Search

CN-121996766-A - Structured data entry method and system oriented to business rule constraint

CN121996766ACN 121996766 ACN121996766 ACN 121996766ACN-121996766-A

Abstract

The invention discloses a structured data entry method and a structured data entry system for business rule constraint. The method comprises the steps of firstly receiving multi-mode semantic input, retrieving a service object identifier, loading a historical state matrix containing field-level state locking identification according to the multi-mode semantic input and retrieving the service object identifier, calling a semantic extraction module to analyze and acquire service intention and target field candidate values, and allowing an execution value to cover and reset a time stamp only when an error correction semantic signal containing error correction features is detected under the condition that a target field is locked through a multi-source data conflict resolution module, otherwise maintaining an original value. Meanwhile, the system judges the integrity based on a preset business constraint knowledge base and generates structured business data or an inquiry instruction. The whole process records a decision attribution evidence chain comprising rule tracks and fault feature identifications. The invention realizes the controlled management of the field updating behavior in the multi-round structured data input process through the field-level state locking mechanism and the automatic-attribution evidence-storing mechanism, and supports the traceability analysis of the data input process and the abnormal state, thereby being applicable to the business scene with higher requirements on consistency and stability.

Inventors

  • ZHOU YUFEI

Assignees

  • 北京铸睿科技有限公司

Dates

Publication Date
20260508
Application Date
20260128

Claims (14)

  1. 1. A structured data entry method, comprising: in the multi-round interaction process, maintaining session state data corresponding to a service object, wherein the session state data comprises a plurality of service fields and field-level locking identifiers corresponding to the service fields, and the field-level locking identifiers are used for distinguishing first-time confirmation writing behaviors of the fields from updating behaviors based on error correction intention in the multi-round interaction process; When the target service field is detected to be written for the first time, storing a corresponding field value into the session state data, and activating a field-level locking identifier corresponding to the target field to limit the subsequent updating of the field; When receiving new semantic input and analyzing to obtain a target field candidate value, if a field-level locking identifier corresponding to the target field is in an activated state, performing error correction intention judgment on the semantic input; And only when the judgment result shows that the semantic input contains the error correction intention aiming at the target field, releasing the field-level locking identification corresponding to the target field, and allowing the field value coverage operation to be executed on the field.
  2. 2. The method of claim 1, wherein the field level state management comprises: loading corresponding historical session draft state data based on the business object identifier, wherein the session state data at least comprises the session draft state data; When the resolved target field does not exist in the current session draft state data, performing field creation writing operation, and activating a field-level locking identification of the field while finishing field writing.
  3. 3. The method of claim 1, wherein the determination of the error correction intent comprises: invoking an error correction semantic recognition unit to analyze semantic features of semantic input; judging whether error correction semantic features exist in the semantic input or not based on a preset error correction semantic feature template or semantic feature matching rule; when the judging result meets the error correction judging condition, generating an error correction intention identifier corresponding to the target field; the error correction semantic features at least comprise one of error correction keyword features, semantic negative structural features or correction instruction features.
  4. 4. The method as recited in claim 1, further comprising: when field value coverage operation is allowed to be executed, synchronously resetting a field-level locking timestamp corresponding to the target field, so that subsequent conflict judgment for the field is performed based on the last valid modification time of the field, and interference of historical semantic input on a newly confirmed field is avoided.
  5. 5. The method as recited in claim 1, further comprising: triggering derivative calculation logic for complementing the associated field related to the current business intention when detecting that the field state has a change event based on a preset business constraint knowledge base; and performing an integrity decision on the current structured data to determine whether a preset field completion condition is satisfied; When it is determined that the integrity condition is not satisfied, a structured query instruction corresponding to the missing field is generated.
  6. 6. The method of claim 1 or 5, further comprising the step of decision auditing and anomaly attribution: Writing the triggered rule mark, field state change and field processing result into an audit and verification unit in the structured data entry process to form a decision attribution record containing a rule track, multi-mode semantic input text, field state change and field processing result; And monitoring the response state of the external semantic extraction interface, and when an abnormality is detected, attributing and marking the abnormality type based on a preset abnormality characteristic, wherein the abnormality type at least comprises communication timeout abnormality and internal processing abnormality.
  7. 7. A structured data entry system, comprising: The session state management unit is used for maintaining session state data in the multi-round interaction process, wherein the session state data comprises a plurality of service fields and field-level locking identifiers corresponding to the service fields; The multi-round service logic processing engine is used for writing a corresponding field value into the session state management unit when detecting that a target service field is written for the first time, and activating a field level locking identifier corresponding to the target field; the multi-source data conflict resolution module is used for judging the error correction intention of the received semantic input when the field-level locking identifier corresponding to the target field is in an activated state; the multi-source data conflict resolution module is further used for unlocking the locking identification of the corresponding field and allowing the field value coverage operation to be executed only when the judgment result shows that the semantic input contains the error correction intention aiming at the target field.
  8. 8. The system of claim 7, wherein the multi-source data conflict resolution module performs a field creation write operation and synchronously activates a field level state lock identification of a corresponding field when the target field does not exist in a current session draft state.
  9. 9. The system of claim 7, wherein the multi-source data conflict resolution module is configured to synchronously reset a field level lock timestamp corresponding to the target field when field value overwriting is allowed.
  10. 10. The system of claim 7, wherein the multi-source data conflict resolution module includes an error correction semantic recognition unit configured to determine whether error correction features exist in the user input semantics based on a preset error correction semantic feature template or semantic feature matching rules.
  11. 11. The system of claim 7, wherein the multi-round business logic processing engine is configured to trigger derivative computation logic based on a pre-set business constraint knowledge base to complement associated fields in the structured data upon detection of a field state change event.
  12. 12. The system of claim 7 or 10, further comprising an anomaly monitoring and auto-attribution module for monitoring a response status of the external semantic extraction interface and attributing an anomaly type based on a preset anomaly characteristic when an anomaly is detected.
  13. 13. The method of claim 6, further comprising a data playback step of driving reconstruction of state data in time sequence based on the full snapshot data and corresponding rule tracks in the decision attribution record to reconstruct an input decision logic evolution path of a particular business object during multiple rounds of interactions.
  14. 14. The system of claim 12, further comprising a data playback module configured to drive reconstruction of state data in a time-series order based on the full snapshot data and corresponding rule tracks in the decision attribution record to reconstruct an entry decision logic evolution path of a particular business object during multiple rounds of interactions.

Description

Structured data entry method and system oriented to business rule constraint Technical Field The invention relates to the technical field of computer data processing and natural language processing, in particular to a business rule constraint-oriented multi-round interactive structured data input method and system. Background With the rapid development of Artificial Intelligence (AI) and Natural Language Processing (NLP) technologies, an automated input scheme for converting unstructured multimodal semantic input (such as voice, text, etc.) into structured business data has been widely used in the fields of finance, industry, medical treatment, etc. However, in practical application scenarios (especially complex multi-round interaction scenarios), existing structured data entry techniques still have the following technical drawbacks: 1. The problem of information miscoverage in multi-round interaction is that uncertainty exists in the analysis of the context intention by the AI model in the multi-round semantic extraction process. When a user enters new semantic information in a subsequent interaction, it is often difficult for existing systems to accurately identify whether the information is "a complement to a new field" or "a misinterpretation of an entered field. Due to the lack of a state management mechanism based on field-level locking identification, the newly generated extraction result is very easy to carry out error coverage on the correct data confirmed by the user before, so that the consistency of the structured data is difficult to guarantee. 2. Error correction logic lacks an explicit technical decision mechanism-existing systems typically employ a simple "time-first" or "probability score-first" strategy when processing data updates. This approach does not effectively distinguish between "normal traffic updates" and "error correction behavior for errors". Under the condition of lack of linkage of error correction semantic recognition and field-level locking identification, the system cannot ensure that the core field is allowed to be updated only when an explicit error correction signal is received, so that the structural input accuracy is reduced. Traditional database concurrency control techniques (e.g., optimistic lock or multi-version concurrency control MVCC) are capable of resolving data write conflicts, but their core logic is based on 'chronological' or 'transaction isolation', and cannot understand business semantics. In a human-computer interaction scenario, the user's ' corrective ' action does not always occur at the ' last moment '. Therefore, the existing database lock mechanism can not solve the problem of intent conflict in AI semantic understanding, and a special technical scheme integrating semantic understanding (error correction recognition) and state control (locking mechanism) is needed. 3. The decision process is opaque and the fault attribution is difficult, the existing logging system is mostly 'black box' processing flow, and the link records for AI extraction process, rule triggering logic and field state transition are lacked. When deviation occurs in the structured result or an external semantic interface (such as an LLM interface) responds to abnormality, development and operation staff are difficult to trace back decision basis, and also cannot accurately ascribe different types of faults such as communication overtime and logic matching abnormality, and cannot meet the requirements of auditing and stability under a high compliance scene. Therefore, a technical scheme capable of guaranteeing the input accuracy and stability through field-level state control, accurate error correction intention recognition and full-link decision evidence storage is needed. Particularly, under the background that artificial intelligence and entity economy are greatly promoted to be deeply fused and autonomous and controllable digital economy bases are built in China, a structured input system which can effectively solve the uncertainty of AI semantic extraction and has complete decision attribution capability is developed aiming at the fields of finance, government affairs, high-end manufacturing and the like which have strict requirements on data full-link safety and autonomous intellectual property rights, and has important practical significance for realizing digital transformation in the key business field. Disclosure of Invention First technical problem The invention aims to solve the technical problem that in the existing unstructured data entry process, semantic extraction results based on Artificial Intelligence (AI) often have uncertainty. Particularly in a multi-round interaction scenario, the newly generated extraction results are prone to erroneously overlaying the confirmed correct data due to the lack of an effective state management and conflict resolution mechanism. In addition, the existing recording system lacks transparent record of decision process and effect