A SYSTEM FOR AUTOMATIC GENERATION OF BUSINESS PROCESS-FOCUSED DOCUMENTS FROM CODE CHANGES AND PUBLICATION TO THE CORPORATE KNOWLEDGE BASE.

Abstract

This invention relates to a system (1) developed to enable the mapping of code changes to business processes and / or business capabilities through multilingual static and / or dynamic analysis and semantic interpretation supported by a large language model, and to automatically update them on a part-by-part basis throughout their lifecycle in the enterprise knowledge base.

Description

1 TARIFF CODE CHANGES TO BUSINESS PROCESS-FOCUSED DOCUMENTS AUTOMATED GENERATION AND CORPORATE KNOWLEDGE BASE PUBLICATION SYSTEM Technical Area This invention enables multilingual static and / or dynamic analysis of code changes and large language segmentation. model-supported semantic interpretation to business processes and / or business capabilities mapping and automatic tracking throughout the lifecycle in the corporate knowledge base 10 It relates to a system that enables section-by-section updates. Previous Technique Current software analysis and documentation methods are mostly focused on a one-time migration. 15 or the business logic of the existing system in order to complete the transformation project It focuses on removing the existing business rules in the system. However, this approach involves removing the existing business rules in the system. decision-making mechanisms, dependencies, and the reasons behind them are persistent and It does not guarantee systematic capture. Therefore, the application The departure of the people who developed or knew the system from the organization, the systems deteriorate over time. critical business information as a result of changes or migration to different technologies It can get lost. Especially if the existing code, database and technical documentation are all the same. why and under what conditions business knowledge is applied in a particular way inability to explain, loss of business knowledge in software development processes and the reuse of that knowledge This leads to significant technical problems, such as high analysis costs for obtaining it. 25 It opens. 2 Therefore, code changes require multilingual static and / or dynamic analysis and large language processing. model-supported semantic interpretation to business processes and / or business capabilities mapping and automated tracking throughout the lifecycle in the corporate knowledge base. It appears that a system is needed that allows for section-by-section updates. According to Chinese patent document CN120560715, which is included in the prior art, Automatic synchronization of large model-based code and development documentation. The method in question is discussed. This invention is a large model (AI) based code. and the development document relates to the automatic synchronization method; the document with Code association and AI-powered automatic synchronization 10 by using this mechanism, the document and the code always remain consistent. It ensures and prevents manual update errors or neglect of updates. It prevents various development problems arising from this. According to this technical solution, the method; Adding a link to the development document to the code description, including the code address. Adding it to the development document and establishing a bidirectional relationship are steps 15 It includes; an AI sync button after the development document is modified. When clicked, artificial intelligence automatically analyzes the change in the document. This involves generating updated code along with the code's business logic, and the code After making the changes, the AI ​​code can be synchronized by clicking the AI ​​synchronization button again. It automatically analyzes the change and includes it in the development document content 20 It enables the creation of updated development documentation. Brief Description of the Invention The purpose of this invention is to enable multilingual static and / or dynamic analysis of code changes and 25 Business processes and / or business through semantic interpretation supported by the Big Language Model. mapping their capabilities and their lifecycle within the corporate knowledge base. 3 A system developed to enable automatic, section-by-section updates. to accomplish. Detailed Description of the Invention 5 The “Business Process Focused from Code Changes” technology was developed to achieve the purpose of this invention. "System for Automated Document Generation and Publication to the Corporate Knowledge Base" This is shown in the attached figure; Figure 1 shows a schematic view of the system that is the subject of the invention. The parts shown in the figure are individually numbered, and the corresponding numbers correspond to these numbers. It is given below. 1. System 2. Application 3. Server Code changes are analyzed using multilingual static and / or dynamic methods and a large language model. 20 mapping to business processes and / or business capabilities through supported semantic interpretation and automatically segment-based throughout the lifecycle in the corporate knowledge base The system in question, developed to ensure its updating (1); -user's software development and approval or comment processes at least one application (2) structured to enable it to perform, 25 -to communicate with the application (2) and receive data using any communication protocol performing transactions, saving, merging, or pulling code from the code repository. the request involves listening to events, recording changes, and which 4 Detecting if files or functions have been changed, code extracting business meaning from the changes, summarizing business meaning in natural language. and creating documents in a format suitable for knowledge bases, the generated document reading, creating, and managing content on platforms using code Adding it to the relevant page via application programming interfaces or 5 updating, with each record's unique identification number in the document section matching and generating information about who did what and when, and the user to gather their approval or comments and to improve the model It includes at least one configured server (3). The application (2) in the system (1) which is the subject of the invention does not require any communication protocol. to communicate with the server (3) and exchange data using It is structured. The application (2) is structured after the user develops their code. to enable it to upload its code to the versioning platform It is being structured. 15 In the system that is the subject of the invention (1), the server (3), webhook or CI / CD pipeline trigger codes with asynchronous processing algorithms using technologies and simple event queue management. listening to recording, merging or withdrawal request events made in the repository It is configured to provide AST 20 in the structural analysis of the code (3). (Abstract Syntax Tree), CFG (Control Flow Graph) and in understanding the flow of business logic. Diff parsing algorithms are used to understand row-based changes. to enable the identification of which functions are affected by the changes It is configured. The server (3) uses LLM-based code interpretation, word embeddings and Cosine similarity is a corporate business concept dictionary with NLP-based matching 25 and LDA or BERTopic, which enable code changes to be linked to business processes. topic modeling algorithms such as these derive business meaning from code changes It is configured to enable extraction. The server (3) is transformer-based. Text summarization in the form of models, templates for standard formatting, and LLM hybrid. Temple-based NLG approach involving updating only the changed sections Delta Update provides the longest common subsequence and patch generation method. Summarizing the meaning of the job in natural language using algorithms and adding it to the knowledge base. It is configured to ensure that documents are created in the appropriate format. Server 5 (3), section anchoring and document with ID-based updates for in-page sections document produced using version control algorithms that involve managing versions reading, creating, and managing content on platforms using code or added to the relevant page via application programming interfaces that provide It is configured to enable updating. Server (3), hash mapping and 10 Graph database algorithms generate a unique identification number for each record in the document. matching it with the relevant section and creating information about who did what and when. It is structured to provide active learning and reinforcement. The server (3) learning with human feedback algorithms to obtain user approval or comments It is configured to collect data and improve the model. Server 15 (3), corporate using any communication protocol through any communication channel To communicate and exchange data with the information repository and document platform. to perform, records of changes made to the code through the relevant code document to ensure that it is located and that the places where changes were made are found is structured. The server (3) artificially creates static and dynamic semantic interpretation of the code. interpreting it through intelligence or machine learning techniques and understanding business logic and to ensure that business rules are extracted from the work done by the code is being configured. The server (3) is configuring the business purpose, scope, and affected by the change. Creating a summary of the parts and data fields and conducting a risk or impact analysis. It is configured to enable its production. The server (3) provides 25 documents to the document platform. By connecting, the relevant work section is found if an existing document exists, and the diff algorithm is used. with the detection of changes, the generative pre-training of the detected changes the data is transmitted to the transformer model and the changes are implemented, thus creating the system. 6 the new document is saved again to the document platform and the user can review it. to ensure that the application (2) interface is sent for approval in order to pass through is structured. The server (3), in a preferred arrangement of the invention, calls graph And dependency analysis allows us to see the real impact in multi-service architectures. Cross-repository or monorepo support and coverage and risk of affected functions 5 impact analysis using test scope correlation techniques that provide the score It is structured to enable the deepening of the invention. The server (3) prefers the invention. In one of the arrangements, code change log, code merge request and job log. Concept mining, ontology, or the process of extracting new business terms from text. Versioning and 10, which also enable tracking and documentation of taxonomy changes. Multilingual term matching that provides automatic alignment between Turkish and English. It is structured to provide concept management and a rich ontology through its techniques. The server (3) converts the LLM output to a JSON schema in a preferred arrangement of the invention. Schematic-oriented manufacturing, which involves locking the template and then casting it, is systematic. 15 errors detected through statements and rules, as well as personal identifying information and keyword scanning. Content validation and automated styling that enable claim checks to be performed. to ensure quality assurance through terminology control techniques is structured. The server (3), in a preferred arrangement of the invention, approval and feedback interface and quality metrics with human loop learning It is structured to enable its realization. The server (3) prefers the invention 20 In one of the implemented features, caching and repetition prevention, batch processing, and Performance and cost through windowing and in-situ small model techniques. It is configured to enable balancing. Server (3), In a preferred arrangement of the invention, role-based access control and attributes access control based on data tracing, audit trail hardening, secret key management, and input or 25 It is structured to ensure security and compliance through output masking techniques. The server (3) provides end-to-end tracking with opentelemetry in a preferred arrangement of the invention. 7 driving, service level indicator or service level target and business impact metrics It is structured to allow for monitoring of the observability of the flow. Industrial Applicability The system (1) in question enables code changes to be made in a multilingual static and / or Business processes through semantic interpretation supported by dynamic analysis and large language models. and / or mapping to job capabilities and their lifecycle in the corporate knowledge base. It is automatically updated section by section throughout. Around these fundamental concepts, the invention focuses on "Business Process-Oriented Code Changes." System for Automated Document Generation and Publication to the Corporate Knowledge Base It is possible to develop a wide variety of applications related to (1)”, and the invention is here It cannot be limited to the examples given; it is essentially as stated in the claims.

Claims

8 REQUESTS 1. Multilingual static and / or dynamic analysis of code changes and major language requirements. model-supported semantic interpretation to business processes and / or business capabilities mapping and automated tracking throughout the lifecycle in the corporate knowledge base. It enables updating on a section-by-section basis; -user's software development and approval or comment processes at least one application structured to enable it to perform (2) including and - Communicate with the application (2) using any communication protocol 10 and perform data exchange, saving and merging in the code repository. or listening to withdrawal request events, saving changes to retrieve and identify which files or functions have been changed the business meaning to be extracted from code changes, the business meaning summarized in natural language and documented in a format suitable for knowledge bases. 15 the creation of the generated document, coding the content on the platforms. Application programming that enables reading, creating, and managing files. adding or updating the relevant page through its interfaces, each matching the record's unique identification number with the document section and The creation of information about who did what and when, and the user's approval or 20 It is characterized by collecting feedback and enabling model improvement. a system (1).

2. To communicate with the server (3) using any communication protocol and 25 characterized by the application (2) which is structured to carry out data exchange A system like the one in Request 1 (1). 9 3. After the user develops their code, they can submit it to a versioning platform. The application is characterized by (2) which is configured to enable it to load. A system like the one in Request 1 or 2 (1).

4. Webhook or CI / CD pipeline trigger technologies and simple event queuing 5 with management and asynchronous processing algorithms performed in the code repository enable listening to save, merge or pull request events. from the above requests characterized by the server (3) configured for a system like any other (1).

5. In the structural analysis of the code, the AST (abstract syntax tree) represents the flow of business logic. CFG (control flow graph) and row-by-row changes in understanding Diff parsing algorithms are used to understand which changes are involved. Server configured to enable the determination of whether it affects functions (3) A system like any of the above-mentioned systems characterized by 15 (1).

6. LLM-based code interpretation, word embeddings, and cosine similarity. Corporate business concepts glossary with NLP-based matching and coding. LDA or BERTopic 20 enables changes to be linked to business processes. topic modeling algorithms such as these derive business meaning from code changes characterized by the server (3) configured to enable its extraction a system like any of the above requests (1).

7. Text summary in the form of Transformer-based models, standard format 25 Temple-based NLG and only, including a template and LLM hybrid approach. Longest Common Subsequence, which enables updating of changing sections. Business meaning is obtained by using delta update algorithms in the form of patch generation. summarized in natural language and documented in a format suitable for the knowledge base. characterized by the server (3) configured to enable its creation a system like any of the above requests (1).

8. On-page section anchoring and documentation with ID-based updates. 5 produced with version control algorithms that involve managing versions the document describes how content on platforms can be read, created, and processed using code. through application programming interfaces that enable its management, the relevant configured to allow addition or update to the page as in any of the above requests characterized by the server (3) a system (1). 10 9. Hash mapping and graph database algorithms identify each record as unique. matching the number with the document section and identifying who did what and when. with the server (3) configured to enable the creation of information A system like any of the above-mentioned demands characterized (1). 15 10. Active learning and reinforcement learning with human feedback algorithms with the collection of user approvals or comments and the model characterized by the server (3) configured to ensure its improvement a system like any of the above requests (1). 20 11. By using any communication method and any communication protocol To communicate with the corporate information repository and document platform and access data. to complete the transaction, records of changes made to the code in the relevant code 25 the above characterized by the server (3) configured to provide a system like any of the requests (1).

12. Static and dynamic semantic interpretation of code by artificial intelligence or machine. making sense of learning techniques and business logic and business 30 11 rules are configured to ensure that the work done by the code is removed. as in any of the above requests characterized by the server (3) a system (1).

13. The purpose, scope, affected parts, and data fields of the change are outlined in section 5. to ensure the creation of a summary and the production of a risk or impact analysis any of the above requests characterized by the configured server (3) a system like one of them (1).

14. By connecting to the document platform, if there is an existing document, the relevant work section 10 finding and detecting changes using the diff algorithm, detection transmitting the acquired changes to the generative pre-trained transformer model and implementing the changes, creating the new document. to be re-registered on the platform and for the user's review 15 configured to ensure that the application (2) interface is sent for approval. as in any of the above requests characterized by the server (3) a system (1).

15. In a preferred arrangement of the invention, the call graph and dependency analysis are very Cross-store or 20 that allows the real effect to be seen in service-oriented architectures. monorepo support and the coverage and risk score of affected functions. impact analysis using test scope correlation techniques characterized by the server (3) configured to enable deepening. a system like any of the above-mentioned requests (1).

16. In a preferred arrangement of the invention, code change registration, code merging A concept that enables the extraction of new business terms from job request and job registration text. mining, ontology or taxonomy changes are also monitored and documented. Versioning that enables reflection and automatic Turkish-English formatting. Concept management and 30 with multilingual term matching techniques that provide alignment 12 characterized by the server (3) which is configured to provide a rich ontology a system like any of the above requests (1).

17. In a preferred arrangement of the invention, the LLM output is converted to a JSON schema. Schematic-oriented manufacturing, which allows the design to be locked and then cast to a template, 5 regular expressions and rules, personal identifying information and keywords Content that enables error verification through scanning. Quality through verification and automated style and terminology control techniques. characterized by the server (3) configured to ensure its security a system like any of the above requests (1). 10 18. In a preferred arrangement of the invention, an approval and feedback interface and Enabling human learning through quality metrics. from the above requests characterized by the server (3) configured for a system like any other (1). 15 19. In a preferred arrangement of the invention, caching and iteration blocking, batch processing and windowing, and in-situ small model techniques. to ensure that performance and cost balance is achieved Any of the above requests characterized by the configured server (3) 20 a system like one of them (1).

20. In a preferred arrangement of the invention, role-based access control and Attribute-based access control, audit trail hardening, secret key management and input or output masking techniques for security and compliance 25 the above characterized by the server (3) configured to provide a system like any of the requests (1).

21. In a preferred arrangement of the invention, end-to-end tracking with open telemetry, 30 with service level indicator or service level target and business impact metrics 13 structured to enable monitoring of the observability of the flow as in any of the above requests characterized by the server (3) a system (1).