HTTP Interfaces and External Systems for Paper Manufacturing

What the data for this category looks like

Paper manufacturing financial report data primarily comes from periodic reports of domestic stock exchanges and publicly available industry data from the China Paper Association. There are three disclosure schedules:

Annual reports are disclosed by April each year
Quarterly reports are released within one month after the end of each quarter
Monthly industry data is updated by the 10th of the following month

Document structures include core financial indicators, detailed production capacity and output data, raw material procurement proportions, and unit energy consumption indicators. Fields include:

Operating revenue (unit: ten thousand RMB)
Paper and paperboard output (unit: ten thousand tons)
Comprehensive unit product energy consumption (unit: tons of standard coal per ten thousand tons of paper)
Wood pulp procurement cost (unit: RMB per ton)

What constraints do these characteristics impose on HTTP interfaces and external systems

The three different disclosure schedules for annual, quarterly, and monthly data require the interface to support the report_type parameter to distinguish data sources for corresponding periods. Split data for business segments (pulp and paper, packaging paper, tissue paper) require the interface to support the business_segment parameter to filter the corresponding segment. The specific unit system (ten thousand tons, tons of standard coal per ten thousand tons of paper) requires the interface to retain original field units when returning data, and not perform forced uniform conversion. The relatively long length of individual financial report documents requires the file parsing interface's timeout threshold to be adapted for long text processing. Multi-dimensional industry segmented data requires batch query interfaces to support pagination parameters to adapt to multi-page data returns.

How to set configurations

Configuration Item	Recommended Approach	Basis for This Approach
`report_type_filter`	Optional values: `annual`/`quarterly`/`monthly`, default `annual`	Adapts to the multi-cycle disclosure characteristics of paper manufacturing financial reports, and accurately matches data sources for corresponding cycles
`PARSE_FILE_TIMEOUT_SECONDS`	`600 seconds`	Individual paper manufacturing financial report documents have relatively long lengths, requiring adaptation to the time consumption demands of long text parsing
`recall_top_k`	`Top 8–12 entries`	Paper manufacturing financial reports include multi-segment detailed data, requiring recall of more relevant fragments to cover business dimensions
`similarity_threshold`	`0.72–0.80`	Financial report data has strong professionalism, requiring a relatively high similarity threshold to filter irrelevant content
`api_return_format`	`json + text`	Supports returns in two formats, adapting to the call demands of different external systems
`kb_id`	Fill in according to the actual ID of the target knowledge base	Binds the unique identifier of the corresponding paper industry financial report knowledge base to avoid cross-library call errors

The parameter values provided on this page are all common recommendations used as starting points for configuration. Actual values are affected by material form, data volume, and business rules. Specific issues require case-by-case analysis, and it is recommended to conduct actual tests on your own samples before finalizing.

Three common errors

Calling the financial report recall interface returns {"code":514,"statusText":"invalid kb config"}. The cause is that the kb_id configuration item was not filled correctly, or the bound knowledge base did not upload the exclusive paper industry financial report data source.
The API return result only includes structured fields, with no natural language analysis text. The cause is that the text format parameter of api_return_format was not enabled, and only structured data from original vector matching was returned.
The number of returned results is lower than expected when querying multiple financial reports in batch. The cause is that a reasonable page_size parameter was not set, and the default pagination number is insufficient to cover segmented data across multiple business segments.

How to confirm the configuration is properly set

Execute a curl command to call the test interface, replace kb_id with the target knowledge base ID, and check whether the returned code field is 0.
Specify api_return_format=text when calling the interface, and check whether the returned result includes natural language formatted financial report analysis fragments.
Pass the business_segment=packaging_paper parameter, and check whether the returned data only includes content related to the packaging paper segment.
View the interface call logs to confirm that the PARSE_FILE_TIMEOUT_SECONDS configuration did not trigger a timeout error.

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.