashboard: A data monitoring platform that displays data changes in real time. It typically includes information such as monitored data source entities, monitored events, and the impact between monitored data sources. It plays an important role in the collection and control of information on business operations.

[0189] Palm scan system: A system that uses palm print for identity authentication, thereby granting corresponding permissions to perform various tasks. This type of system can serve a large number of identity-based business systems. For example, in subway transportation scenarios, a palm scan system can be used for passenger boarding.

[0190] On the technology side, such as Figure 7 The diagram shown is a schematic representation of the overall architecture of a high-quality data flow dashboard method based on deep learning and domain models. Figure 7 The high-quality data flow dashboard method based on deep learning and domain models shown in the figure consists of four main modules: data source module, pre-trained deep neural network module, domain model module, and data dashboard low-code generation module.

[0191] 1. Data source module

[0192] This module primarily interfaces with business systems to obtain raw data from data source entities. In the context of the subway transportation business scenario serving the swipe-based payment system, the raw data obtained from the data source entities of the swipe-based payment system includes, but is not limited to, subway gate passage data, swipe-based payment system recognition status, and recognition speed. The raw data is used in two parts: one part serves as the input layer data vector needed for training the pre-trained deep neural network module, thus contributing to the generation of the training set; the other part serves as data for extracting high-quality data source entity relationships.

[0193] The raw data can be actively obtained using methods such as logs and event tracking. The final data is displayed in JSON format, and the required fields are generally the data source entity name, time, and business fields.

[0194] The steps involved in generating a dataset by the data source module include:

[0195] Step 1: Obtain the raw JSON data based on logs, event tracking, and business rules, and convert it into vector format;

[0196] The second step is to aggregate the raw data according to multiple dimensions, such as time, entity, and scenario.

[0197] Step 3: Cleaning and adjusting relational data of data entity vectors, including handling bidirectional data, entities with duplicate names, and null value assignments;

[0198] Step 4: Configure the training set. For example, passenger gate speed and palm recognition speed are strongly correlated. The correlation coefficient can be calculated based on past experience. You can configure the influence factors of the two vectors, gate speed and palm recognition speed. Here, the influence factor corresponds to the implicit relationship between multiple data source entities that need to be explored later. The influence factor refers to the influence parameter in the deep learning model. In essence, all parameters of the deep learning model will affect each other. This is a black box process inside the model.

[0199] 2. Domain Model Module

[0200] This module is used to parse the domain model graph of a business system. Its main function is to convert the domain model graph into a computable vector form, extracting vectors that simply express the relationships between entities. In this embodiment, the domain model graph only retains the relationships between entities from the data source, serving as an influence factor that the pre-trained deep neural network module can compare.

[0201] like Figure 8 The diagram shown is the domain model diagram of the data source entities in the swipe palm system. It can be understood that... Figure 8 As a simplified entity domain model diagram of a swipe palm system data source, the server can... Figure 8 The domain model graph shown generates basic entity relationship vectors. That is, the server can extract the vectors of interest from the domain model graph according to preset rules, for example, from... Figure 8 The domain model diagram shown extracts the following entity relationship vectors: {(Swipe User, Natural Person, Direct Association 0.1, Aggregate 0.9, 0.2, 0.06, 0.7, 0, 0, 0, 0), (Swipe User, Status, User Status, Whether Swipe Service is Activated, xx), (Natural Person, Recognition Result, Whether Recognition Successful, xx, xx, xx), (Recognition Session, Recognition Result, Whether Recognition Successful, Recognition Status), (Recognition Session, Device)}. In these entity relationship vectors, 0.1 and 0.9 represent the influence parameters, i.e., influence factors, of different association relationship positions. Similarly, multiple domain model diagrams of a business system can generate multiple vectors expressing entities and data source relationships. They can also express basic relationship data such as the relationships between entities, the quantitative relationships between entities, and the indirect relationships between entities. This relationship data can be obtained through rule configuration.

[0202] 3. Pre-trained deep neural network module

[0203] This module features a typical neural network structure for deep learning, consisting of an input layer, multiple hidden layers, and an output layer. Internal details include neurons and activation functions. This module is responsible for extracting relationships between data source entities and influencing factors between data streams from the data stream itself, which are then used as parameters for the backend configuration of the subsequent low-code data dashboard generation module.

[0204] The input to the data source entity neurons is a vector converted from JSON business data. An initial weight is set for each vector; this initial weight can be a fixed value or an empirical value. To adjust the impact of the weights on the final output data of the neurons, the entire module uses the classic deep learning technique of backpropagation for parameter tuning. The neuron composition formula used in the swipe palm business system is: multi-dimensional swipe palm system data vector + weights + ReLU activation function + implicit relation vector output to the hidden layer. The relation vector in the hidden layer refers to the vector in the intermediate stage of the deep learning model.

[0205] It is understood that the activation function used in this application embodiment is a leaky linear rectified function, i.e., the activation function is Parametric ReLU. The most important function of the leaky linear rectified function is to adjust the learning parameter α. For positive numbers, the learning parameter α remains unchanged; for negative numbers, the learning parameter α is adjusted according to the loss value, the purpose of which is to control the change of the learning parameter. The reason why the deep learning model in this application uses a leaky linear rectified function is that the positively expressed data should be preserved, for example, if there is a relationship between two entities. If the expression indicates that there is no relationship between two entities, it may be a misjudgment. This judgment process occurs internally within the deep learning model, and it is hoped that the impact of misjudgment can be reduced. Figure 9 The diagram shown is a schematic of a multi-layer neural network structure for subway traffic scenarios. Figure 9 The input vector of node b1 in the middle input layer includes the passage speed, recognition status, recognition speed, and comparison accuracy. Each node in the hidden layer is essentially a vector, and nodes in the next layer are influenced by all nodes in the previous layer. For example, Figure 9 Hidden layer nodes b2 and b3 are essentially vectors. Hidden layer node b2 is affected by all nodes of the previous layer b1, and hidden layer node b3 is affected by all nodes of the previous layer b2.

[0206] The entity relationship input layer is fully connected to form hidden layers, where forward computation and weight calculation are performed. Each entity relationship neuron processes the input entity relationship vector from different angles, extracting a portion of the information from the input vector. This information is then used in the classification task to provide a basis for decision-making. Therefore, a single-layer entity relationship neuron essentially processes the input of the entity relationship input layer neurons from different angles; these angles can be understood as gradients.

[0207] like Figure 10The diagram illustrates the parameter tuning process in deep learning. The neural network propagates the output and calculates the loss value layer by layer through forward propagation, while backward propagation transmits the error back to adjust the calculated parameters. In this embodiment, the weight parameters of the neurons and the λ variable of the activation function are adjusted. This process is a classic procedure for deep learning parameter tuning and is only briefly described as a technique used in this application.

[0208] 1. Forward computation

[0209] For example, [H11,H12,H13] represents the result of multiplying the first vector of the entity relationship input variable matrix with the weight matrix W, as shown below:

[0210] H11=W(1)11*X1+W(1)21*X2+W(1)31*X3+B1

[0211] H12=W(1)12*X1+W(1)22*X2+W(1)32*X3+B2

[0212] H13=W(1)13*X1+W(1)23*X2+W(1)33*X3+B3

[0213] Applying the adjusted ReLU function to [H11,H12,H13] yields [F(H11),F(H12),F(H13)]. [OUT1,OUT2] = [F(H11),F(H12),F(H13)] * W2 + B2, where W2 is the weight matrix. Applying the adjusted ReLU function to [OUT1,OUT2] again yields [F(OUT1),F(OUT1)]. This is a forward computation process; substituting these values ​​into the loss function yields the loss value.

[0214] 2. Reverse computation and propagation

[0215] The key to training a neural network is training its weight matrix and biases. Backpropagation involves continuously adjusting the error generated by the sample based on the current parameters to ensure that the error is as small as possible. The commonly used strategy for backpropagation in neural networks is gradient descent, which achieves backpropagation through the chain rule of partial derivatives. The loss function selected in this application can be cross-entropy information entropy.

[0216] For vectorized data, the loss function derivation needs to be expanded layer by layer. For a single weight parameter, the corresponding error gradient can be obtained in the forward calculation. From this, a matrix representing the error gradient can be constructed. This matrix can be expressed as the sum of the loss values ​​propagated from the next layer in the backpropagation, and participates in the outer product of the activation function output in the forward calculation.

[0217] (1) Obtain the loss function, multiply it by a constant coefficient to express the error, and obtain the error expression E.

[0218] (2) Taking the partial derivative of the error term with respect to the error expression E, based on the chain rule of two derivatives, the expression can be similar to the following formula (1):

[0219]

[0220] The derivative of a neuron's output with respect to its input is the partial derivative of the activation function, and its expression can be obtained recursively.

[0221] (3) Select a learning rate α and use gradient descent to update the weight parameters and activation function λ.

[0222] 4. Data Dashboard Low-Code Generation Module

[0223] This module is mainly divided into a configuration system backend and a visual frontend for business configuration. This module can automatically generate a visual data flow dashboard based on the configuration data.

[0224] The backend is mainly responsible for configuring the data source entities and the influencing factors between entities. This can be viewed as a polynomial formula, with the influencing factors being the coefficients in the polynomial formula. These configurations can be manually configured by backend developers or automatically configured through relationships mined by deep learning modules. The configurations affect the quality of the generated visual data dashboards, including but not limited to whether they can more realistically display the relationships between entities, such as promoting or inhibiting relationships, or whether there are intermediary entities playing a role in the relationships between entities.

[0225] The front-end is mainly responsible for generating visual data curves and providing Excel report downloads based on the data configured on the back-end.

[0226] The beneficial effects of the method provided in this application embodiment include:

[0227] As business grows, the subway transportation scenario generates a large amount of unpredictable data, sent to the backend by different data generators. Therefore, real-time data management on the backend requires additional development work. Traditional data flow dashboards inevitably contain redundant and outdated data. Furthermore, due to varying levels of business understanding among product and development personnel, the resulting dashboards do not fully represent the comprehensive data flow situation. For example, they may fail to capture the relationships between multiple data flows or the deep relationships between multiple observed entities. The method provided in this application leverages past experience with data source entities—such as the domain model diagram of the business system and manually labeled known data source entity influence parameters—to generate higher-quality data source entity relationships and data flow relationships. This method expresses the influence of multiple entities implicit in multi-layered relationships and obtains more accurate influence parameters. Simultaneously, combined with a low-code data flow dashboard generation platform, it provides higher-quality data flow dashboards for business development, reduces labor costs, minimizes redundant data, improves the maintainability of data flow dashboards, and more effectively empowers business exploration. This significantly improves the quality of data flow dashboards, providing higher-quality and more accurate dashboards.

[0228] It should be understood that although the steps in the flowcharts of the embodiments described above are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the embodiments described above may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.

[0229] Based on the same inventive concept, this application also provides a data flow dashboard generation apparatus for implementing the data flow dashboard generation method described above. The solution provided by this apparatus is similar to the implementation described in the above method; therefore, the specific limitations of one or more data flow dashboard generation apparatus embodiments provided below can be found in the limitations of the data flow dashboard generation method described above, and will not be repeated here.

[0230] In one embodiment, such as Figure 11 As shown, a data flow dashboard generation device is provided, comprising: an acquisition module 1102, a determination module 1104, and a generation module 1106, wherein:

[0231] The acquisition module 1102 is used to acquire the domain model diagram of the business system and different data source entities.

[0232] The determination module 1104 is used to determine the relationship data between domain entities in the domain model graph based on the first representation vector of the domain model graph; and to determine the data source entity relationship, data flow relationship and influencing factor between the data source entities based on the relationship data and the data source entities.

[0233] The generation module 1106 is used to generate a visual data flow dashboard corresponding to the business system based on the entity relationship, data flow relationship and influencing factors of the data source.

[0234] In one embodiment, the apparatus further includes: a processing module, configured to perform vectorization processing on the domain model graph to obtain a first representation vector of the domain model graph; and a determining module, configured to determine, based on the first representation vector, relationship data between a target domain entity and combinations of other domain entities in the domain model graph; or, based on the first representation vector, determine relationship data between various domain entities in the domain model graph.

[0235] In one embodiment, the apparatus further includes a selection module and an aggregation module. The acquisition module is further configured to acquire raw data corresponding to different data source entities; the selection module is configured to select target data from the raw data; the aggregation module is configured to aggregate the data source entity relationships, data flow relationships, and influencing factors between the various data source entities with the target data to obtain first aggregated data; the generation module is further configured to generate a visual data flow dashboard corresponding to the business system based on the first aggregated data.

[0236] In one embodiment, the acquisition module is further configured to acquire the original data corresponding to different data source entities; the aggregation module is further configured to aggregate the data source entity relationships, data flow relationships and influencing factors between the data source entities with the original data to obtain second aggregated data; the generation module is further configured to generate a visual data flow dashboard corresponding to the business system based on the second aggregated data.

[0237] In one embodiment, the generation module is further configured to use the aggregated data as configuration data for the platform backend, so that the platform frontend generates a visual data flow dashboard corresponding to the business system based on the configuration data; wherein, the aggregated data is the first aggregated data or the second aggregated data; the platform backend and the platform frontend are respectively the backend and frontend of the data flow dashboard platform.

[0238] In one embodiment, the apparatus further includes: a display module for displaying the data stream dashboard through the platform front end; the data stream dashboard includes a visual data chart and a download button; the processing module is further configured to download the visual data chart in response to a trigger operation of the download button; or, export the data stream dashboard.

[0239] In one embodiment, the processing module is further configured to process the relational data and the identification information of the data source entities using a deep neural network model to obtain the data source entity relationships and influence factors between each of the data source entities; the determining module is further configured to determine the data flow relationships between each of the data source entities based on the data source entity relationships between each of the data source entities, the influence factors, and the original data corresponding to different data source entities.

[0240] In one embodiment, the acquisition module is further configured to acquire the initial influence factors of each of the data source entities; the processing module is further configured to process the relational data, each of the data source entities, and the initial influence factors of each of the data source entities through the deep neural network model to obtain the data source entity relationships and influence factors between each of the data source entities.

[0241] In one embodiment, the selection module is further configured to select a target impact factor that meets the threshold condition from the impact factors; the acquisition module is further configured to acquire the data source entity relationship corresponding to the target impact factor, and use the data source entity relationship as the target data source entity relationship.

[0242] In one embodiment, the apparatus further includes a configuration module and a training module. The acquisition module is further configured to acquire entity samples from a data source and convert the entity samples into vector data; the aggregation module is further configured to aggregate the vector data according to different dimensions to obtain aggregated vector data; the configuration module is configured to configure the relationships between different aggregated vector data to obtain relationship data samples; and the training module is configured to train the deep neural network model based on the aggregated vector data and the relationship data samples.

[0243] In one embodiment, the acquisition module is further configured to acquire domain model graph samples of different business systems; the determination module is further configured to determine entity relationship data between domain entities in the domain model graph sample based on the second representation vector of the domain model graph sample; and the configuration module is further configured to configure the relationship between different aggregate vector data based on the entity relationship data to obtain a relationship data sample.

[0244] Each module in the aforementioned data flow dashboard generation device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the operations corresponding to each module.

[0245] In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 12 As shown, this computer device includes a processor, memory, input / output interfaces (I / O), and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores data for generating data flow dashboards. The I / O interfaces are used for exchanging information between the processor and external devices. The communication interface is used for communicating with external terminals via a network connection. When the computer program is executed by the processor, it implements a method for generating data flow dashboards.

[0246] Those skilled in the art will understand that Figure 12 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0247] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.

[0248] In one embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above method embodiments.

[0249] In one embodiment, a computer program product or computer program is provided, the computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and executes the computer instructions, causing the computer device to perform the steps in the above method embodiments.

[0250] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data shall comply with the relevant laws, regulations and standards of the relevant countries and regions.

[0251] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, quantum computing-based data processing logic devices, etc., and are not limited to these.

[0252] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0253] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.

Claims

1. A method for generating a data flow dashboard, characterized in that, The method includes: Obtain the domain model diagram of the business system and different data source entities; Based on the first representation vector of the domain model graph, the relationship data between each domain entity in the domain model graph is determined; the first representation vector is obtained by vectorizing the domain model graph; the relationship data is used to express the relationship between each domain entity; the domain model graph includes a UML diagram. Using a pre-trained deep neural network model, based on the relational data and each of the data source entities, the data source entity relationships, data flow relationships, and influencing factors among the data source entities are determined. The data source entity relationships refer to the associations between the data source entities, and these associations include at least unidirectional associations, generalization relationships, combination relationships, derivation relationships, and aggregation relationships. The influencing factors are parameters used to reflect the relationships between the data source entities. The relational data is used as a parameter tuning factor for the deep neural network model. Based on the data source entity relationships, the data flow relationships, and the influencing factors, a visual data flow dashboard corresponding to the business system is generated.

2. The method according to claim 1, characterized in that, The step of determining the relationship data between domain entities in the domain model graph based on the first representation vector of the domain model graph includes: The domain model graph is vectorized to obtain the first representation vector of the domain model graph. Based on the first representation vector, determine the relationship data between the target domain entity and combinations of other domain entities in the domain model graph; or... Based on the first representation vector, determine the relationship data between entities in each domain in the domain model graph.

3. The method according to claim 1, characterized in that, The method further includes: Obtain the original data corresponding to the different data source entities; Select the target data from the original data, and aggregate the data source entity relationships, data flow relationships, and influencing factors between the data source entities with the target data to obtain the first aggregated data; The step of generating a visual data flow dashboard corresponding to the business system based on the data source entity relationship, the data flow relationship, and the influencing factors includes: Based on the first aggregated data, a visual data flow dashboard corresponding to the business system is generated.

4. The method according to claim 1, characterized in that, The method further includes: Obtain the original data corresponding to the different data source entities; The data source entity relationships, data flow relationships, and influencing factors among the various data source entities are aggregated with the original data to obtain the second aggregated data; The step of generating a visual data flow dashboard corresponding to the business system based on the data source entity relationship, the data flow relationship, and the influencing factors includes: Based on the second aggregated data, a visual data flow dashboard corresponding to the business system is generated.

5. The method according to claim 1, characterized in that, The step of generating a visual data flow dashboard corresponding to the business system based on the data source entity relationship, the data flow relationship, and the influencing factors includes: The aggregated data is used as configuration data for the platform backend, so that the platform frontend can generate a visual data flow dashboard corresponding to the business system based on the configuration data; wherein, the aggregated data is either first aggregated data or second aggregated data; the platform backend and the platform frontend are respectively the backend and frontend of the data flow dashboard platform.

6. The method according to claim 5, characterized in that, The method further includes: The data flow dashboard is displayed on the platform's front end; the data flow dashboard includes visual data charts and a download button. In response to the triggering of the download button, download the visualized data chart; or export the data stream dashboard.

7. The method according to claim 1, characterized in that, The step of determining the data source entity relationships, data flow relationships, and influencing factors among the data source entities based on the relational data and each of the data source entities includes: The relational data and the identification information of the data source entities are processed by a deep neural network model to obtain the data source entity relationships and influencing factors among the data source entities. Based on the data source entity relationships between the data source entities, the influencing factors, and the original data corresponding to different data source entities, the data flow relationships between the data source entities are determined.

8. The method according to claim 7, characterized in that, The process of processing the relational data and the identification information of the data source entities using a deep neural network model to obtain the data source entity relationships and influencing factors between each data source entity includes: Obtain the initial impact factor for each of the aforementioned data source entities; The deep neural network model is used to process the relational data, each data source entity, and the initial influence factors of each data source entity to obtain the data source entity relationships and influence factors between each data source entity.

9. The method according to claim 8, characterized in that, The method further includes: Select target impact factors that meet the threshold conditions from among the aforementioned impact factors; Obtain the data source entity relationship corresponding to the target influence factor, and use the data source entity relationship as the target data source entity relationship.

10. The method according to claim 1, characterized in that, The data source entity relationships and the influencing factors are obtained by processing the relationship data and each of the data source entities using a deep neural network model; the method further includes: Obtain entity samples from the data source and convert the entity samples from the data source into vector data; The vector data is aggregated according to different dimensions to obtain aggregated vector data. Configure the relationships between different aggregated vector data to obtain relationship data samples; The deep neural network model is trained based on the aggregated vector data and the relational data samples.

11. The method according to claim 10, characterized in that, The configuration of different relationships between the aggregated vector data yields a sample of relational data, including: Obtain domain model graph samples from different business systems; Based on the second representation vector of the domain model graph sample, determine the entity relationship data between each domain entity in the domain model graph sample; Based on the entity relationship data, different aggregated vector data are configured to obtain relationship data samples.

12. A data stream dashboard generation device, characterized in that, The device includes: The acquisition module is used to acquire the domain model diagram of the business system and different data source entities; A determination module is used to determine the relationship data between domain entities in the domain model graph based on a first representation vector of the domain model graph; the first representation vector is obtained by vectorizing the domain model graph; the relationship data is data used to express the relationships between the domain entities; the domain model graph includes a UML diagram; and a pre-trained deep neural network model is used to determine the data source entity relationships, data flow relationships, and influencing factors between the data source entities based on the relationship data and the data source entities; the data source entity relationships refer to the association relationships between the data source entities, and the association relationships include at least unidirectional association relationships, generalization relationships, combination relationships, derivation relationships, and aggregation relationships; the influencing factors are parameters used to reflect the relationships between the data source entities; wherein, the relationship data is used as a parameter tuning factor for the deep neural network model; The generation module is used to generate a visual data flow dashboard corresponding to the business system based on the data source entity relationship, the data flow relationship, and the influencing factors.

13. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 11.

14. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 11.

15. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 11.

Images