How royashcenazi/parsigs works
Main Components
a) Core Processing Pipeline:
• Input Layer (Raw Sig Text)
• NER Model Layer
• Rule-Based Processing Layer
• Structured Output Layer
b) Supporting Components:
• Resource Management System (for drug_names.txt and latin_frequency.json)
• Testing Framework
• Research/Training Infrastructure
[Raw Sig Text Input] → [NER Processing] → [Rule-Based Processing] → [Structured Output]
Specific Implementation Details
a) Data Flow:
• How the SigParser class (parse_sig_api.py) orchestrates the processing
• How the StructuredSig object is populated
• Demonstrate resource file utilization
b) Component Integration:
• Integration with SpaCy NER model
• Configuration management (config files in research/config)
• Testing framework integration
Key Architectural Elements
Specific components:
[Resource Management]
• Drug Names Database
• Latin Frequency Dictionary
[Core Parser]
• NER Model
• Rule Engine
• Attribute Extractor
[Output Structure]
• StructuredSig Object
• Drug Information
• Dosage Details
• Frequency Data
• Period Information
Technical Details to Highlight
• Privacy-first architecture (no external APIs)
• Model training pipeline
• Configuration management
• Test coverage integration