What this category's data looks like
In the scenario of signing research project task books, the materials include three types: project task books, topic task breakdown tables, and budget books. The project task book provides the topic name, undertaking unit, start and end dates, and signature fields; the topic task breakdown table breaks out sub-topics, responsible persons, and delivery milestones; the budget book lists budget subjects, amounts, years, and approvals. The three types of materials often come from different departments with differing update rhythms: the task book is finalized according to the signing milestone, the breakdown table is adjusted after the topic starts, and the budget book rolls over annually. When entering the multilingual materials and terminology processing function, the task book and breakdown table serve as terminology sources and reference drafts, while the budget book provides the basis for cross-verification of quantity and date fields. Field characteristics are concentrated in topic number, topic name, responsible person, start and end dates, budget subjects, signatures, and receipts. The name, quantity, and date of the same record in the three materials need to be able to refer back to each other, and the version and signature status need to be synchronized.
Configuration and implementation checks
This part is executed by the indexing service and vector database, outputting vectors and index status (whether recall is possible, whether data in the same group is merged). "Unified terminology table and reference draft" is then completed by subsequent 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 the data table of 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, and the highest score is taken for the same group of data.
| Object | Check | Verification action |
|---|---|---|
| Project task book | Check whether the topic name, responsible person, start and end dates, signature, and receipt are consistent with the breakdown table and budget book | Open the task book data row, check whether the index field records multiple vector identifiers, and confirm that the index model is consistent with the vector model actually used by the knowledge base, with status indexed |
| Topic task breakdown table | Check whether the sub-topic name, quantity, and delivery date match the task book and budget book | Check the index model in the knowledge base settings, view the data range covered by the rebuild task record, and confirm that after changing the model, rebuild is triggered only for data that has already been indexed |
| Budget book | Check whether the budget subjects, amounts, years, and approval dates are consistent with the task book and breakdown table | Perform one retrieval, see whether the same group of data is merged and whether the vector score takes the highest score, and confirm that the data row status is indexed |
| Terminology table and reference draft | Check whether the same term is written consistently throughout and whether the translated name can refer back to the original text | Sample three terms to check their writing throughout, record the number of citations and the tokens budget of the cited text, and do not compare the two for equality |
Three common errors
- Symptom: duplicate entries appear in retrieval results. Inspect the following: the same group of data corresponds to multiple vector identifiers, the merge step does not merge by identifier, or the data row index field is missing some vector identifiers.
- Symptom: recall quality drops after changing the model. Inspect the following: the rebuild task does not cover all data that has already been indexed, or the index model is inconsistent with the vector model actually used by the knowledge base.
- Symptom: the name, quantity, and date of the same record do not match across two materials. Inspect the following: versions are not synchronized, signatures or receipts are missing, and the terminology table and reference draft are not updated along with the materials.
How to verify the setup
Acceptance of this function: sample one data row and look at the index field to confirm that multiple vector identifiers are all present and the status is indexed; then perform one retrieval to see whether the same group of data is merged and whether the vector score takes the highest score, and view the data range covered by the rebuild task record, checking the values of defaultToken and maxToken.
Business test: sample three terms to check whether their writing is consistent throughout, confirm that the translated names can refer back to the original text, record the number of citations and the tokens budget of the cited text separately, and do not compare limit for equality.
- In the knowledge base settings, check whether the index model is consistent with the vector model actually used by the knowledge base, and view the data range covered by the rebuild task record.
- Open a data row and check the index field to confirm that multiple vector identifiers are all present and the status is indexed, then perform one retrieval to see whether the same group of data is merged.