Skip to content

4 Possible Improvements of AGDISTIS

Ricardo Usbeck edited this page Nov 17, 2017 · 2 revisions

Ideas

  • add ideas from TAGME TO WAT Paper - MISSING
  • java garbage collection mössenböck 2014 - MISSING
  • add machine learning to AGDISTIS on a sample of benchmark texts instead of empirically designing - MISSING
  • use different string similarities - DONE PARTIALLY
  • build a benchmark on two domain different corpora (medicine, music?) - MISSING
  • get chinese texts and transform them to NIF corpus for benchmarking - DONE PARTIALLY
  • use Multi-objective learning of hybrid classifiers - Rok Piltaver, Mitja Luštrek, Jernej Zupančič, Sašo Džeroski, and Matjaž Gams to explain results in FOX and AGDISTIS - MISSING
Clone this wiki locally