Method for extracting open-issue data from textual specifications using natural language text processing

• US 5,500,796 A
• Filed: 07/22/1993
• Issued: 03/19/1996
• Est. Priority Date: 07/22/1993
• Status: Expired due to Fees

• Current Assignee: General Electric Company
• Original Assignee: General Electric Company
• Inventors: Black, James E. Jr.
• Primary Examiner(s): Weinhardt, Robert A.
• Application Number:
  US08/094,740
• Time in Patent Office: 971 Days
• Field of Search: 395/600, 364/419.19, 364/419.03, 364/419.08, 364/419.01, 364/419.13, 364/419.06
• US Class Current: 704/9
• CPC Class Codes: G06F 40/253 Grammatical analysis; Style critique

Abstract

A method for extracting open-issue data from a textual specification using a computer is disclosed. The method includes searching for predetermined trigger token morphemes in an enhanced text corpus based upon the textual specification and constituted of respective enhanced sentences so as to electronically mark respective ones of the enhanced sentences based upon the presence of predetermined trigger token morphemes. Linguistic analysis can then be executed upon each remaining unmarked sentences to detect an interrogatory construct in respective ones of the remaining sentences. Sentences marked based either upon the presence of the trigger token morpheme or upon the presence of the interrogatory construct constitute the open-issue data extracted from the textual specification undergoing analysis.

Claims

1. A natural language text processing method using a computer for extracting open-issue data from a textual specification, said method comprising the steps of:

providing an enhanced text corpus based upon said textual specification to said computer, said enhanced text corpus comprising electrical signals representative of enhanced sentences each constituted of a stream of enhanced corpus words;

scanning each of said enhanced sentences for determining the presence of a predetermined trigger token morpheme in each said stream of enhanced corpus words;

marking said enhanced sentences having the presence of said predetermined trigger token morpheme, the marked enhanced sentences indicative of one portion of said open-issue data;

executing predetermined linguistic analysis upon each remaining unmarked sentence for determining the presence of a predetermined interrogatory construct in said remaining sentences;

marking said remaining sentences having the presence of said interrogatory construct, the marked remaining sentences indicative of another portion of said open-issue data; and

extracting the marked portions indicative of said open-issue data.

2. A method in accordance with claim 1 wherein said predetermined linguistic analysis includes detecting a predetermined query signal in said remaining sentences.

3. A method in accordance with claim 1 wherein said predetermined linguistic analysis includes detecting a question mark signal in said remaining sentences.

4. A method in accordance with claim 1 wherein said predetermined linguistic analysis includes detecting a linguistic void signal in said remaining sentences.

5. A method in accordance with claim 1 wherein said predetermined linguistic analysis includes detecting semantically incongruous signals in said remaining sentences.

6. A method in accordance with claim 1 wherein said predetermined linguistic analysis comprises:

detecting a question mark signal in said remaining sentences;

detecting a predetermined query signal in said remaining sentences;

detecting a linguistic void signal in said remaining sentences; and

detecting semantically incongruous signals in said remaining sentences;

wherein detection in any of the foregoing recited detecting steps establishes the presence of said interrogatory construct.

7. A natural language text processing method using a computer for extracting open-issue data from a textual specification, said method comprising the steps of:

providing an enhanced text corpus based upon said textual specification to said computer, said enhanced text corpus comprising electrical signals representative of enhanced sentences each constituted of a stream of enhanced corpus words;

executing predetermined linguistic analysis upon each enhanced sentence for determining the presence of a predetermined interrogatory construct in said enhanced sentences, said predetermined linguistic analysis comprising;

detecting for a question mark signal in said enhanced sentences;

detecting for a predetermined query signal in said enhanced sentences;

detecting for a predetermined cue signal in said enhanced sentences;

detecting for a linguistic void signal in said enhanced sentences; and

detecting for semantically incongruous signals in said enhanced sentences;

wherein detection in any of the foregoing recited detecting steps establishes the presence of said interrogatory construct;

marking said enhanced sentences having the presence of said interrogatory construct; and

extracting the marked sentence indicative of said open-issue data.