Knowledge Base Retrieval for Research Outcome Registration and Intellectual Property: Maintain Document Versions and Changes

Research IP: review retrieval for version tracking. Compare the source records, required inputs and downstream results before handing the task over.

What this category's data looks like

This category involves three core types of materials: achievement registration forms, patent disclosure documents, and software copyright materials. Achievement registration forms contain the achievement name, contributors, registration date, achievement number, and organizational seal; patent disclosure documents contain the technical solution, inventors, filing date, drawings, and inventor signature; software copyright materials contain the software name, version number, development completion date, copyright holder information, and seal receipt. The three types of materials come from scattered sources, and their update cadence changes with nodes such as application, examination, correction, and registration approval, with fields most prone to discrepancies in names, quantities, dates, and signatures. When entering the knowledge base retrieval and recall function, achievement registration forms serve as the source of the achievement master record, patent disclosure documents serve as the source of technical associations, and software copyright materials serve as the source of versions and ownership, each providing candidate snippets for subsequent nodes or manual comparison and ledger generation.

Configuration and implementation checks

This step is executed by the knowledge base retrieval service (retrieval node and search test), outputting candidate snippets and similarity scores (after pattern, merge, rerank, filter, and upper-limit trimming). The version ledger and change records are then completed by subsequent nodes or manually.

ObjectCheckVerification action
Achievement registration formWhether the name, quantity, registration date, and seal are consistent with the patent disclosure document and software copyright materialsSwitch searchMode in the search test to obtain candidates, and check the same achievement name and date fields
Patent disclosure documentWhether the inventors, filing date, and technical solution name match the corresponding record in the achievement registration formEnable usingReRank to compare the ordering of the top few results, and check whether the fields of the disclosure document and the registration form are synchronized
Software copyright materialsWhether the software name, version number, development completion date, and seal receipt are consistent with the registration form versionAdjust similarity to filter out low-scoring candidates, and check the version number and receipt fields one by one
Version ledger and change recordsThe current version is unique, old versions are traceable, and references are synchronizedAfter changing one piece of content, rerun retrieval to see whether the candidate snippets and references update with the version

Three common errors

  • Symptom: The name and quantity of the same achievement do not match across two materials. Cause: The registration form and the disclosure document are maintained separately, without field alignment against the same record.
  • Symptom: Old-version materials are still referenced, and the current version is not synchronized. Cause: Retrieval was not rerun after the change, so candidates still hit historical snippets.
  • Symptom: The disclosure document or copyright materials lack signatures or receipts. Cause: Signature and receipt fields were not included in the retrieval verification scope and were omitted from the candidates.

How to verify the setup

Acceptance for this function: test once each with synonym-rewritten questions, compare the number of candidates and scores, then toggle rerank on and off to observe the ordering of the top few results.

Business test: change one piece of content and see whether the version number and references are updated synchronously.

  • In the search test, run each of the three searchMode options—semantic, full-text, and hybrid—once, record the number of candidates and scores, and check whether both semantic and full-text results are included.
  • After changing one field of the same record, rerun retrieval, check the tokens and array length of the referenced text, and confirm that old-version snippets no longer appear in the top few results.