real trend calculation and weighting logic

[CodingSelim]

Summary

Implements real trend calculation and weighting for get_trending_topics in the MCP data service, replacing the previous placeholder logic.

Changes Made

• Keyword Statistics: Build detailed stats including frequency, best rank, and matched news items
• Trend Labeling: Compare current vs historical frequencies to label keywords as "rising", "falling", "new", or "stable"
• Weight Scoring: Compute normalized weight scores using configurable weights from config/config.yaml (rank: 60%, frequency: 30%, hotness: 10%)
• Helper Methods: Added 5 internal methods to support the new logic:
  • _calculate_keyword_stats() - Computes keyword statistics from titles
  • _get_historical_keyword_frequency() - Fetches historical data for trend comparison
  • _calculate_trend() - Determines trend labels based on frequency changes
  • _calculate_weight_score() - Calculates normalized weight scores
  • _get_weight_config() - Reads weight configuration from YAML

Bug Fix

• 🐛 Fixed NameError in get_trending_topics: Changed word_frequency to word_groups on line 398

Testing

• Manual verification with actual historical data (2025-11-15)
• Validated weight calculation with config/config.yaml settings
• Confirmed trend detection logic with various frequency scenarios
• Python compilation and import checks passed

File Changes

• mcp_server/services/data_service.py: +167 insertions, -45 deletions