Natural Language Processing

Fit characters into a small table

A journey through hashing characters with Botie.

Natural Language Processing · A journey with Botie

First visit? Learn the workshop controls

Botie's notebook

Could We Allocate Space Before Knowing Every Feature?

Part 1 of 3

A growing text collection can keep introducing new items. If each one requires a new dictionary entry and feature coordinate, the representation grows with the collection. We may instead need a fixed amount of memory and a rule that can handle an item immediately.

Sources for this history

Weinberger and colleagues (2009), Feature Hashing for Large Scale Multitask Learning

← Previous lesson: Finite-State Morphology Read the full lesson and derivations