What This Category’s Data Looks Like
Water utility marketing-related data is primarily sourced from pipeline network operation and maintenance records, water fee collection ledgers, water conservation promotional materials, offline event plans, official public documents, and other channels. Update frequency varies by business scenario: marketing event plans are updated alongside event cycles, operation and maintenance records are archived monthly, and promotional materials are adjusted based on seasons or policies. Document formats include long-text PDFs, structured table reports, scanned posters, and more. Fields cover water supply area scope, water unit price (yuan/cubic meter), pipeline pressure parameters, event participation rules, and other items. Some documents contain unstructured hand-drawn annotations or image captions.
Constraints on Document Parsing and Chunking
The multi-source and multi-format nature of water utility marketing documents requires the parsing process to adapt to content extraction needs across different formats. Long-text event plans must be split along semantic boundaries to avoid mixing rules across chapters. Structured reports must retain associations between fields and units to prevent disconnection of information such as unit prices and usage quantities after chunking. Scanned promotional materials require OCR to extract text, otherwise retrievable content cannot be generated. Additionally, some documents contain specialized terminology and custom units; context associations must be preserved during chunking to ensure full matching of business keywords during retrieval.
Configuration Settings
| Configuration Item | Recommended Setting | Rationale |
|---|---|---|
PARSE_OCR_ENABLE | Enabled | Water utility marketing documents often include scanned posters, pipeline network diagrams, and other non-pure text content, requiring OCR to extract valid text |
MAX_SEGMENT_LENGTH | 800–1200 characters | Water utility documents contain long sections of event rules and operation instructions. This range preserves semantic integrity and avoids retrieval logic breaks caused by excessive chunking |
PARSE_FILE_TIMEOUT_SECONDS | 120 seconds | Large water utility marketing manuals (including multi-page reports and high-resolution images) take longer to parse. This duration covers parsing needs for most files |
ENABLE_SEGMENT_SPLIT_BY_HEADING | Enabled | Water utility documents often organize marketing activities, service terms, and operation instructions by chapter. Splitting chunks by headings improves recall accuracy |
UPLOAD_FILE_MAX_SIZE | 500 MB | Some water utility marketing materials include high-resolution pipeline maps, requiring support for large file uploads to fully preserve asset information |
RECALL_SIMILARITY_THRESHOLD | 0.75–0.85 | Water utility-related terminology has high recognition specificity. This threshold filters irrelevant recall results and improves retrieval precision |
The parameter values provided on this page are common starting points for configuration. Actual values are affected by material format, data volume, and business rules. Specific issues require case-by-case analysis, and testing on locally collected samples is recommended before finalizing settings.
Three Common Configuration Mistakes
- Issue: Uploading a scanned PDF of water utility marketing content results in no search results and a
500 Parsing Failedstatus code in the interface. Cause: ThePARSE_OCR_ENABLEconfiguration is not enabled, so no valid text is extracted from the scanned document. - Issue: Enabling the
pdf markerhigh-precision mode triggers anOCR Errorerror during parsing. Cause: The document contains water utility terminology annotations using non-standard fonts, such as handwritten notes for pipeline pressure, and the OCR engine cannot match training set features. - Issue: In version 4.8.9’s simple mode, a single-page water conservation promotional flyer is not automatically parsed. Cause: The
PARSE_TRIGGER_CONDITIONparameter has not been adjusted. The default parsing rule only triggers for long-text files, so short documents are not included in the parsing scope.
How to Verify Correct Configuration
- Upload a single-page PDF of a water utility marketing poster, check if the parsed text blocks include the event time, participation conditions, and unit information from the poster, to confirm the OCR parsing configuration is active.
- Import a document containing a structured water fee report, verify that the chunked content is split according to table fields, to confirm the heading-based chunking configuration is active.
- Initiate a knowledge base search, enter water utility-related terminology such as water supply radius, yuan/cubic meter, check the relevance of recall results, and adjust the similarity threshold to a range that meets business requirements.
- Upload a large water utility marketing manual under 500 MB, confirm that the parsing task completes within the set timeout period with no timeout errors.
Question material comes from public community discussions. Configuration values are common starting points and should be measured against your own samples. Verified on 2026-09-14.