Anticipated meaning natural language interface

• US 5,682,539 A
• Filed: 09/29/1994
• Issued: 10/28/1997
• Est. Priority Date: 09/29/1994
• Status: Expired due to Fees

• Current Assignee: Leverance, Inc.
• Original Assignee: Charles Cosby, Donovan Conrad
• Inventors: Conrad, Donovan, Cosby, Charles
• Primary Examiner(s): Weinhardt, Robert A.
• Assistant Examiner(s): Shingala, Gita
• Application Number:
  US08/315,240
• Time in Patent Office: 1,125 Days
• Field of Search: 364/419.08, 364/419.07, 364/419.13, 364/419.14, 395/600
• US Class Current: 704/9
• CPC Class Codes: G06F 40/40 Processing or translation of natural language natural language analysis G06F40/20; semantic analysis G06F40/30

Abstract

An anticipated meaning natural language interface system for computer applications. A knowledge engineer anticipates the general meaning of each sentence that a user is likely to enter and builds a structure of general meaning nodes. For each node, the knowledge engineer enters one or more typical sentences to represent the general meaning. A knowledge engineer abstracts the typical sentences and stores the abstractions in a knowledge base. When a user enters a sentence, it is abstracted by the system and compared to abstracted typical sentences in the knowledge base. This information, and other available information, is used by an algorithm to determine which of the general meaning nodes is intended by the user.

Claims

1. A method of creating a natural language interface for a computer application, the computer, interface, and application together comprising a system, which interface allows a user to direct functions of the system with natural language input, comprising:

(a) based on a plurality of functions of the system, anticipating the general meaning of each of a plurality of likely user input sentences and storing in the computer a plurality of general meaning nodes, one for each anticipated user input general meaning;

(b) for a plurality of general meaning nodes, associating each node with a function, entering at least one typical anticipated user-input sentence which conveys the general meaning of the node, generating a pattern from the words of the typical sentence, and storing the typical sentence pattern in the computer;

(c) receiving from a user an input sentence and generating a pattern from the words of the input sentence;

(d) applying an algorithm stored in the computer to select which one of the plurality of general meaning nodes is intended by the user by comparing the input sentence pattern to the typical sentence patterns; and

(e) executing the function associated with the selected general meaning node.

2. The method of claim 1 further comprising:

(a) identifying an object of the system that can be directed by the user, anticipating object terms that a user is likely to use to refer to the object, and storing the object terms in the computer with an associated reference to the object;

(b) in the process of generating a typical sentence pattern from a typical sentence, replacing the object terms with a general category code and;

(c) in the process of generating an input sentence pattern from an input sentence, replacing the object terms with the general category code.

3. The method of claim 2 further comprising:

(a) adjusting the algorithm to also consider whether object terms are contained in the input sentence.

4. The method of claim 2 further comprising:

(a) for a general meaning node, specifying an action to be taken by the system when the general meaning node is selected, which action specifies the general category code, and;

(b) when the general meaning node is selected, replacing the general category code within the action with the associated reference to the object and executing the action.

5. The method of claim 3 further comprising:

(a) anticipating node terms that a user is likely to use to refer to a general meaning node and storing the node terms in the computer;

(b) in the process of generating a pattern from a typical sentence, replacing the node terms with a general category code;

(c) in the process of generating a pattern from an input sentence, replacing the node terms with the general category code; and

(d) adjusting the algorithm to also consider whether the input sentence includes the node terms.

6. The method of claim 3 further comprising:

(a) anticipating node group terms that a user is likely to use to refer to the subject matter of a group of general meaning nodes and storing the node group terms in the computer with a reference to the group of general meaning nodes;

(b) in the process of generating a pattern from a typical sentence, replacing the node group terms with a general category code;

(c) in the process of generating a pattern from an input sentence, replacing the node group terms with the general category code; and

(d) adjusting the algorithm to also consider whether the input sentence includes the node group terms.

7. The method of claim 1 further comprising:

(a) organizing the general meaning nodes with determinable relationships between them;

(b) as a node is selected, storing in a memory the last previously selected node;

(c) adjusting the algorithm to also consider the relationship between the last previously selected node and the general meaning nodes.

8. The method of claim 7 further comprising:

(a) adjusting the algorithm to also consider the relationships between nodes selected prior to the last previously selected node and the general meaning nodes.

9. The method of claim 8 further comprising:

(a) associated with each of a plurality of the general meaning nodes, storing information regarding each node;

(b) adjusting the algorithm to also consider the information stored in the computer in association with the last previously selected node or nodes selected prior to the last previously selected node.

10. The method of claim 3 further comprising:

(a) storing a record of object terms entered by the user in a plurality of successive input sentences;

(b) adjusting the algorithm to also consider object terms entered by the user in input sentences prior to the current input sentence.

11. The method of claim 5 further comprising:

(a) storing a record of node terms entered by the user in a plurality of successive input sentences;

(b) adjusting the algorithm to also consider node terms entered by the user in input sentences prior to the current input sentence.

12. The method of claim 6 further comprising:

(a) storing a record of node group terms entered by the user in a plurality of successive input sentences;

(b) adjusting the algorithm to also consider node group terms entered by the user in input sentences prior to the current input sentence.

13. The method of claim 4 further comprising:

(a) before executing the action, presenting to the user a selected typical sentence and seeking confirmation from the user that the input sentence has the same general meaning as the typical sentence.