What this category of data looks like
Data for oil and gas exploration and development ledgers comes from sources including single well exploration operation records, well logging curve reports, oil test analysis documents, monthly development performance reports, and more. Update frequency follows operation nodes: ledgers for individual wells are updated immediately after well completion, while monthly summary ledgers are updated according to the natural month. Document structures are primarily made up of structured fields, including core indicators such as well ID, well depth, permeability, and porosity, alongside unstructured scanned reports and original well logging data files. Field units include professional exploration units such as meters, millidarcys, cubic centimeters per second, and megapascals.
What constraints do these characteristics impose on the knowledge base retrieval and recall link
The large number of structured fields and specialized units require retrieval to support precise field matching and unit verification, to avoid retrieving invalid data with mismatched units. Unstructured documents have long length, so chunk length must be controlled appropriately to retain contextual semantics. Unfixed update frequency requires incremental knowledge base synchronization to avoid full-volume repeated imports. Coexistence of multi-source heterogeneous data requires clear retrieval scope to prevent accidental recall of cross-category documents.
How to set the configurations
| Config Item | Suggested Value | Basis for This Value |
|---|---|---|
similarity | 0.72–0.88 | Structured fields of oil and gas exploration ledgers have high semantic similarity requirements. This range filters low-correlation results while retaining valid data for precise matching |
limit | Top 8–12 entries | A single ledger report has relatively large content. A small number of retrieved results meets the contextual needs of ledger summary and report generation, while avoiding exceeding the context window limit |
usingReRank | Enabled | Semantic similarity differentiation of unstructured well logging reports is low. Reranking optimizes result sorting logic and improves the recall priority of core documents |
searchMode | hybrid | Balances precise matching of structured fields and semantic retrieval of unstructured documents, adapting to the characteristics of multi-source heterogeneous ledger data |
datasetSearchUsingExtensionQuery | Enabled | Exploration and development terms are mostly professional abbreviations. Extended queries supplement synonyms and related terms to improve the coverage of retrieval and recall |
collectionIds | Fill in the specified oil and gas exploration and development ledger collection ID | Clarifies the retrieval scope to prevent documents of other categories from being accidentally included in retrieval results |
Three common mistakes
- Retrieval results include non-oil and gas exploration and development ledger documents, and the number of results exceeds expectations. The cause is that
collectionIdsis not configured, causing the retrieval scope to cover the entire knowledge base. - Exploration reports imported in the online environment have garbled characters, while imports in local deployments are normal. The cause is that the file encoding of the online server is not set to UTF-8, and non-standard characters such as Greek letters and professional symbols in well logging reports are not handled.
- Unable to retrieve ledger data for a specified well ID. The cause is that the
similaritythreshold is set too high, making semantic matching of structured fields fail, or the precise retrieval mode for structured fields is not enabled.
How to confirm the configuration is complete
- Log in to the FastGPT v4.17.0 knowledge base management interface, view the associated collection list, and confirm that the oil and gas exploration and development ledger collection corresponding to
collectionIdsis correctly selected. - Initiate a test query using a single well ID, and check that all returned document results belong to the oil and gas exploration and development category, with no cross-category documents mixed in.
- Adjust the
similarityparameter value, verify that the relevance of retrieval results changes as expected with the parameter, with no obvious accidental recall or missed recall. - After enabling
usingReRank, view the sorted results after reranking, and confirm that the priority of core relevant documents is improved.
The values provided are common starting points and should be measured against the reader's own samples.
Question material comes from public community discussions. Configuration values are common starting points and should be measured against your own samples. Verified on 2026-09-28.