A Structured Language Model

  Request Code

  API Access   Call/Text an Expert

The paper presents a language model that develops syntactic structure and uses it to extract meaningful information from the word history, thus enabling the use of long distance dependencies. The model assigns probability to every joint sequence of words - binary-parse-structure with headword annotation. The model, its probabilistic parametrization, and a set of experiments meant to evaluate its predictive power are presented.