What this category's data looks like
This page addresses the scenario of academic misconduct report investigation and processing. Material types include report materials, investigation reports, and expert appraisal opinions. Report materials are usually submitted by the reporter and contain original leads, evidence lists, and timelines; investigation reports are written by the investigation team and record the verification process, factual findings, and handling recommendations; expert appraisal opinions are issued by domain experts and provide professional judgments on specific academic content. All three types of materials must be annotated with fields, dates, and signatures, and often involve multilingual materials and terminology comparisons. After materials enter this feature, they are first converted into searchable text and vectors for subsequent terminology unification and comparison draft generation. The update pace advances with the investigation stages, and versions may iterate multiple times. Attention must be paid to whether the names, quantities, and dates of the same record are consistent across different materials, as well as version synchronization and the completeness of signature receipts.
Configuration and implementation checks
This step 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 glossary and comparison 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 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 verify the values of defaultToken and maxToken |
| Index field of the data row | All multiple vector identifiers of one piece of data are present, and the status is indexed | Open a data row and check the index field, confirm that multiple vector identifiers are complete and the status is indexed |
| Rebuild task record | After changing the model, only trigger for data that has already been indexed | View the data range covered by the rebuild task record, and confirm that it only involves data that has already been indexed |
| Report materials, investigation reports, and expert appraisal opinions | The name, quantity, and date of the same record are consistent across the three materials, versions are synchronized, and signature receipts are complete | Compare the fields, dates, and signatures of the three materials item by item, verify whether the name, quantity, and date of the same record are consistent, and check whether 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 they are not merged by original data, or the highest score is not taken during merging.
- Symptom: The name, quantity, and date of the same record do not match between the report materials and the investigation report. Inspect the following: Multilingual terminology is not unified, or versions are not synchronized, resulting in citation of an old draft.
- Symptom: Recall quality declines after changing the model. Inspect the following: The rebuild task record 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.
How to verify the setup
Feature acceptance: Sample one data row and check the index field to confirm that multiple vector identifiers are complete and the status is indexed; then retrieve once to see whether data in the same group is merged and whether the highest score is taken, and verify the values of limit in the index and citation budget.
Business test: Sample three terms and check whether their spelling is consistent throughout the entire text, and confirm that translated names can refer back to the original text; the number of citations is the array length, quoteMaxToken is the citation content budget, the two are not compared for equality, verify the tokens of the citation text, and record the array count separately.
- Check the index model in the knowledge base settings, confirm it is consistent with the vector model actually used by the knowledge base; open a data row and check the index field, confirm that multiple vector identifiers are complete and the status is indexed.
- View the data range covered by the rebuild task record, and confirm that after changing the model, it is only triggered for data that has already been indexed; sample three terms and check whether their spelling is consistent throughout the entire text, and verify whether translated names can refer back to the original text.