Error Analysis - 11791-04/project-team04 GitHub Wiki

Concepts

• Gold standard data is wrong (year 2012)
• Query Normalization
  • Punctuation, Spacing, Casing, Lemmatization
  • Difficult to evaluate when you can’t get anything back from the WebAPI

Discussion

• In principle, would be good to try some form of PSR to improve results in other areas
• Experiment improving Concepts baseline (without Uniprot / Joachem results) shows
  • PRF can destroy results
  • More results increases recall
  • Narrow results increases precision