What this category's data looks like
Research achievement registration and intellectual property management involves three types of materials: achievement registration forms, patent disclosure documents, and software copyright materials. Achievement registration forms record the achievement name, contributors, dates, and signatures; patent disclosure documents carry the technical solution, terminology definitions, and inventor confirmation; software copyright materials include version numbers, functional descriptions, and receipts. The three types of materials often circulate in multilingual versions, and the same term needs to be able to refer back to the source text across different languages. When the materials enter the knowledge base retrieval and recall function, achievement registration forms provide cross-reference clues for name and date fields, patent disclosure documents provide terminology definitions and context, and software copyright materials provide version and receipt information. The update cadence advances with registration, application, and publication milestones, and the field characteristics manifest as high terminology density, many cross-language correspondences, and date and signature fields that need to be checked item by item.
Configuration and implementation checks
This step is executed by the knowledge base retrieval service (retrieval node and search test), outputting candidate chunks and similarity scores (after pattern, merge, re-rank, filter, and upper-limit trimming). The "unified terminology glossary and comparison draft" is completed by subsequent nodes or manually.
| Object | Check | Verification action |
|---|---|---|
| Achievement registration form | Name, quantity, date, and signature fields need to be aligned item by item with the patent disclosure document and software copyright materials | In the search test, switch searchMode, query once each using different language phrasings of the same term, record the number of candidates and the scores, then check the registration form fields |
| Patent disclosure document | Terminology definitions and translated names need to be able to refer back to the source text, and the same term must be written consistently throughout | Compare the ranking of the top few results before and after enabling usingReRank, and check whether the candidate chunks cover the terminology definitions and context |
| Software copyright materials | Version numbers, functional descriptions, and receipts need to stay consistent with the registration form and disclosure document; missing signatures or receipts need to be flagged | Adjust the similarity threshold to observe changes in candidates, separately record the number of entries in the citation array, and check the tokens budget of the cited text |
| Unified terminology glossary and comparison draft | The same term must be written consistently throughout, translated names must be able to refer back to the source text, and spot checks must be completed before external release | Spot-check three terms to see whether they are written consistently throughout, and check whether dates, quantities, and signature fields are in sync |
Three common errors
- Symptom: The name, quantity, or date of the same achievement does not match across two materials. Cause: Registration and disclosure are maintained separately, versions are not synchronized, and retrieval candidates only present chunks without field-level comparison.
- Symptom: The translated name of a term is written differently within the same material. Cause: Multilingual materials are entered in separate batches, the terminology glossary is not unified, and recall candidates do not cover all variants.
- Symptom: Software copyright materials are missing signatures or receipts. Cause: Process nodes do not send back data, and retrieval only returns candidate chunks and does not perform authorization or receipt verification.
How to verify the setup
Acceptance of this feature: test once each with synonym-rewritten questions, compare candidates and scores, then toggle re-ranking on and off to inspect the ranking of the top few results.
Business test: spot-check three terms to see whether they are written consistently throughout.
- In the search test, switch
searchModeand query the same term using semantic, full-text, and hybrid modes respectively, recording the number of candidates and the scores. - Compare the ranking of the top few results before and after enabling
usingReRank, adjust thesimilaritythreshold to observe changes in candidates, separately record the number of entries in the citation array, and check the tokens budget of the cited text.