A Bayesian Framework for Cross-Situational Word Learning
Michael C. Frank, Noah D. Goodman, & Joshua B. Tenenbaum (W33)

Graphical model
O objects referential lexicon l I intention

d W wor!s  situations

Word bear bigbird bird birdie book bottle bunnies bunnyrabbit hand hat hiphop kittycat lamb laugh meow mhmm mirror moocow oink on pig put ring sheep

Object bear bird duck duck book bear bunny bunny hand hat mirror kitty lamb cow baby hand mirror cow pig ring pig ring ring sheep

Quantitative results
Bayesian model IBM Translation Model I Mutual information Transitional probability Association frequencies

Inclusion of reference results in a higherprecision lexicon

Corpus data from CHILDES annotated with objects