Reference and Traceability for DTP Pharmacy Registration and

Data Characteristics for This Category

DTP pharmacy registration and declaration documents originate primarily from drug registration approvals, drug inserts, quality standards, clinical trial reports, pharmaceutical research data, non-clinical research data, and the pharmacy's GSP certification, business licenses, and personnel qualifications. Update frequencies vary by document type. Drug approvals and inserts may update with National Medical Products Administration (NMPA) approvals, while pharmacy qualifications undergo periodic review. Document structures are highly standardized, adhering to NMPA's unified formats, such as the CTD format. Fields include generic drug name, dosage form, strength, manufacturer, approval number, indications, dosage and administration, adverse reactions, precautions, and for the pharmacy, business license number, GSP certificate number, legal representative, and number of licensed pharmacists. Data units are strict, such as milligrams (mg) and milliliters (ml) for dosage, batch numbers, and expiry dates.

Constraints Imposed by These Characteristics on "Reference and Traceability"

The highly standardized nature and strict update cycles of DTP pharmacy registration and declaration documents require precise referencing to original texts, with traceability to specific versions and revision dates. Consistent document structure enables chapter or paragraph-level referencing, but requires meticulous text preprocessing. The strictness of fields and units means the system must accurately identify and cite relevant numerical values in responses, avoiding vague statements. For example, citing a drug approval requires both the approval number and its approval date. Pharmacy qualification documents must link to specific certificate numbers and validity periods. Since document updates can involve additions, revisions, or invalidations, the referencing system must support version management to ensure citations always point to the latest or a specific historical valid version. Reference display needs to support linking to original documents or providing clear path descriptions.

Configuration Settings

Configuration Item	Recommended Value	Rationale for this Setting
`maxContext`	`3000 characters`	DTP pharmacy document paragraphs are often long, requiring more context for complete referencing.
`Similarity threshold` (Similarity Threshold)	`0.78`	Registration and declaration documents use highly standardized terminology and descriptions; a higher threshold reduces irrelevant recall.
`Rerank result count` (Reranked Return Count)	`Top 5 entries` (Top 5)	Ensures that among many candidate references, the few most relevant ones are prioritized for processing and display.
`Chunk size` (Segment Length)	`800 characters`	Balances semantic completeness and retrieval efficiency, preventing overly long segments from leading to imprecise retrieval.
`Recall count` (Recall Count)	`15 entries`	Ensures enough potential reference items for the reranking model to choose from, improving final citation accuracy.
`PARSE_FILE_TIMEOUT_SECONDS`	`600 seconds`	Registration and declaration documents can be large, requiring more time for parsing, preventing processing failures due to timeouts.

Three Common Mistakes

Symptom: Knowledge base responses do not display specific source documents or paragraphs for citations. Reason: Configuration items like Reference Source Output or Display Knowledge Base Citations are not enabled, or the association between the cited content and the original text is lost during data ingestion.
Symptom: Cited content returned by the system deviates from information in the original file, such as inconsistent drug dosage or expiry dates. Reason: Knowledge base data lacks effective version management, leading to citations of outdated or revised document versions.
Symptom: Content appears garbled after uploading a .csv file. Reason: File encoding format is inconsistent with the system's default encoding; for example, the original file uses GBK encoding while the system processes it as UTF-8.

How to Confirm Correct Configuration

Select a typical registration and declaration document containing various fields and units. After submitting a query, check if critical information like batch numbers, strengths, and expiry dates in the response's citations match the original text, and verify the cited original paragraphs.
For a document that has undergone version updates, verify that the system accurately cites information from the latest version and can trace its revision history.
Upload a declaration document file containing special characters or multilingual content. Check if the system exhibits garbled text or information loss during parsing and citation, and ensure all characters display correctly.

The values provided are common starting points and should be measured against specific samples.

Question material comes from public community discussions. Configuration values are common starting points and should be measured against your own samples. Verified on 2026-09-21.