Data

Collection

Articles scraped

(JSON / CSV)

LLM-generated summaries (GPT-4o)

(CSV)

LLM-generated evaluations (Llama)

• Old prompt evaluation_results_aggregated_2.csv
  • = llm_evalution_2.csv
• New prompt: evaluation_results_aggregated_0503.csv
  • = llm_evaluation_updated_prompt.csv
  • ✅ processed: llm_evaluation.csv
    • (added col 1 "ID-Global",
    • dropped col 2 "Article File Name",
    • dropped 3 rows)

Human-generated evaluations

• Batch #1: Human_Evaluation_Results.csv
  • = human_eval_1_all.csv
  • processed: human_evalutation.csv
    • (dropped col 1 "Evaluator Name",
    • dropped articles_number 320, 39, 86, 153, 42, 132, 39, 21, 64, 7)
• Batch #2: Human_Evaluation_Results_differences.csv
  • = human_eval_2_all.csv

Processed data

[wip]