A system that enables the automatic detection of database relationships through AI-powered source code analysis.

Abstract

This invention relates to a system (1) that enables the automatic detection of implicit relationships between database tables and columns by analyzing application source codes with the support of artificial intelligence.

Description

1 TARIFF DATA THROUGH ARTIFICIAL INTELLIGENCE-SUPPORTED SOURCE CODE ANALYSIS AUTOMATIC DETECTION OF BASE RELATIONSHIPS A SYSTEM THAT PROVIDES Technical Area This invention analyzes application source code with the support of artificial intelligence to generate data. Automatic detection of implicit relationships between base tables and columns. It is related to a system that enables this. 10 Previous Technique In many applications with a poorly designed database in the known state of the art Data loss occurs for many reasons. Manually extracting data from a table 15 When that data is deleted, the data in another table that it is related to remains, which This leads to data inconsistencies. When data is deleted from a table, it becomes inconsistent with that table. The parallel-running function becomes unusable. Therefore, considering the studies and shortcomings in the current technology, 20 When this is the case, the application source code can be analyzed with the support of artificial intelligence. by automatically establishing implicit relationships between database tables and columns. to be identified as having no physical foreign key establishing logical relationships in environments, parent-child relationships extracting hierarchies and modeling these relationships with confidence scores data 25 a system is needed that ensures the improvement of integrity and testing processes It seems that it was heard. United States Regulation US20250036602A1, which is included in the known state of the art. The patent document specifies that primary, unique, and / or 30 datasets are required for relational datasets. Techniques for discovering foreign keys are explained. These techniques involve the relevant data. 2 profiling relational datasets to obtain their profiles; the first relational the initial data profile of the dataset and the first trained machine learning model using one or more primary keys for the initial relational dataset determining the candidate; the second relational dataset to the first relational dataset According to the subset analysis, one or more primary key candidates are identified. 5 using one or more foreign keys for the second relational dataset determining the proposal; the first data profile, the second relational dataset's second data profile and a second model different from the first trained machine learning model using a trained machine learning model for the second relational dataset or the identification of more foreign key candidates; and the primary key 10 This includes providing the candidates and foreign key candidates as output. Brief Description of the Invention The aim of this invention is to analyze application source code with the support of artificial intelligence. by automatically establishing implicit relationships between database tables and columns. to be identified as such in environments where no physical foreign key is found. the creation of logical relationships, the derivation of hierarchical structures, and this Data integrity and testing processes are modeled using trust scores for relationships. The goal is to implement a system developed to ensure its improvement. 20 Another purpose of this invention is to integrate application source code with relational databases. a multi-layered artificial structure to explore the implicit relationships between its structures the use of intelligence architecture, static code analysis, semantic interpretation, and machine learning. By working together with techniques in the form of learning, database tables and columns 25 and to enable the automatic detection of relationships between these elements. The goal is to implement a system developed for this purpose. Detailed Description of the Invention 3 The "AI-Powered Resources" project was carried out to achieve the purpose of this invention. Automatic Detection of Database Relationships Through Code Analysis A "System Providing" is shown in the attached figure; Figure 1; Schematic view of the system that is the subject of the invention. 5 The parts shown in the figure are individually numbered, and these numbers... The corresponding answers are given below. 1. System 10 2. Application 3. Server Using artificial intelligence, application source code is analyzed and a database is created. automatic detection of implicit relationships between tables and columns 15 The system in question, developed for the purpose of providing (1); - software that programmers write in human-readable languages source code, which are text files containing instructions for operation compiled and translated into machine language, the final version executed by the user. at least one application (2) structured to be software, 20 - application (2) source codes, Object-Relational Mapping configurations and entity definitions, Structured Query Language queries and join statements, method calls and service flows, data the base table and column metadata and relationship usage frequencies and Working with data in the form of access patterns, application 25 (2) source codes, Object-Relational Mapping mappings and embedded Structured Query Language queries can be analyzed using static and semi-dynamic methods. by examining the code's data access behaviors using various techniques creating a multi-layered program graph, source code; 30 in the form of control flow chart, call chart and data flow chart. modeling, using artificial intelligence algorithms on these graphical structures 4 Applying the graph representation-based learning technique involves method calls, parameter passes, repository usage, and join statements Modeling in terms of nodes and edges, the textual content of the code without being tied to, the actual working logic and data access To enable the learning of patterns, the program graphs created are 5. pattern recognition and artificial intelligence unsupervised learning using algorithms, iterative changes between tables and columns Identifying relationship structures, together within the same data flow chain the table and column pairs included, potential foreign keys and parent-child pairs Marking them as relationship candidates, AI algorithm data stream 10 direction, call chain continuity, and centrality within the graph. to enable it to evaluate its measurements, throughout the entire code. Data structures that are used together consistently as relationship candidates. It includes at least one server (3) configured to determine. The application (2) in the system (1) that is the subject of the invention, programmers by human hands Text files written in readable languages ​​that contain instructions for how the software is used. The source code is compiled and translated into machine language, then executed by the user. It is being configured to be the final software. The source codes of the server (3), application (2) included in the system (1) are the subject of the invention, Object-Relational Mapping (ORM) configurations and Entity definitions, Structured Query Language (SQL) queries and join statements, method calls and service flows, database Table and column metadata, relationship usage frequencies, and access patterns 25 Working with data in the form of application (2) source code, Object- Relational Mapping mappings and embedded Structured Query Language queries By examining the code's data access using static and semi-dynamic analysis techniques... Creating a multi-layered program graph that represents its behavior, resources code; control flow graph (CFG), call graph 30 and modeling it as a data flow graph, artificial intelligence with the algorithm, graph representation-based learning is performed on these graph structures (Graph Applying the Representation Learning technique involves analyzing method calls and parameters. Nodes handle transitions, repository usage, and join statements. and modeling it as edges, without being bound by the textual content of the code, 5. To enable learning the actual working principles and data access patterns. It is structured in such a way. The server (3) displays the generated program graphics. pattern recognition (pattern mining) and unsupervised artificial intelligence learning algorithms Using this method, identify repetitive relationship structures between tables and columns. to, pairs of tables and columns that are located together within the same data flow chain, Potential foreign key and parent-child relationship candidates 10 to mark as, the data flow direction of the artificial intelligence algorithm, call chain evaluating continuity and centrality measures within the graph to ensure data structures are used consistently throughout the entire code. It is structured to determine each relationship candidate. The server (3) determines each relationship For the candidate, a multidimensional and data flow-intensive system consisting of graphic-based features, 15 graph link strength, call chain depth, column data type fit, and repeat rate. Creating a feature vector containing metrics such as these, using an artificial intelligence model Using these vectors, generate a confidence score for each relationship, based on a defined threshold. Relationship candidates above the value, logical foreign key or parent-child to consider it as a relationship, 20 obtained from new code analyses over time by updating itself with graphic patterns, it achieves higher levels of accuracy. It is configured to enable access to the server (3), graph representation. High-security relationship candidates obtained from the learning-based layer, physical Logical (soft) foreign key structures independent of database constraints. Modeling it in this way, with tables as nodes and relationships as directed edges. 25 by defining them as such and creating a hierarchical relationship graph It is structured. The server (3) uses an artificial intelligence algorithm to plot the graph. By analyzing data flow direction and connection density, it can identify upstream and downstream relationships and multiplexes. It is configured to automatically extract multi-level data chains. Server (3) updates changes made to application (2) source code, new 30 By monitoring the added service flows and updated data access patterns, the relationship 6 dynamically re-evaluating the graph, over time, with an artificial intelligence model. by comparing the changing code behaviors within it with previous graphical patterns It is being configured to recalculate the validity of the relationships. Server (3), thanks to the feedback mechanisms, those with low confidence scores or no longer weakening or disabling unused relationships, consistently 5 It is structured to strengthen the relationships used. Server (3), graph The logical relationship model created with AI based on; test data generation, data masking, data integrity auditing, unauthorized data recording, and anomaly detection. to be used as direct decision support input in processes such as these to provide, thanks to upper-lower hierarchies, synthetic or manual test data 10 to ensure that relational violations are automatically prevented during creation is configured. Server (3), relational database management systems, automatic (D) Implementation with external tools in the form of test tools and corporate applications. programming interface (Application Programming Interface - API) or service It is configured to work in an integrated manner through the layer. 15 Industrial application of the invention The system in question has (1) application (2) code and data structures in its resources being examined, the relationships between artificial intelligence, codes and databases are being made sense, 20 A confidence score is calculated, and based on this confidence score, the database... Relational situations are created, and continuous learning takes place based on the obtained codes. The mechanism is triggered, database relationships are revealed, and outputs are generated. A structure is being designed to systematize the process. Around these fundamental concepts, the invention's subject is "AI-Powered Source Code". Automatic Detection of Database Relationships Through Analysis It is possible to develop a wide variety of applications related to "A System Providing (1)". and the invention cannot be limited to the examples described here, but mainly to the claims as stated. 30

Claims

7 REQUESTS 1. Data is extracted by analyzing application source code with the support of artificial intelligence. automatic implicit relationships between base tables and columns Developed to enable detection; 5 - software that programmers write in human-readable languages source code, which are text files containing instructions for operation compiled and translated into machine language, the final version executed by the user. and includes at least one application (2) which is structured to be software; - application (2) source codes, Object-Relational Mapping 10 configurations and entity definitions, Structured Query Language queries and join statements, method calls and service flows, data the base table and column metadata and relationship usage frequencies and Working with data in the form of access patterns, application (2) source codes, Object-Relational Mapping mappings and embedded 15 Structured Query Language queries can be analyzed using static and semi-dynamic methods. by examining the code's data access behaviors using various techniques creating a multi-layered program graph, source code; in the form of control flow chart, call chart and data flow chart modeling, 20 on these graphical structures with artificial intelligence algorithms Applying the graph representation-based learning technique involves method calls, parameter passes, repository usage, and join statements Modeling in terms of nodes and edges, the textual content of the code without being tied to, the actual working logic and data access To enable the learning of patterns, the program graphs created are 25. pattern recognition and artificial intelligence unsupervised learning using algorithms, iterative changes between tables and columns Identifying relationship structures, together within the same data flow chain the table and column pairs included, potential foreign keys and parent-child pairs Marking them as relationship candidates, AI algorithm data stream 30 direction, call chain continuity, and centrality within the graph. 8 to enable it to evaluate its measurements, throughout the entire code. Data structures that are used together consistently as relationship candidates. characterized by at least one server (3) configured to determine a system (1).

2. Software that programmers write in human-readable languages. source code, which are text files containing instructions for operation compiled and translated into machine language, the final software that the user runs. The application (2) is characterized by the Request structured to be (2) A system like the one in 1 (1). 10 3. Application (2) source codes, Object-Relational Mapping configurations and entity definitions, Structured Query Language queries, and join statements, method calls and service flows, database table and column metadata, and Using data in the form of relationship usage frequencies and access patterns, 15 to work, application (2) source codes, Object-Relational Mapping their mappings and embedded Structured Query Language queries are static and By examining the code's data access using semi-dynamic analysis techniques creating a multi-layered program graph that represents its behavior, Source code; control flowchart, call graph and data flow graph 20 modeling in this way, using artificial intelligence algorithms to create these graphical structures. applying a learning technique based on graph representation, method calls, parameter passes, repository usage, and participation Modeling expressions as nodes and edges, textual code Regardless of the content, the actual working logic and data access 25 with the server (3) configured to enable learning of patterns as in any of the above characterized claims system (1).

4. Pattern recognition and artificial intelligence on the generated program graphics 30 Using unsupervised learning algorithms, tables and columns 9 Identifying recurring relationship structures between them, within the same data flow chain pairs of tables and columns that are included together, potential foreign entities Marking as key and parent-child relationship candidates by artificial intelligence the algorithm's data flow direction, call chain continuity, and graph within To enable the evaluation of centrality metrics, the entire code is 5. data structures that are used consistently together throughout as relationship candidates. characterized by the server (3) configured to determine a system like any of the above requests (1).

5. For each potential relationship candidate, a multidimensional and 10-dimensional system consisting of graphic-based features. data flow density, graph connection strength, call chain depth, column data a feature vector containing metrics such as type fit and repetition rate to create, using artificial intelligence models, these vectors for each relationship generating a confidence score, the relationship above the determined threshold value. 15 candidates considered as logical foreign keys or parent-child relationships to do, graphs obtained from new code analyses over time by updating itself with its patterns, it achieves higher levels of accuracy. characterized by the server (3) configured to enable access a system like any of the above-mentioned requests (1).

6. High confidence obtained from the learning layer based on graph representation. relationship candidates, regardless of physical database limitations. Modeling tables as logical foreign key structures, nodes, By defining the relationships as directed edges, a hierarchical relationship graph can be created. 25 characterized by the server (3) configured to create a system like any of the above requests (1).

7. Using an artificial intelligence algorithm, the direction of data flow and connections on the graph are determined. by analyzing density, upstream and downstream relationships and multi-level data 30 with server (3) configured to automatically extract chains as in any of the above characterized claims system (1).

8. Application (2) changes made in the source code, new By monitoring the added service flows and updated data access patterns 5 Dynamically re-evaluating the relationship graph, artificial intelligence model with, code behavior that changes over time compared to previous graphical patterns. to reassess the validity of relationships by comparing them from the above requests characterized by the configured server (3) a system like any of them (1). 10 9. Thanks to feedback mechanisms, those with low confidence scores or weakening or deactivating relationships that are no longer in use, consistent Server configured to strengthen the relationships used in this way (3) a 15 as in any of the above claims characterized by system (1).

10. Testing the logical relationship model created with graph-based artificial intelligence. data generation, data masking, data integrity checking, orphaned data Direct decision support in processes such as registration and anomaly detection 20 to enable its use as input, thanks to upper-lower hierarchies, Automated relational violations when generating synthetic or manual test data. server (3) configured to ensure that it is prevented as such as in any of the above characterized claims system (1). 25 11. Relational database management systems, automated testing tools, and enterprise Application programming tools with external tools in the form of applications (D) to work in an integrated manner via the interface or service layer 11 from the above requests characterized by the configured server (3) a system like any other (1). 10 20 30