An AI conversation-oriented enterprise collaboration system intelligent distribution method and system

Abstract

The application discloses an AI conversation-oriented enterprise collaboration system intelligent distribution method, receives a natural language distribution instruction input by a user in an AI conversation interface, analyzes addressee description information from the natural language distribution instruction, extracts key feature words, queries an organization structure table through a preset collaboration system database connection module, matches the addressee description to a unique identification of a user in the collaboration system, and obtains a report file generated in the current AI conversation as a distribution attachment; a distribution task is created through an interface of the collaboration system, full-link logs from AI report generation to collaboration distribution are recorded, and a complete audit trace chain across systems is formed. The application solves the system fragmentation problem between AI report generation and OA collaboration distribution, and a user can complete cross-system distribution through only one natural language instruction.

Description

Technical Field

[0001] This invention relates to an intelligent distribution method and system, particularly an intelligent distribution method and system for enterprise collaboration systems oriented towards AI dialogue, belonging to the field of computer technology. Background Technology

[0002] With the increasing application of large-scale AI models in enterprise scenarios, AI agents are now capable of autonomously generating various business reports (such as sales analysis reports, audit reports, and due diligence reports). However, when users wish to distribute these AI-generated reports to other personnel within the enterprise, they face the following technical obstacles:

[0003] (1) Fragmented distribution process - Users need to download the report file from the AI ​​system first, then switch to the OA collaboration system, manually upload attachments, select recipients, and fill in distribution instructions. The whole process requires 4-6 system switching operations.

[0004] (2) Recipient identification needs to be accurately matched - Users in the OA system are usually identified by their employee ID or login name, while users are used to using Chinese names or job titles in natural language (such as "Manager Wang", "Engineer Zhang from the Finance Department"), so it is necessary to manually search for and match in the OA address book.

[0005] (3) Lack of intelligent distribution auditing - AI report generation and OA distribution are two independent systems operating independently, making it impossible to form a complete traceability chain from "who asked the question → what AI generated → who it was sent to".

[0006] (4) Unable to implement conversational distribution - The distribution operation of the existing OA system must be completed in its own interface and cannot be triggered by natural language commands in the AI ​​dialogue interface. Summary of the Invention

[0007] The technical problem to be solved by this invention is to provide an intelligent distribution method and system for enterprise collaborative systems oriented towards AI dialogue. By parsing the user's distribution intent and the recipient's natural language description, the system automatically queries the organizational structure data of the enterprise collaborative office system to match the target recipient, and creates a distribution task through the collaborative system interface by attaching the AI-generated report file.

[0008] To solve the above-mentioned technical problems, the technical solution adopted by the present invention is as follows: A smart distribution method for enterprise collaboration systems oriented towards AI dialogue includes the following steps: S1. Receive natural language distribution instructions input by the user in the AI ​​dialogue interface. The natural language distribution instructions include distribution target and recipient description information. S2. Parse the recipient description information from the natural language distribution instructions and extract key feature words, which include Chinese name, job title and department; S3. Query the organizational structure table through the pre-set collaborative system database connection module and match the recipient description with the user's unique identifier in the collaborative system; S4. Obtain the report file generated in the current AI dialogue and distribute it as an attachment; S5. Create a distribution task through the interface of the collaborative system. The distribution task is associated with a unique user identifier and distribution attachments. S6. Record the entire chain of logs from AI report generation to collaborative distribution, forming a complete audit traceability chain across systems.

[0009] Further, step S2 specifically includes: S2.1 Keyword extraction: Extract three types of keywords from the recipient description information using regular expressions; S2.2 OA User Table Query: The organizational structure data of the OA system is mainly stored in the following three tables: user basic information table, department information table, and user-department association table. The system establishes a connection to the OA database through the oa_vpd.py shortcut module, constructs multi-condition SQL joint queries, and uses LIKE fuzzy matching for names and departments in the query conditions. Status filtering only queries current employees. If the user command does not specify a department, the department filtering condition is omitted, expanding the search scope to the entire company. S2.3 Fuzzy matching and disambiguation: If the query returns multiple matching results, the system selects the best match according to the following priority: First priority: exact match of job title keywords; Second priority: name edit distance; Third priority: most recent active time; The result ranked first is used as the suggested recipient and is confirmed to the user in the AI ​​reply. After the user confirms, the distribution process begins; If the user denies it, the top 3 candidate recipients are displayed for selection. S2.4. Unique match with automatic confirmation: If the query returns only one matching result, the system skips the user confirmation step and automatically confirms the recipient's identity.

[0010] Furthermore, the three types of keywords in step S2.1 include: (a) Chinese name: 2-3 consecutive Chinese characters, the core name is obtained by removing the job title suffix; (b) Job keywords: Match with a predefined dictionary of job keywords, which contains common corporate job titles; (c) Department keywords: Match against a predefined dictionary of department keywords, which contains common company department names.

[0011] Furthermore, in step S2.2, the user basic information table includes user ID, login_name, user name, email, and employment status; the department information table includes department ID, department name, and parent department ID; and the user-department association table includes user ID, department ID, and whether it is a parent department flag.

[0012] Further, step S4 specifically involves: the report files generated during the AI ​​dialogue are stored in the static / files / directory of the web server, with the filenames containing the conversation_id prefix of the current dialogue; the system scans all files in this directory, filters out files whose filenames match the current dialogue ID and are in common report formats, sorts them in descending order of file creation time, and takes the most recently created file as the distribution attachment; if the AI ​​generates multiple files in this dialogue, the latest one is selected; if the user needs to distribute a specific file, the file type is specified in the distribution instruction.

[0013] Furthermore, step S5 specifically involves: the OA system supporting the creation of collaborative work items via direct database write, without requiring the OA system to provide an API interface; and the system writing distribution records into the following three tables of the OA database: (a) Insert a collaborative event record into the COLLABORATION table using INSERT INTO VALUES, with fields including initiator ID, title, body content, and creation time; (b) Insert an attachment association record into the COLLABORATION_ATTACHMENT table using INSERT INTO VALUES, with fields including the collaboration item ID, file name, file storage path on the server, and file size; (c) Insert a recipient record into the COLLABORATION_RECEIVER table using INSERT INTO VALUES. The fields include the collaborative item ID, the recipient user ID, the reading status, and the receiving time. The write operation to all three tables is completed within the same database transaction; After the distribution task is successfully created, the system returns a distribution result summary to the AI ​​dialogue. The distribution result summary includes the recipient's name and department, OA collaborative task number, sending time and file name.

[0014] Further, step S6 specifically includes: S6.1 AI-side audit records: Add audit records for distribution event types to the existing audit_log table. The audit_log table will add the following extended fields: matched recipient OA user ID distribution_target, distributed report file name distribution_file, distribution status distribution_status, and OA collaboration item number oa_collaboration_id. Each distribution operation will be recorded as an audit log entry in the audit_log. S6.2 OA side distribution logs: Create an independent AI_DISTRIBUTION_LOG table in the OA database. The AI_DISTRIBUTION_LOG table records complete information for each AI distribution and is associated with the AI ​​system's audit_log through two foreign keys, conversation_id and collaboration_id. S6.3 Traceability Query: Administrators can view the complete distribution traceability chain through the Web audit panel: User query content → AI-generated report summary → Recipient parsing process → OA distribution status → Recipient reading status; The audit panel supports combined queries by sender, recipient, and time range, and supports one-click export of the complete distribution audit report.

[0015] A system for implementing an intelligent distribution method for an AI-oriented dialogue-based enterprise collaboration system, comprising The instruction parsing module is used to extract recipient descriptions and distribution intents from user natural language instructions; The recipient matching module is used to query the collaborative system's organizational structure database to match the recipient description with the system user identifier. The file location module is used to retrieve the report files generated in the current AI dialogue. The distribution execution module is used to create distribution tasks containing recipients and attachments through the collaboration system interface; The audit association module is used to record and associate the distribution logs of the AI ​​system and the collaborative system, and supports cross-system traceability and query.

[0016] Compared with the prior art, the present invention has the following advantages and effects: 1. Distribution efficiency: Traditional solutions require 4-6 system switching operations, taking an average of 3-5 minutes; this invention only requires inputting a single natural language command, taking about 10-20 seconds, improving efficiency by more than 90%. 2. Recipient matching accuracy: Based on a recipient parsing algorithm with multi-condition SQL queries and priority sorting, the matching accuracy can reach over 95%; 3. Audit integrity: A dual-system audit link is established through conversation_id and collaboration_id to form a complete traceability chain from user inquiry to recipient reading; 4. Low user operation threshold: The distribution operation is simplified to a single natural language command, and non-technical users do not need OA operation training; 5. System integration cost: The OA system is integrated through direct database writing, eliminating the need for OA vendor API interfaces or custom development. Attached Figure Description

[0017] Figure 1 This is an overall architecture diagram of an intelligent distribution method for an enterprise collaboration system oriented towards AI dialogue, according to the present invention.

[0018] Figure 2 This is a flowchart illustrating step S2 of an intelligent distribution method for an AI-oriented enterprise collaboration system according to the present invention.

[0019] Figure 3 This is a task distribution sequence diagram for step S5 of the intelligent distribution method for an AI-oriented enterprise collaboration system according to the present invention.

[0020] Figure 4 This is a schematic diagram illustrating step S6 of the intelligent distribution method for an AI-oriented enterprise collaboration system according to the present invention. Detailed Implementation

[0021] To illustrate in detail the technical solutions adopted by the present invention to achieve the intended technical objectives, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Furthermore, the technical means or technical features in the embodiments of the present invention can be replaced without creative effort. The present invention will be described in detail below with reference to the accompanying drawings and embodiments.

[0022] like Figure 1 As shown, this invention adds an OA collaborative distribution bridging layer to the existing AI dialogue platform, forming a four-layer architecture.

[0023] The first layer is the user dialogue layer. Users interact with the AI ​​agent through a web browser using natural language. During the dialogue, users can issue report distribution commands in natural language at any time, such as "Please send this sales analysis report to Manager Wang in the finance department." This layer is responsible for receiving user input and displaying the AI ​​response to the user in real time, supporting SSE streaming push and Markdown rendering.

[0024] The second layer is the AI ​​agent layer. Based on Claude Code CLI subprocess bridging technology, the AI ​​agent has the following capabilities: understanding the user's distribution intent, parsing the recipient's natural language description in the instruction, locating the generated report file, and calling the OA bridging layer interface. The AI ​​model can autonomously generate SQL statements to query the OA organizational structure through the OA data dictionary index (934 tables) injected into the system prompts.

[0025] The third layer is the OA bridging layer. The newly added OADistributionService module is the core innovative component of this invention, responsible for: extracting key feature words (name, position, department) from the recipient's description; performing organizational structure queries by connecting to the OA database via the oa_vpd.py shortcut module; matching feature words as unique identifiers for OA system users; and creating collaborative distribution tasks in the OA database. This module acts as a bridging adapter between the AI ​​agent and the OA collaborative system, encapsulating the details of OA database read / write operations and exposing a concise Python function interface.

[0026] The fourth layer is the OA Collaboration System layer. The Oracle database of the Seeyon Collaborative OA system (192.168.1.122:1521 / jhcsoa26) stores organizational structure information (ORG_USER table, ORG_DEPARTMENT table, ORG_MEMBER table) and collaborative work item data (COLLABORATION table, COLLABORATION_ATTACHMENT table, COLLABORATION_RECEIVER table). This layer requires no code modification; task distribution can be created simply through standard SQL read and write operations.

[0027] The present invention provides an intelligent distribution method for an AI-driven dialogue-based enterprise collaboration system, comprising the following steps: S1. Receive natural language distribution instructions input by the user in the AI ​​dialogue interface. The natural language distribution instructions include distribution target and recipient description information.

[0028] S2. Parse the recipient description information from the natural language distribution instructions and extract key feature words, which include Chinese name, job title and department.

[0029] like Figure 2 As shown, when a user enters a distribution instruction like "Please send this sales report to Manager Wang in the Finance Department" in the AI ​​dialogue, the system first extracts key information through AI semantic understanding: action="send", target file="sales report", and recipient description="Manager Wang in the Finance Department". Then, it executes the following recipient parsing steps.

[0030] S2.1, keyword extraction: three types of keywords are extracted from the recipient description information by regular expression: (a) Chinese names: 2-3 consecutive Chinese characters, the core name is obtained by removing postfixes of positions such as "manager", "supervisor", "director" and "director", for example, "Manager Wang" is extracted as "Wang"; (b) position keywords: matching is performed with a predefined position keyword dictionary, and the core name is obtained by removing postfixes of positions such as "manager", "supervisor", "director" and "director", for example, "Manager Wang" is extracted as "Wang"; (c) department keywords: matching is performed with a predefined department keyword dictionary, and the dictionary includes common enterprise department names such as Finance Department, Human Resources Department, Administration Department, Sales Department, Marketing Department, R&D Department, Technology Department, Production Department, Purchasing Department, Quality Department, etc.

[0031] S2.2, OA user table query: the organizational structure data of the OA system is mainly stored in the following three tables: basic user information table, department information table and association table of users and departments; the basic user information table includes user ID, login name login_name, user name user_name, email email and in-service status status; the department information table includes department ID, department name dept_name and superior department ID; the association table of users and departments includes user ID, department ID and the flag indicating whether it is the primary department.

[0032] The system establishes a connection to the OA database through the shortcut module oa_vpd.py (from oa_vpd import get_connection; conn, cur = get_connection()), and constructs a multi-condition SQL joint query: SELECT u.id,u.login_name, u.user_name, d.dept_name FROM ORG_USER u JOIN ORG_MEMBER m ON u.id=m.user_id JOIN ORG_DEPARTMENT d ON m.dept_id=d.id WHERE u.user_name LIKE'%Wang%' AND d.dept_name LIKE '%Finance%' AND u.status='In-service'. In the query conditions, the name uses LIKE fuzzy matching (%Wang%), the department uses LIKE fuzzy matching (%Finance%), and the status filter only queries in-service employees. If the user's instruction does not specify a department, the department filtering condition is omitted, and the search scope is expanded to the whole company.

[0033] In step S2.2, the user basic information table includes user ID, login_name, user name, email, and employment status; the department information table includes department ID, department name, and parent department ID; the user-department association table includes user ID, department ID, and whether it is a parent department flag.

[0034] S2.3 Fuzzy Matching and Disambiguation: If the query returns multiple matching results (e.g., two users with the surname Wang, Wang Jianguo and Wang Lihua, in the same department), the system selects the best match based on the following priorities: First priority – exact match of job title keywords (if the user said "Manager Wang," then users whose job title field contains "Manager" are prioritized); Second priority – name edit distance (calculate the Levenshtein edit distance between the extracted name and the database user's name, prioritizing the closest user with the smallest distance); Third priority – most recent active time (users who recently logged into the OA system are prioritized). The top-ranked result is used as the suggested recipient, and the AI ​​reply confirms with the user: "We have matched you with Wang Jianguo (username wangjg) in the Finance Department. Confirm sending?" After the user confirms, the distribution process begins; if the user denies, the top 3 candidate recipients are displayed for selection.

[0035] S2.4. Unique Match Automatic Confirmation: If the query returns only one matching result, the system skips the user confirmation step and automatically confirms the recipient's identity, directly informing the system in the AI ​​reply: "The report has been sent to Wang Jianguo in the Finance Department." This automatic confirmation mechanism reduces one round of dialogue when the recipient description is accurate enough, improving the user experience.

[0036] S3. Query the organizational structure table through the pre-installed collaborative system database connection module and match the recipient description with the user's unique identifier in the collaborative system.

[0037] S4. Obtain the report file generated in the current AI dialogue and distribute it as an attachment.

[0038] Report files generated during AI dialogues are stored in the `static / files / ` directory on the web server, with filenames prefixed with the `conversation_id` of the current dialogue. The system scans all files in this directory, filters out files whose filenames match the current dialogue ID and are in common report formats (.html, .docx, .xlsx, .pdf), sorts them in descending order of creation time, and selects the most recently created file as the distribution attachment. If the AI ​​generates multiple files in this dialogue, the latest one is selected. If the user needs to distribute a specific file, they specify the file type in the distribution instruction (e.g., "Send the Excel spreadsheet to Manager Wang").

[0039] S5. Create a distribution task through the interface of the collaborative system. The distribution task is associated with a unique user identifier and distribution attachments. like Figure 3 As shown, the OA system supports creating collaborative work items via direct database writes, without requiring the OA system to provide an API interface. The system writes distribution records to the following three tables in the OA database: (a) Insert collaborative event records into the COLLABORATION table using INSERT INTO VALUES. Fields include initiator ID (mapped from the NC65 user code of the current AI dialogue user to the user ID in the OA system), title (formatted as "AI Agent Report: {Report File Subject}", extracted from the file name or dialogue title), body content (including report summary description and user distribution instructions), and creation time (taken as the current system time). (b) Insert the associated attachment record into the COLLABORATION_ATTACHMENT table using INSERT INTO VALUES. The fields include the collaborative item ID (the auto-incrementing ID returned by the previous INSERT), the file name, the file's storage path on the server, and the file size (in bytes). (c) Insert a recipient record into the COLLABORATION_RECEIVER table using INSERT INTO VALUES. The fields include the collaborative item ID, the recipient user ID (parsed OA user ID), the read status (initially "unread"), and the receipt time. The three tables are written within the same database transaction, ensuring the atomicity of the task distribution—either all three tables are written successfully, or all are rolled back.

[0040] After the distribution task is successfully created, the system returns a distribution result summary to the AI ​​dialogue. The summary includes the recipient's name and department, the OA collaborative task number (formatted as "XT-2026-0616-001"), the sending time, and the file name. After logging into the OA system, the recipient will see the collaborative notification in the "To-Do Items" area on the homepage. Clicking on it allows them to view the AI-generated report online (HTML format rendered directly in the OA system) or download the attachment file (DOCX / XLSX / PDF format). Simultaneously, the OA system's message notification mechanism (such as integration with WeChat / DingTalk) will also push the distribution notification to the recipient.

[0041] S6. Record the entire chain of logs from AI report generation to collaborative distribution, forming a complete audit traceability chain across systems.

[0042] S6.1 AI-side audit records: Add audit records for distribution event types to the existing audit_log table. The audit_log table will add the following extended fields: matched recipient OA user ID distribution_target, distributed report file name distribution_file, distribution status distribution_status, and OA collaboration item number oa_collaboration_id. Each distribution operation will record an audit log entry with role='system' and risk_level='low' in the audit_log. S6.2 OA terminal distributes logs. Create an independent AI_DISTRIBUTION_LOG table in the OA database. The structure of the AI_DISTRIBUTION_LOG table is: CREATE TABLE AI_DISTRIBUTION_LOG (id NUMBER PRIMARYKEY, conversation_id VARCHAR2(36), sender_user_id VARCHAR2(100), receiver_user_id VARCHAR2(100), file_name VARCHAR2(500), file_size NUMBER,collaboration_id VARCHAR2(100), status VARCHAR2(50), created_at TIMESTAMPDEFAULT SYSTIMESTAMP).

[0043] The AI_DISTRIBUTION_LOG table records complete information for each AI distribution and is associated with the AI ​​system's audit_log through two foreign keys, conversation_id and collaboration_id, to achieve dual system log association.

[0044] S6.3 Traceability Query: Administrators can view the complete distribution traceability chain through the Web Audit Panel: User Request Content → AI-Generated Report Summary → Recipient Parsing Process (including keyword extraction and matching results) → OA Distribution Status → Recipient Reading Status (obtained by querying the reading status field of the COLLABORATION_RECEIVER table); The Audit Panel supports combined queries by sender (AI dialogue user), recipient (OA user), and time range, and supports one-click export of the complete distribution audit report (JSON format).

[0045] A system for implementing an intelligent distribution method for an AI-oriented dialogue-based enterprise collaboration system, comprising The instruction parsing module is used to extract recipient descriptions and distribution intents from user natural language instructions; The recipient matching module is used to query the collaborative system's organizational structure database to match the recipient description with the system user identifier. The file location module is used to retrieve the report files generated in the current AI dialogue. The distribution execution module is used to create distribution tasks containing recipients and attachments through the collaboration system interface; The audit association module is used to record and associate the distribution logs of the AI ​​system and the collaborative system, and supports cross-system traceability and query.

[0046] Compared with existing technologies, this invention has the following advantages and effects: 1. Distribution efficiency: Traditional solutions require 4-6 system switching operations, taking an average of 3-5 minutes; this invention only requires inputting a single natural language command, taking approximately 10-20 seconds, improving efficiency by over 90%; 2. Recipient matching accuracy: Based on a recipient parsing algorithm using multi-condition SQL queries and priority sorting, the matching accuracy can reach over 95%; 3. Audit integrity: By establishing a dual-system audit association through conversation_id and collaboration_id, a complete traceability chain is formed from user inquiry to recipient reading; 4. User operation threshold: The distribution operation is reduced to a single natural language command, eliminating the need for OA operation training for non-technical users; 5. System integration cost: The OA system is integrated through direct database writing, eliminating the need for OA vendor API interfaces or custom development.

[0047] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention in any way. Although the present invention has been disclosed above with reference to preferred embodiments, it is not intended to limit the present invention. Any person skilled in the art can make some modifications or alterations to the above-disclosed technical content to create equivalent embodiments without departing from the scope of the present invention. Any simple modifications, equivalent substitutions, and improvements made to the above embodiments without departing from the scope of the present invention, based on the technical essence of the present invention and within the spirit and principles of the present invention, shall still fall within the protection scope of the present invention.

Claims

1. A method for intelligent distribution in an AI-driven enterprise collaboration system, characterized in that... Includes the following steps: S1. Receive natural language distribution instructions input by the user in the AI ​​dialogue interface. The natural language distribution instructions include distribution target and recipient description information. S2. Parse the recipient description information from the natural language distribution instructions and extract key feature words, which include Chinese name, job title and department; S3. Query the organizational structure table through the pre-set collaborative system database connection module and match the recipient description with the user's unique identifier in the collaborative system; S4. Obtain the report file generated in the current AI dialogue and distribute it as an attachment; S5. Create a distribution task through the interface of the collaborative system. The distribution task is associated with a unique user identifier and distribution attachments. S6. Record the entire chain of logs from AI report generation to collaborative distribution, forming a complete audit traceability chain across systems.

2. The intelligent distribution method for an AI-driven dialogue-oriented enterprise collaboration system according to claim 1, characterized in that: Step S2 specifically involves: S2.1 Keyword extraction: Extract three types of keywords from the recipient description information using regular expressions; S2.2 OA User Table Query: The organizational structure data of the OA system is mainly stored in the following three tables: user basic information table, department information table, and user-department association table. The system establishes a connection to the OA database through the oa_vpd.py shortcut module, constructs multi-condition SQL joint queries, and uses LIKE fuzzy matching for names and departments in the query conditions. Status filtering only queries current employees. If the user command does not specify a department, the department filtering condition is omitted, expanding the search scope to the entire company. S2.3 Fuzzy matching and disambiguation: If the query returns multiple matching results, the system selects the best match according to the following priority: First priority: exact match of job title keywords; Second priority: name edit distance; Third priority: most recent active time; The result ranked first is used as the suggested recipient and is confirmed with the user in the AI ​​reply. After the user confirms, the distribution process begins. If the user rejects the request, the first three candidate recipients will be displayed for selection. S2.

4. Unique match with automatic confirmation: If the query returns only one matching result, the system skips the user confirmation step and automatically confirms the recipient's identity.

3. The intelligent distribution method for an AI-driven dialogue-oriented enterprise collaboration system according to claim 2, characterized in that: The three types of keywords in step S2.1 include: (a) Chinese name: 2-3 consecutive Chinese characters, the core name is obtained by removing the job title suffix; (b) Job keywords: Match with a predefined dictionary of job keywords, which contains common corporate job titles; (c) Department keywords: Match against a predefined dictionary of department keywords, which contains common company department names.

4. The intelligent distribution method for an AI-driven dialogue-oriented enterprise collaboration system according to claim 2, characterized in that: In step S2.2, the user basic information table includes user ID, login_name, user name, email, and employment status; the department information table includes department ID, department name, and parent department ID; the user-department association table includes user ID, department ID, and whether it is a parent department flag.

5. The intelligent distribution method for an AI-driven dialogue-oriented enterprise collaboration system according to claim 1, characterized in that: Step S4 specifically involves: the report files generated during the AI ​​dialogue are stored in the static / files / directory of the web server, with the filenames containing the conversation_id prefix of the current dialogue; the system scans all files in this directory, filters out files whose filenames match the current dialogue ID and are in common report formats, sorts them in descending order of file creation time, and takes the most recently created file as the distribution attachment; if the AI ​​generates multiple files in this dialogue, the latest one is selected; if the user needs to distribute a specific file, the file type is specified in the distribution instruction.

6. The intelligent distribution method for an AI-driven dialogue-oriented enterprise collaboration system according to claim 1, characterized in that: Step S5 specifically involves: The OA system supports creating collaborative work items via direct database writing, without requiring the OA system to provide an API interface. The system writes distribution records into the following three tables in the OA database: (a) Insert collaborative event records into the COLLABORATION table using INSERT INTO VALUES, with fields including initiator ID, title, body content, and creation time; (b) Insert an attachment association record into the COLLABORATION_ATTACHMENT table using INSERT INTO VALUES, with fields including the collaboration item ID, file name, file storage path on the server, and file size; (c) Insert a recipient record into the COLLABORATION_RECEIVER table using INSERT INTO VALUES. The fields include the collaborative item ID, the recipient user ID, the reading status, and the receiving time. The write operation to all three tables is completed within the same database transaction; After the distribution task is successfully created, the system returns a distribution result summary to the AI ​​dialogue. The distribution result summary includes the recipient's name and department, OA collaborative task number, sending time and file name.

7. The intelligent distribution method for an AI-driven dialogue-oriented enterprise collaboration system according to claim 1, characterized in that: Step S6 specifically involves: S6.1 AI-side audit records: Add audit records for distribution event types to the existing audit_log table. The audit_log table will add the following extended fields: matched recipient OA user ID distribution_target, distributed report file name distribution_file, distribution status distribution_status, and OA collaboration item number oa_collaboration_id. Each distribution operation will be recorded as an audit log entry in the audit_log. S6.2 OA side distribution logs: Create an independent AI_DISTRIBUTION_LOG table in the OA database. The AI_DISTRIBUTION_LOG table records complete information for each AI distribution and is associated with the AI ​​system's audit_log through two foreign keys, conversation_id and collaboration_id. S6.3 Traceability Query: Administrators can view the complete distribution traceability chain through the Web audit panel: User query content → AI-generated report summary → Recipient parsing process → OA distribution status → Recipient reading status; The audit panel supports combined queries by sender, recipient, and time range, and supports one-click export of the complete distribution audit report.

8. A system for executing the intelligent distribution method for an AI-oriented dialogue-based enterprise collaboration system according to any one of claims 1-7, characterized in that: Include The instruction parsing module is used to extract recipient descriptions and distribution intents from user natural language instructions; The recipient matching module is used to query the collaborative system's organizational structure database to match the recipient description with the system user identifier. The file location module is used to retrieve the report files generated in the current AI dialogue. The distribution execution module is used to create distribution tasks containing recipients and attachments through the collaboration system interface; The audit association module is used to record and associate the distribution logs of the AI ​​system and the collaborative system, and supports cross-system traceability and query.

Images