What this category's data looks like
This page focuses on data versioning and change maintenance in the context of research funding budget adjustments and reimbursements, involving three types of materials: budget adjustment applications, reimbursement forms, and contract invoices. Budget adjustment applications record the reason for adjustment, amount, and approval signature; reimbursement forms carry expense details, quantities, and dates; contract invoices provide transaction evidence and invoicing information. The material sources are project team submissions, financial archives, and contract attachments, and the update cadence changes as approval milestones progress. Field characteristics include application number, amounts before and after adjustment, reimbursement reason, invoice number, signature field, and receipt date. When the materials enter the vector model and indexing feature, each is stored as an independent data row, the index field records the corresponding vector identifier, and during retrieval the original data is merged back by identifier, with the highest score taken for the same group of data. The version ledger and change records are completed by subsequent AI nodes, code nodes, or manually, and retrieval and citation trimming only take supporting materials.
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). "Version ledger and change records" are then completed by subsequent AI nodes, code nodes, or manually (retrieval and citation trimming only take supporting materials).
| Object | Check | Verification action |
|---|---|---|
| Budget adjustment application | The application number, amounts before and after adjustment, approval signature, and date must correspond item by item with the reimbursement form and contract invoice, and the name, quantity, and date of the same record in the two materials must be consistent | Open the data row to check whether the index field contains multiple vector identifiers, check whether the values of defaultToken and maxToken are consistent with the knowledge base indexing model, and confirm the status is indexed |
| Reimbursement form | The reimbursement reason, expense quantity, and occurrence date must match the invoicing information and receipt signature of the contract invoice; missing signature or receipt is regarded as inconsistent | Retrieve the records related to this reimbursement form, check whether data in the same group is merged and whether the highest score is taken, and check the coverage of the rebuild task records |
| Contract invoice | The contract name, invoice number, amount, and date must be consistent with the budget adjustment application and reimbursement form; when the version is not synchronized, the old version is still traceable | After changing the model, check whether the rebuild task records are triggered only for data that has completed indexing, and sample one data row to check the completeness of the index field |
| Version ledger and change records | The current version is unique, the old version is traceable, and citations are synchronized; after changing one piece of content, the version number and citations must be updated synchronously | Record the tokens and array count of the retrieved citation text separately, and check whether the limit budgets match |
Three common errors
- Symptom: duplicate entries appear in retrieval results. Inspect the following: the name, quantity, and date of the same record in the two materials do not match, and the vector identifiers are not merged.
- Symptom: recall quality drops after changing the model. Inspect the following: the version is not synchronized, and the rebuild task records do not cover all data that has completed indexing.
- Symptom: a data row has content but cannot be retrieved. Inspect the following: missing signature or receipt makes the material incomplete, and the index field does not record the corresponding vector identifier.
How to verify the setup
This feature's acceptance: sample one data row to inspect the index field, confirm that multiple vector identifiers are all present and the status is indexed, then retrieve once to see whether the same group is merged and whether the highest score is taken; check the values of defaultToken and maxToken.
Business test: change one piece of content, see whether the version number and citations are updated synchronously, and confirm that the current version is unique and the old version is traceable.
- In the knowledge base settings, check whether the indexing model is consistent with the vector model actually used by the knowledge base, and view the data scope covered by the rebuild task records.
- Open a data row to check the index field, confirm the status is indexed, and check that the tokens and array count of the citation text are recorded separately.