Improve prompt engineering for better tag generation results

[itsmattius]

The current prompt engineering in the OpenAI service provider needs improvement to generate better and more accurate tags.

Current State

• Basic prompt structure is in place
• Tag generation works but could be more accurate
• Limited context understanding

Desired Improvements

• Enhance prompt to better understand content context
• Improve tag relevance and accuracy
• Add better examples and few-shot learning
• Consider content type (technical, documentation, etc.)
• Implement dynamic prompt adjustment based on content length

Technical Details

• File: src/services/providers/openai_service_provider.py:28
• Current implementation uses basic prompt structure
• Need to analyze different content types and their optimal prompts

Acceptance Criteria

• Analyze current prompt effectiveness
• Research best practices for tag generation prompts
• Implement content-type-aware prompting
• Add few-shot examples for different content types
• Test with various content samples
• Measure accuracy improvement