Natural Language Processing

Choose the piece that covers more new ground

A journey through greedy coverage with Botie.

Natural Language Processing · A journey with Botie

First visit? Learn the workshop controls

Botie's notebook

Which Addition Gives Us the Most New Coverage?

Part 1 of 3

Suppose we have room for only a few vocabulary additions. Two candidate pieces may each cover many occurrences, but much of their coverage may overlap. Selecting both on their standalone counts can spend the second slot on information the first already supplied.

Sources for this history

Nemhauser, Wolsey and Fisher (1978), An Analysis of Approximations for Maximizing Submodular Set Functions

← Previous lesson: SentencePiece Read the full lesson and derivations