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 register the name, completers, date, and signature; patent disclosure documents contain the technical solution, inventor, application date, and seal; software copyright materials contain the software name, version number, development completion date, and receipt. The three types of materials come from different sources and have different update rhythms: registration forms are produced along with achievements, disclosure documents are submitted along with applications, and copyright materials are added along with registration receipts. When materials enter this function, the achievement registration form serves as the main record providing the name and quantity fields, while patent disclosure documents and software copyright materials serve as comparison sources providing date, signature, and receipt information; together, the three constitute basis entries that can be traced back to the original text.
Configuration and implementation checks
The indexing service and vector database only output vectors and index status (whether recall is possible, whether data in the same group is merged); basis entries that can be traced back to the original text must be handled subsequently by AI nodes, code nodes, or manual work (retrieval and citation trimming only take supporting materials). Vectors are stored in the vector database, while the original data is stored in data tables in the business database, with index fields recording the corresponding multiple vector identifiers; retrieval first recalls vectors, then merges back to the original data by identifier, taking the highest score for the same group of data.
| Object | Check | Verification action |
|---|---|---|
| Index model | Consistent with the vector model actually used by the knowledge base | Check the index model in the knowledge base settings, confirm it is the same as the currently used vector model, and check whether the data row status is indexed |
| Index field of the data row | All multiple vector identifiers of one piece of data are present | Open a data row and check the index field, confirm that the multiple vector identifiers are complete, and that the same group of data can be merged by identifier during retrieval |
| Rebuild task record | After changing the model, trigger only for data that has already completed indexing | Check the data scope covered by the rebuild task record, confirm that it only includes data that has already completed indexing, and that data that has not completed indexing does not enter the rebuild |
| Achievement registration form, patent disclosure document, software copyright material | The name, quantity, and date of the same record match, the version is current, and signatures and receipts are complete | Check the fields, dates, and signatures of the three types of materials item by item, compare whether the name, quantity, and date of the same record are consistent, and check whether the versions are synchronized and whether signatures or receipts are missing |
Three common errors
- Symptom: duplicate entries appear in retrieval results. Inspect the following: the same group of data corresponds to multiple vector identifiers, and after recall it is not merged back to the original data by identifier, so the highest score is not taken for the same group of data.
- Symptom: recall quality declines after changing the model. Inspect the following: the rebuild task is triggered only for data that has already completed indexing, data that has not completed indexing is not covered, and versions are not synchronized, causing recalled content to be inconsistent with the current materials.
- Symptom: a data row has content but cannot be retrieved. Inspect the following: the data row's index field is missing or the vector identifiers are incomplete, the vectors are not written into the vector database, and the corresponding original data cannot be recalled during retrieval.
How to verify the setup
Acceptance of this function: sample a data row and look at the index field, confirm that the multiple vector identifiers are complete and the status is indexed; then retrieve once and see whether the same group of data is merged and whether the highest score is taken.
Business test: ask a basis question that spans materials, see whether the answer gives a traceable source, and check whether the source matches the original text and whether the version is current.
- Open a data row, check whether the multiple vector identifiers in the index field are complete, confirm that the status is indexed, and check the data scope covered by the rebuild task record.
- Check in the knowledge base settings that the index model is consistent with the currently used vector model, check that the tokens and array count of the cited text are recorded separately, and check
limitagainst the citation content budget.