Method of providing computer generated dictionary and for retrieving natural language phrases therefrom

• US 5,383,121 A
• Filed: 05/05/1992
• Issued: 01/17/1995
• Est. Priority Date: 09/11/1991
• Status: Expired due to Term

• Current Assignee: Mitel Networks Corporation
• Original Assignee: Zarlink Semiconductor Inc. (Microchip Technology Incorporated)
• Inventors: Letkeman, Kim D.
• Primary Examiner(s): Weinhardt, Robert A.
• Application Number:
  US07/878,854
• Time in Patent Office: 987 Days
• Field of Search: 364/419, 364/419.11, 364/419.15, 364/419.08, 364/419.04, 364/419.07
• US Class Current: 704/10
• CPC Class Codes: H03M 7/30 Compression speech analysis-synthesis for redundancy reduction G10L19/00; for image communication H04N; Expansion; Suppression of unnecessary data, e.g. redundancy reduction

Abstract

A method of translating a command into natural language is comprised of the steps of: storing compressed words of the language once as either a word or subset of a word or group of adjoining words in a memory; storing a numeric indicator of each group of tokens which form a natural language string, with position locators; looking up the position locators in response to the command; accessing the memory to retrieve the compressed words; and providing a string of the tokens as a natural language string.

Claims

1. A method of providing a computer-generated natural language dictionary, said computer having computer-associated memory, comprising the steps of:

(a) subdividing storage space of said computer-associated memory into a plurality of memory tables;

(b) generating a plurality of natural language dictionary strings comprised of a plurality of separate tokens, respective ones of which tokens comprise groups of characters separated by one or more spaces;

(c) assigning a separate numeric value as a token number to each different token, whereby the same numeric value is assigned as the same token number to common tokens of various strings;

(d) storing the tokens of said natural language dictionary in said computer-associated memory;

(e) storing a first memory table in said computer-associated memory, said first memory table having an index and a list, said index being comprised of a first quantity representative of the number of strings in said list of natural language dictionary strings, and successive second quantities representative of the positions of respective strings in said plurality of natural dictionary strings, each position being immediately followed by a third quantity representative of the number of tokens in an associated dictionary string, said list being comprised of the assigned numeric value for each token in each of said strings; and

(f) storing a second memory table in said computer-associated memory, said second memory table containing a correlation of each token number, a storage location in said computer-associated memory at which each token starts, and the length of the token.

2. A method according to claim 1, further comprising the step of (g) retrieving natural language phrases from said computer-generated natural language dictionary by:

(g1) receiving a command containing tokens;

(g2) correlating tokens in said command with token numbers represented by or responsive to said command;

(g3) accessing said first and second tables in accordance with token numbers correlated in step (g2), so as to read out tokens stored in said computer-associated memory; and

(g4) providing the tokens read out in step (g3) in a prescribed order to a user.

3. A computer-generated natural language dictionary comprising a computer and a computer-associated memory, said memory having a first portion which stores tokens of said natural language dictionary, respective ones of said tokens comprising groups of characters separated by one or more spaces, and wherein a separate numeric value is assigned as a token number to each different token, whereby the same numeric value is assigned as the same token number to common tokens of various natural language dictionary strings of separate tokens, said memory further having a first memory table which stores an index and a list, said index being comprised of a first quantity representative of the number of strings in said plurality of natural language dictionary strings, and successive second quantities representative of the positions of respective strings in said plurality of natural dictionary strings, each position being immediately followed by a third quantity representative of the number of tokens in an associated dictionary string, said list being comprised of the assigned numeric value for each token in each of said strings, said memory further including a second memory table which stores a correlation of each token number, a storage location in said computer-associated memory at which each token starts, and the length of the token,

and wherein said computer is operative, in response to receipt of a command containing tokens, to correlate tokens in said command with token numbers represented by or responsive to said command, and to access said first and second tables in accordance with correlated token numbers so as to read out tokens stored in said computer-associated memory and to provide the tokens read out in a prescribed order to a user.