What the documents look like
Trade and FX administration for procurement and supplier records centers on four document types: foreign exchange receipt advices, trade contracts, invoices, and foreign exchange business registration materials. Unlike letter-of-credit document examination, this category tracks whether foreign exchange funds and the underlying trade background are genuine. A full set of letter-of-credit documents is not examined here. Materials update per transaction and per month, and for any single supplier the receiving entity, contract counterparty, invoice amount, and registration records need line-by-line reconciliation.
Inside a knowledge base, the search unit is the library; collections serve only as classification and do not affect search results. One group of source data can be linked to multiple indexes, and the index content is checked separately from the original text. Knowledge bases linked together for retrieval must use the same vector model, so the actual model and input type are confirmed first. A vector score reflects similarity only; retrieval does not perform authorization checks.
Constraints at the vector indexes step
Supplier records commonly carry entity names, contract numbers, invoice numbers, and registration status fields. Semantic retrieval suits natural-language questions, while full-text retrieval suits numbers and proper nouns. Hybrid retrieval merges multiple result paths through RRF. Reranking requires an available model and must be enabled in settings. The reference cap controls total referenced tokens. Minimum relevance takes effect only in semantic retrieval or when result reranking is used, and it is a filter threshold rather than a final sort rule.
Image retrieval also depends on whether image descriptions or image vectors were generated at ingestion time; checking the query side alone is not sufficient. Libraries built on different vector models cannot be linked together. The identifier is checked first, non-matching libraries are excluded, and the actual observation is recorded. A retrieval result can have length zero, and the downstream path still completes.
Configuration and implementation checks
Hypothetical paper exercise sample: a trade finance staff member organizes supplier A's receipt materials. The source text reads: "Supplier A received payment of USD 50,000 from overseas buyer B on 10 March 2026, corresponding to contract HT-2026-031, invoice INV-0310, receipt advice number SH-20260310." The draft sentence to be judged reads: "Supplier A has completed settlement and the funds have been credited." The source text records only the receipt of payment. It does not record completed settlement or credited status.
| Configuration item | Purpose on this page | How to check |
|---|---|---|
| Linked knowledge bases | Select supplier record libraries using the same vector model | Compare model identifiers and input types; exclude non-matching ones |
| Search mode | Use full-text or hybrid retrieval for number-type fields | Use contract or invoice numbers as query terms and compare hits against source text |
| Index content | Build focused indexes for the same source data | Compare index text against the entity, object, and status in the source |
| Reference cap | Control total referenced tokens | Set a hypothetical starting value for this exercise and compare against other values |
| Minimum relevance | Filter low-score results in semantic retrieval or reranking | Record citation differences before and after a threshold change |
| Result reranking | Secondary sorting when candidates are numerous | Confirm the model is available and enabled in settings |
Three common errors
- Reading "received payment" as "completed settlement." The source text records only the fact of receipt. Completed settlement and credited funds are separate statuses. When evidence is missing, the status remains pending verification.
- Treating a vector similarity score as a permission or entity filter. Similarity does not exclude unrelated suppliers or unauthorized records. Retrieval results still require manual checking of the entity and the authorization boundary.
- Linking knowledge bases across different vector models. Libraries on different vector models cannot be linked together. The identifier is checked first, non-matching libraries are excluded, and the actual observation is recorded.
How to verify the setup
- Enter the hypothetical source text and the draft sentence. Compare the receipt fact recorded in the source against the completed-settlement wording in the draft. The source does not record settlement or crediting, so the judgment is unsupported rather than contradictory or negated.
- Enter contract number HT-2026-031 as the query term. Compare whether the returned citation contains the source text with that number. A hit whose cited source matches the sample satisfies the condition. A miss is recorded as a gap without inferring a cause.
- Enter the paper input [] for an empty result. Check whether the downstream path completes. An output of length zero with the path continuing satisfies the condition. No claim is made that a fabricated rule necessarily returns an empty array.
The publicly documented technical facts used to identify relevant content in this document are: linked retrieval knowledge bases must use the same vector model, and the actual model and input type are checked first. One source data block can be linked to multiple indexes, and index content is checked separately from the source text. Indexes are built from complete and focused information, and retrieval coverage and index length are weighed against the sample. Image retrieval also depends on whether image descriptions or image vectors were generated at ingestion time, not on the query side alone. After a model or index change, the same set of questions is used to compare cited source text, and no effect is promised from the vector score alone.
This page addresses only trade and FX administration / procurement and supplier records / vector models and indexes. During operational handover, the authorization status of materials, operations, and requests must be distinguished. Retrieval and forms do not perform authorization.