Search

CN-121659977-B - Multi-source data knowledge base supporting real-time synchronization and intelligent agent system

CN121659977BCN 121659977 BCN121659977 BCN 121659977BCN-121659977-B

Abstract

The invention relates to the technical field of knowledge graph real-time synchronization and discloses a multisource data knowledge base supporting real-time synchronization and an intelligent agent system. The system comprises an event access module, a content deconstructing module, an entity matching module, a data anchoring module, a synchronous decision module and a network executing module. The system deconstructs semantics, data, and identifiers by receiving the multi-source event stream in real time and matches or creates entities and relationships in a dynamic entity relationship network. By anchoring the data version to the entity and intelligently generating a synchronous decision instruction according to the data source and the operation type, the instant reconstruction and update of the network topology are driven. The system realizes millisecond-level real-time evolution of the knowledge base, ensures logic consistency of multi-source data fusion through strategic synchronous coordination, and improves instantaneity and reliability of knowledge support.

Inventors

  • LUO GENSHENG

Assignees

  • 杭州开云集致科技有限公司

Dates

Publication Date
20260512
Application Date
20260205

Claims (9)

  1. 1. A multi-source data intelligent agent system supporting real-time synchronization, comprising the following processing steps: The event access module receives original event streams containing time stamps from a plurality of heterogeneous data access terminals, wherein the original event streams carry data source identifications and event operation types; the content deconstructing module carries out content deconstructing on the original event stream and separates a semantic description field, a structured data main body and an associated identifier set; The entity matching module is used for matching or creating entity nodes and relationship edges in a dynamic entity relationship network based on the semantic description field and the association identifier set, wherein the dynamic entity relationship network is a core data topology maintained during system operation; the data anchoring module integrates the structured data main body and the timestamp into a versioned attribute block and anchors the versioned attribute block to corresponding entity nodes; the synchronous decision module determines a synchronous decision mode to be adopted according to the data source identification and the event operation type, and generates a synchronous decision instruction which is used for driving the state transition of the entity node and the relation edge in the dynamic entity relation network; The network execution module is used for executing the synchronous decision instruction, completing the topology reconstruction and attribute update of the dynamic entity relation network and outputting a network state snapshot to the proxy decision unit; The synchronous decision mode comprises an instant strong synchronous mode, a delayed asynchronous synchronous mode and a conflict arbitration mode; When the comprehensive synchronization priority belongs to the highest threshold interval, adopting an instant strong synchronization mode, generating a synchronization decision instruction which requires to lock related entity nodes immediately, executing attribute updating and relation edge creation in an exclusive mode, and blocking other operation requests of the related nodes and the relation edges before execution is finished; When the comprehensive synchronous priority score belongs to the middle threshold interval, adopting a delay asynchronous synchronous mode, packaging related operations into asynchronous tasks by using a generated synchronous decision instruction, putting the asynchronous tasks into a priority task queue for scheduling and executing, and immediately returning a response received by the operations; When the comprehensive synchronization priority belongs to the lowest threshold value area or version conflict of attribute update aiming at the same entity node is detected, adopting a conflict arbitration mode; the specific substeps of generating a synchronization decision instruction in a collision arbitration mode include: extracting the structured data body, the data source identification, the time stamp and the version serial number from each candidate event which generates conflict; sorting the data sources of all the conflicting parties according to the data source credibility table; Preferentially adopting event versioning attribute blocks corresponding to the highest-ranking data sources; If the highest data source credibility is the same, adopting an event versioning attribute block with the latest time stamp; if the time stamps are the same, adopting an event versioning attribute block with a larger version serial number; An explicit arbitration result instruction is generated that anchors the adopted versioned attribute block to the subject node and logs the arbitration log.
  2. 2. The intelligent proxy system for multi-source data supporting real-time synchronization according to claim 1, wherein said implementation of content deconstructing the original event stream comprises: invoking a predefined event mode rule set to filter an original event stream, and screening out candidate events related to a preset entity model; Performing nested structure expansion on the original load of the candidate event, and recursively extracting key value pairs of all leaf levels; Grouping key value pairs according to the name space to form a structured data main body set, wherein keys of the structured data main body comprise field, entity type and attribute name information; Resolving a semantic description field from the metadata header of the candidate event, wherein the semantic description field comprises event abstract text, a causal chain identifier and an operation intention code; And extracting a set of associated identifiers from the reference field of the candidate event, wherein the set of associated identifiers at least comprises a main entity identifier, a related entity identifier list and a context session identifier.
  3. 3. The intelligent agent system for multi-source data supporting real-time synchronization according to claim 2, wherein the step of matching or creating entity nodes and relationship edges in the dynamic entity relationship network based on the semantic description field and the associated identifier set specifically comprises: comparing the main entity identifier in the association identifier set with the unique identifiers of all entity nodes existing in the dynamic entity relationship network; If the comparison is successful, locating the entity node as a main body node of the current operation; if the comparison fails, creating an entity node with a new unique identifier in the dynamic entity relation network according to the field and entity type information in the semantic description field, and setting the new entity node as a main body node; Traversing a related entity identifier list in the associated identifier set, and executing a comparison and creation process for each identifier in the list to obtain a related object node list; Defining or updating a directed relationship edge pointing to each object node from the host node in the dynamic entity relationship network according to the operation intention code and the event abstract text in the semantic description field; the type, intensity weight, creation or update time information of a directed relationship edge is bound as an attribute to the directed relationship edge.
  4. 4. A multi-source data intelligent agent system supporting real-time synchronization according to claim 3, wherein said step of integrating structured data bodies with time stamps into versioned property blocks and anchoring to corresponding entity nodes comprises: Generating a globally unique version sequence number for each candidate event; Extracting event time stamps accurate to millisecond level from an original event stream; packaging the event time stamp, the version serial number and the structured data main body corresponding to the candidate event to form a version attribute block; the versioned property block has immutable properties whose content, once generated, is not allowed to be modified; Determining a main body node of the current operation, and linking the generated versioned attribute block to an attribute history chain of the main body node in an additional mode; The attribute history chain is a linked list which is arranged in strict ascending order according to the version serial number and is used for completely recording all change histories of the entity node attributes.
  5. 5. The intelligent agent system for multi-source data supporting real-time synchronization according to claim 4, wherein said step of generating a synchronization decision command according to the data source identifier and the event operation type comprises: Maintaining a data source reputation table, wherein the historical accuracy, data freshness grade and weight coefficient corresponding to each data source identifier are recorded in the table; maintaining an operation type influence matrix, wherein the matrix defines influence weights of different event operation types on different elements in a dynamic entity relation network; inquiring a data source credibility table to obtain a data freshness grade and a data source weight coefficient corresponding to the current data source identifier; Inquiring an operation type influence matrix to obtain operation influence weights of the current event operation type on the main body node, the directed relation edge and the attribute history chain; based on the data freshness level, the data source weight coefficient and the operation influence weight, calculating to obtain the comprehensive synchronous priority score of the current candidate event; Comparing the comprehensive synchronous priority score with a plurality of preset score threshold intervals to determine a synchronous decision mode to be adopted; according to the selected synchronous decision mode, combining the information of the current subject node, the object node list, the directed relation edge and the versioning attribute block, and generating a synchronous decision instruction containing specific operation objects, operation actions and execution parameters.
  6. 6. The intelligent agent system for multi-source data supporting real-time synchronization according to claim 5, wherein said executing synchronization decision instructions to complete topology reconfiguration and attribute updating of the dynamic entity relationship network comprises: analyzing the synchronous decision instruction, and acquiring the instruction type, the operation object list and the execution parameters; If the instruction type is in the instant strong synchronous mode, immediately applying a write lock to the entity node appointed in the operation object list, adding the versioned attribute block to the attribute history chain of the node according to the operation action in the instruction, creating or updating the appointed directional relation edge, and releasing the write lock after completion; if the instruction type is a delayed asynchronous synchronous mode, packaging the instruction, the related versioning attribute blocks and the network state context into an asynchronous task unit, and inserting the asynchronous task unit into the corresponding position of a priority task queue according to the comprehensive synchronous priority score of the task; An independent asynchronous execution thread continuously monitors the priority task queue, sequentially takes out the asynchronous task units to execute the operations in the priority task queue, does not apply node locks in the execution process, and needs to check the consistency of node states before and after the operations; If the instruction type is in a conflict arbitration mode, directly applying an arbitration result instruction, linking the adopted versioned attribute block to an attribute history chain of the main body node, and updating the state of the related directed relation edge; After any instruction type operation is completed, the system traverses the modified entity node and the directed relation edge directly related to the modified entity node, recalculates the local consistency hash value of the network element, and stores the local consistency hash value in association with the network global version number.
  7. 7. The intelligent agent system for multi-source data supporting real-time synchronization according to claim 6, wherein said step of outputting a snapshot of network status to a proxy decision unit comprises: Triggering a snapshot generating flow after updating the network global version number; Extracting entity nodes and directed relation edges with all changed states from the last snapshot from a dynamic entity relation network; generating a simplified representation of each changed entity node, wherein the simplified representation comprises a unique identifier of the entity node, a key content abstract of the latest version attribute block, and types and target node identifiers of all directed relation edges directly connected with the entity node; Generating a simplified representation of each changed directed relation edge, wherein the simplified representation comprises a starting point node identification, an ending point node identification, an edge type, a current intensity weight and an updating time of the directed relation edge; encapsulating the simplified representation of all the change elements together with the current network global version number and snapshot generation time stamp into a network state snapshot message; the network state snapshot message is actively pushed to the subscribed proxy decision unit through the network communication interface.
  8. 8. The intelligent agent system for supporting real-time synchronization according to claim 7, wherein when the asynchronous execution thread fetches an asynchronous task unit from a priority task queue and performs a rollback and retry mechanism if it is checked that the current state of the relevant entity node is inconsistent with the network state context stored in the asynchronous task unit, the specific substeps include: immediately suspending the execution of the current asynchronous task unit, and marking the asynchronous task unit as invalid; Recording the related information of the failure task unit, including the operation target, the original synchronous decision instruction and the failure reason, to an abnormal task log; recalculating the current comprehensive synchronous priority score according to the data source identifier and the event operation type stored in the failure task unit; creating a new asynchronous task unit based on the newly calculated comprehensive synchronous priority score, the new asynchronous task unit containing context extracted from the latest network state; reinserting the new asynchronous task unit into the priority task queue, the position of which is determined by the new comprehensive synchronous priority score; If the same original event repeatedly triggers the rollback and retry mechanism within the preset retry times, the system will forcedly raise the next comprehensive synchronization priority score to make it enter the instant strong synchronization mode for processing so as to ensure the final consistency.
  9. 9. A multi-source data knowledge base supporting real-time synchronization, which is applied to a multi-source data intelligent agent system supporting real-time synchronization according to any one of claims 1 to 8, and is characterized in that the knowledge base comprises a dynamic entity relation network constructed and maintained in running and all version attribute blocks thereof, and stores a data source reputation table, an operation type influence matrix, a priority task queue and all network state snapshot histories, so as to provide a structured data storage basis for data synchronization and agent decision of the multi-source data intelligent agent system supporting real-time synchronization.

Description

Multi-source data knowledge base supporting real-time synchronization and intelligent agent system Technical Field The invention relates to the technical field of knowledge graph real-time synchronization, in particular to a multisource data knowledge base supporting real-time synchronization and an intelligent agent system. Background In the prior art, a method for constructing a static knowledge graph based on a batch ETL flow is generally adopted for integrating multi-source heterogeneous data. The method needs to define a complete mode in advance, and periodically executes data extraction, conversion and loading tasks to finally form a knowledge base with a fixed structure. The process causes a significant delay from the generation of data to the availability of knowledge, and the knowledge base state lags behind the real change of the data source, so that the real-time application scene with strict requirements on the timeliness of the data can not be satisfied. Existing schemes typically rely on a timestamp-based "last write priority" rule or simple predefined merge logic when handling multi-source data updates and conflicts. The method regards different data sources as homogeneous information providers, and ignores the essential differences of the data sources in service authority, credibility and event operation intention. This results in the easy generation of data states that violate business logic, compromising the overall consistency and reliability of the knowledge base, and difficulty in supporting intelligent decisions based on accurate, consistent knowledge in the event of concurrent write or operation conflicts. A technical solution is needed that enables the transition of a knowledge base from static storage to a dynamic, real-time evolution system. The scheme can continuously receive event streams and reconstruct an internal knowledge network in real time, and can intelligently decide according to the identity of a data source and the operation type so as to ensure the logic consistency of multi-source data in a high-speed synchronization process. Disclosure of Invention The invention aims to provide a multisource data knowledge base supporting real-time synchronization and an intelligent agent system so as to solve the problems in the background technology. To achieve the above object, the present invention provides a multi-source data intelligent agent system supporting real-time synchronization, the system comprising: The event access module receives original event streams containing time stamps from a plurality of heterogeneous data access terminals, wherein the original event streams carry data source identifications and event operation types; the content deconstructing module carries out content deconstructing on the original event stream and separates a semantic description field, a structured data main body and an associated identifier set; The entity matching module is used for matching or creating entity nodes and relationship edges in a dynamic entity relationship network based on the semantic description field and the association identifier set, wherein the dynamic entity relationship network is a core data topology maintained during system operation; the data anchoring module integrates the structured data main body and the timestamp into a versioned attribute block and anchors the versioned attribute block to corresponding entity nodes; The synchronous decision module generates a synchronous decision instruction according to the data source identification and the event operation type, and the synchronous decision instruction is used for driving the state migration of the entity nodes and the relationship edges in the dynamic entity relationship network; and the network execution module is used for executing the synchronous decision instruction, completing the topology reconstruction and attribute update of the dynamic entity relation network, and outputting a network state snapshot to the proxy decision unit. Preferably, the specific implementation of content deconstructing the original event stream includes: invoking a predefined event mode rule set to filter an original event stream, and screening out candidate events related to a preset entity model; Performing nested structure expansion on the original load of the candidate event, and recursively extracting key value pairs of all leaf levels; Grouping key value pairs according to the name space to form a structured data main body set, wherein keys of the structured data main body comprise field, entity type and attribute name information; Resolving a semantic description field from the metadata header of the candidate event, wherein the semantic description field comprises event abstract text, a causal chain identifier and an operation intention code; And extracting a set of associated identifiers from the reference field of the candidate event, wherein the set of associated identifiers at least comprises a main entity i