Word predicting method, voice recognition method, and voice recognition apparatus and program using the same methods
AssigneeInternational Business Machines Corporation
InventorsShinsuke Mori, Masafumi Nishimura, Nobuyasu Itoh
Life of this patent
Estimating the term…
Abstract
A word predicting method for use with a voice recognition using a computer includes the steps of specifying a sentence structure of a history up to a word immediately before the word to be predicted, referring to a context tree stored in arboreal context tree storage section having information about possible structures of a sentence and a probability of appearance of words with respect to the structures at nodes, and predicting words based on the context tree and the specified sentence structure of the history.
Claims6
A word predicting method of predicting words in a predetermined sentence by using a computer, comprising the steps of:
- retrieving a history tree of word prediction covering words before a word to be predicted to be used in predicting words from a word prediction history storage means where a word prediction history of partial parse trees is stored;
- dynamically selecting a shape of a partial parse tree as a word prediction reference range by specifying a sentence structure of said history;
- acquiring an arboreal context tree for word prediction from a context tree storage means, which stores said context tree having information about possible structures of a sentence and a probability of appearance of words with respect to said structures at nodes;
- comparing at least one node of said arboreal context tree to said history tree;
- identifying a node of said arboreal context tree having a partial tree matched with said history tree;
- using a probability distribution appended to the identified node of the arboreal context tree to predict a word to be predicted;
- and
- outputting the predicted word.
A data processing method comprising the steps of:
- retrieving a processing history tree of word prediction covering words before a word to be predicted to be used in predicting a predetermined element from a word prediction history storage unit storing said processing history of partial parse trees for an array;
- dynamically selecting a range of processing history for use as a word prediction reference range by acquiring a stochastic model from a stochastic model storage unit storing the stochastic model for the tree structure having predetermined partial trees and a probability distribution associated with said partial trees at nodes;
- comparing at least one node of said stochastic model to said history tree to identify a node of said stochastic model having a partial tree corresponding to the tree structure of said processing history for said stochastic model,
- using a probability distribution appended to the identified node of the stochastic model to predict said predetermined element;
- and
- outputting the predicted element.
A computer-readable program storage medium readable by computer and tangibly embodying a program of instructions stored on the medium and executable by the machine for causing the computer to perform a method for predicting words in a predetermined sentence by controlling a computer,
- wherein said method comprises the steps of:
- retrieving a history tree of word prediction covering words before a word to be predicted to be used in predicting words from a word prediction history storage means where a word prediction history of partial parse trees is stored;
- dynamically selecting a shape of a partial parse tree as a word prediction reference range by specifying a sentence structure of said history;
- acquiring an arboreal context tree for word prediction from a context tree storage means, which stores said context tree having information about possible structures of a sentence and a probability of appearance of words with respect to said structures at nodes;
- comparing at least one node of said arboreal context tree to said history tree;
- identifying a node of said arboreal context tree having a partial tree matched with said history tree;
- using a probability distribution appended to the identified node of the arboreal context tree to predict a word to be predicted;
- and
- outputting the predicted word.
A computer-readable program storage medium readable by computer and tangibly embodying a program of instructions stored on the medium and executable by the machine for causing the computer,
- wherein said method comprises the steps of:
- retrieving a processing history tree of word prediction covering words before a word to be predicted to be used in predicting a predetermined element from a word prediction history storage unit storing said processing history of partial parse trees for an array;
- dynamically selecting a range of processing history for use as a word prediction reference range by acquiring a stochastic model from a stochastic model storage unit storing the stochastic model for the tree structure having predetermined partial trees and a probability distribution associated with said partial trees at nodes;
- comparing at least one node of said stochastic model to said history tree to identify a node of said stochastic model having a partial tree corresponding to the tree structure of said processing history for said stochastic model,
- using a probability distribution appended to the identified node of the stochastic model to predict said predetermined element;
- and
- outputting the predicted element.
Dependencies are inferred from the claim text.