Inference approach to grammar construction

A system of computer assisted grammar construction (CAGC) is presented in this paper. The CAGC system is designed to generate broad-coverage grammars for large natural language corpora by utilizing both an extended inside-outside algorithm and an automatic phrase bracketing (AUTO) system which is designed to provide the extended algorithm with constituent information during learning. This paper demonstrates the capability of the CAGC system to deal with realistic natural language problems and the usefulness of the AUTO system for constraining the inside-outside based grammar re-estimation. Performance results, including coverage, recall and precision, are presented for a grammar constructed for the Wall Street Journal (WSJ) corpus using the Penn Treebank.

Efficient decoding with generative score-spaces using the expectation semiring

State-of-the-art speech recognisers are usually based on hidden Markov models (HMMs). They model a hidden symbol sequence with a Markov process, with the observations independent given that sequence. These assumptions yield efficient algorithms, but limit the power of the model. An alternative model that allows a wide range of features, including word- and phone-level features, is a log-linear model. To handle, for example, word-level variable-length features, the original feature vectors must be segmented into words. Thus, decoding must find the optimal combination of segmentation of the utterance into words and word sequence. Features must therefore be extracted for each possible segment of audio. For many types of features, this becomes slow. In this paper, long-span features are derived from the likelihoods of word HMMs. Derivatives of the log-likelihoods, which break the Markov assumption, are appended. Previously, decoding with this model took cubic time in the length of the sequence, and longer for higher-order derivatives. This paper shows how to decode in quadratic time. © 2013 IEEE.