Data Characteristics
Data for home healthcare products primarily comes from product manuals, user guides, official website product detail pages, and medical device registration certificates. This data updates infrequently, typically with new product models or regulatory changes. Documents are semi-structured, containing extensive text descriptions of product parameters, usage instructions, contraindications, precautions, and maintenance. They also include images and charts. Common fields include product name, model, manufacturer, registration number, scope of application, technical specifications (e.g., measurement accuracy, battery life), and material composition. Units vary, covering length (mm, cm), weight (g, kg), time (hours, days), numerical values (mV, Pa), and specific medical units (e.g., mmHg, mmol/L).
Constraints Imposed by These Characteristics on Forms and Interactions
The low update frequency of home healthcare product data means less daily maintenance pressure after knowledge base construction. However, initial data entry and cleaning require significant effort. Semi-structured documents require flexible form designs to adapt to unique product fields and support rich text input for detailed explanations. Diverse units and technical specification fields require forms to include unit validation and range restrictions to prevent invalid user input. For example, blood pressure monitor measurement ranges and blood glucose meter calibration fluid concentrations need strict numerical constraints. Product involvement with user health requires interaction design to emphasize clarity, accuracy, and ease of use, avoiding ambiguity, especially when consulting about contraindications and precautions.
Configuration Settings
| Configuration Item | Recommended Value | Rationale |
|---|---|---|
maxContext | 2000 characters | Home healthcare product manuals are often long. Sufficient context length is needed to provide complete product information and usage guidance. |
minSimilarity | 0.8 | This ensures retrieved knowledge points are highly relevant to user queries, reducing recall of inaccurate information, especially for precise matches involving product models and technical parameters. |
segmentLength | 500 characters | A moderate segment length effectively captures key information blocks in product manuals. It avoids overly long segments diluting the topic or overly short segments losing context. |
recallTopN | top 5 | This provides enough relevant knowledge points for the model to choose from while ensuring recall quality. It avoids excessive redundant information that increases processing burden. |
variableNames | Product Name, Model, Registration No. | This ensures forms can directly reference and match core product information frequently asked by users, improving consultation efficiency. |
inputValidationRules | numerical range validation, unit matching, enumeration value restriction | For numerical parameters like blood pressure and blood glucose, set reasonable input ranges. Validate units against standards (e.g., mmHg). For specific parameters (e.g., battery type), use enumeration value restrictions to ensure data accuracy. |
Common Mistakes
- When selecting a knowledge base for search, if no optional values appear when referencing variables, the corresponding entity variables are not configured or extracted in the knowledge base.
- If the system prompts "Invalid parameter input" after a user enters numerical parameters in a form, the form lacks unit validation or range restrictions for the input value, causing it to exceed the preset valid range.
- When integrating with external systems like Feishu, if form input functionality fails to transmit data correctly, the API interface parameter names might not match, or the data format might not meet the target system's requirements.
How to Verify Configuration
- Submit simulated queries with different product models and technical parameters. Check if the returned results accurately mention the corresponding product information and specifications.
- Test inputting values outside reasonable ranges or text that does not conform to preset formats. Confirm that the form correctly triggers validation prompts and prevents invalid submissions.
- Call the form function through an external platform (e.g., Feishu). Check if data is transmitted completely and accurately to FastGPT and triggers the corresponding knowledge base query.
- For specific products, simulate user questions about contraindications or precautions. Verify that FastGPT accurately recalls and presents relevant warning information.
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 2026-09-21.