What this category's data looks like
In the scenario of district and county teaching research activities and lesson observation and evaluation, the materials that enter the business basis query are mainly three types: teaching research activity notices, lesson observation records, and lesson evaluation minutes. A teaching research activity notice usually contains the activity name, time, location, participating kindergartens, activity agenda, and signature receipt column; a lesson observation record contains the observation date, class, teaching teacher, activity segments, observation points, and signature; lesson evaluation minutes contain the evaluation date, participants, key points of speeches, conclusions, and receipt. The three types of materials are kept separately by different kindergartens or by different segments of the same kindergarten, and the update rhythm is mostly supplementary entry one by one before and after the activity. The names, quantities, and dates of the same record in different materials often fail to match, and unsynchronized versions, missing signatures, or missing receipts are also common. When materials enter this function, the notice bears the activity background and scope definition, the lesson observation record bears on-site factual support, and the lesson evaluation minutes bear the conclusions and improvement basis. Together, the three constitute the source of basis entries that can be traced back to the original text, to be called upon when replying to parents or writing them into plans.
Configuration and implementation checks
This step is executed by the index service and vector database, outputting vectors and index status (whether recall is possible, whether data in the same group is merged). The basis entries that can be traced back to the original text are then completed by subsequent AI nodes, code nodes, or manual work (retrieval and citation trimming only take supporting materials). The index service and vector database are only responsible for writing the vectors of the segmented materials into the vector database, and during retrieval merging them back to the original data by identifier.
| Object | Check | Verification action |
|---|---|---|
| Index model | Consistent with the vector model actually used by the knowledge base; only data that has already been indexed triggers rebuilding after changing the model | Check the index model in the knowledge base settings, and view the data scope covered by the rebuild task |
| Index field of the data row | Multiple vector identifiers of one data row are all present, and data in the same group can be merged during retrieval | Open a data row and check the index field, confirming the status is indexed and the rebuild task record |
| Teaching research activity notice | The activity name, date, participating kindergartens, and signature receipt column can match the lesson observation record and lesson evaluation minutes | Check the notice date and signature, and check whether the name and quantity of the same record are consistent with the other two materials |
| Lesson observation record and lesson evaluation minutes | The observation date, class, teaching teacher, evaluation conclusion, and receipt can refer back to each other | Check the signature and receipt, and check whether the versions are synchronized and whether the entries can match the original text |
Three common errors
- Phenomenon: duplicate entries appear in retrieval results. Inspect the following: the same group of data corresponds to multiple vector identifiers, and the highest score is not taken when merging back to the original data.
- Phenomenon: the name, quantity, and date of the same record in two materials do not match. Inspect the following: the notice, lesson observation record, and lesson evaluation minutes are supplemented separately, without version synchronization.
- Phenomenon: a data row has content but cannot be retrieved. Inspect the following: the index field lacks a vector identifier, or missing signatures or receipts cause the material not to enter the index.
How to verify the setup
This function's acceptance: take one data row and look at the index field, confirm that multiple vector identifiers are all present 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; after changing the model, only trigger rebuilding for data that has already been indexed, and view the coverage scope of the rebuild task record.
Business test: ask a basis question that spans materials, see whether the answer gives a traceable source, and check the number of citations and the tokens of the cited text; the number of array entries is recorded separately, and limit are only used as reference for the citation content budget.
- Open a data row and check the index field, confirm the status is indexed and the rebuild task record, then retrieve once to see whether data in the same group is merged.
- Check the dates, signatures, and receipts of the teaching research activity notice, lesson observation record, and lesson evaluation minutes, and check whether the name and quantity of the same record are consistent and whether the versions are synchronized.