Identiy Resolution: Webdata Commons and Spotify

[jjakubassa]

Baseline

Baseline: https://github.com/jjakubassa/WDI-Project/commit/15a639dcbdc1babaef7f5d0a009f55e41274035d

Report

• Content and size of your gold standard
  • Which classes/data sets are included?
  • WDC and Spotify albums
  • What „corner cases“ did you include?
  • ...
• Which matching rules did you try?
  • What happens with P/R/F1?
  • Precision: 1,0000
  • Recall: 0,6667
  • F1: 0,8000
  • Which attribute comparators / similarity measures did you use?
  • Album Title
    • First character of each token
  • Artist Name
  • Track titles
  • Total tracks
  • Duration in s
• What blockers have you tried?
  • What happened with runtime and number of matches?
  • from memory error (no blocker) to 1s with reduction ratio: 0.9997320283848182
  • What blockers / blocking functions have you used?
  • Album Title
    • First character of each token
  • Artist Name
  • Track titles
  • Total tracks
  • Duration in s
  • How do P/R/F1 change, and why?
  • no comparison to noblocker possible, since noblock throws error

[jjakubassa]