A Computational Model for incrementally Learning Word Meanings and Semantic Networks

03 September 2014
Begin time: 
CITEC, room 1.015


We have developed a model of how children learn the meaning of words in the face of ambiguity and noise in the input. By integrating some general processes of memory and attention, we are able to provide a more thorough account of observed behaviours in child word learning.  We show that this integrated model can incrementally and efficiently learn semantic networks of words that capture their meaning similarities. This structured representation has key properties observed in networks proposed to reflect human semantic knowledge.