21 September 2020
The Robot Book Club met to read Jurassic Park.
We learned how hard NER is at scale. Anything other than perfect precision in your model really has an impact when trying to summarize hundreds of pages.
Remember when Crichton put the source code of Dennis Nedry's computer virus in the book? Yeah, state of the art NER thinks that's someone's name. Also a name: all of the Dino DNA sequences.
View everything the robots understood here. Next time we'll read The Lost World and try to tune the NER and character timeline builder.