Inference machine

• US 20070011125A1
• Filed: 06/02/2006
• Published: 01/11/2007
• Est. Priority Date: 12/03/2003
• Status: Active Grant

• Current Assignee: SemEO Services GmbH, The Ohio State University Research Foundation
• Original Assignee: Ontoprise GmbH
• Inventors: Angele, Jurgen
• Granted Patent: US 7,333,970 B2
• Time in Patent Office: Days
• US Class Current: 706/46
• CPC Class Codes: G06N 5/045 Explanation of inference; Explainable artificial intelligence [XAI]; Interpretable artificial intelligence

Abstract

The invention relates to an inference machine which is used to answer questions relating to predetermined data with the aid of an ontology, which is used to structure data, and a declarative system of rules, which reproduces additional knowledge. During the evaluation of a query, inference protocols, relating to the instantiation of the regulations which appear during evaluation, are read in an evaluation unit associated with the inference unit. Explanations relating to the evaluation of the rules are generated in the evaluation unit according to the inference protocols. A logical derivation of the answer becomes transparent. The system can answer technical questions and can carry out and explain technical considerations.

Claims

1. A data processing system

a) having at least one information generating unit for storing or generating data forming a database;

b) having at least one computer unit for generating an ontology comprising a class structure and a declarative system of rules,

b1) wherein the data can be structured by means of the ontology, and

b2) wherein the rules combine elements of the class structure and/or data;

c) having an input/output unit for inputting a query and for outputting an answers to the query; and

d) having a first inference unit in which the rules are evaluated for generating the answer to the query;

wherein

e) an evaluation unit is allocated to the first inference unit, wherein the evaluation unit comprises a second inference unit;

f) within the second inference unit an explanation of the answer to the query is generated by evaluating predefined explanation rules;

g) wherein inference protocols can be read into the evaluation unit, which state, which rules were evaluated for answering the query, and which contain the facts, which were used for instantiation of the rules for answering the query; and

h) wherein for the evaluation of the explanation rules for generating an explanation of the answer the inference protocols form instances of the explanation rules.

2. The data processing system as claimed in claim 1,

wherein

the explanation rules are expressed in F-Logic.

5. The data processing system as claimed in claim 1,

wherein

a selection of various explanation possibilities is performed by means of explanation rules.

6. The data processing system as claimed in claim 1,

wherein

the explanations are dependent on external data.

7. The data processing system as claimed in claim 6,

wherein

the external data are structured by means of an ontology.

8. The data processing system as claimed in claim 6,

wherein

the external data are formed by user profiles of the users performing the queries.

9. The data processing system as claimed in claim 6,

wherein

the external data form instances for the explanation rules.

3. The data processing system as claimed in claim

wherein

one explanation is allocated to a plurality of rules.

4. The data processing system as claimed in claim 3,

wherein

for generating an explanation for a predetermined number of rules, an explanation rule or a group of explanation rules is provided by means of which the rules are described summarily.

10. A method for querying information, wherein the following steps are performed, by using a data processing system:

a) a database is generated or stored;

b) an ontology is generated which comprises a class structure and a declarative system of rules,

b1) wherein the data can be structured by means of the ontology, and

b2) wherein the rules combine elements of the class structure and/or data;

c) a query is input;

d) a first inference unit evaluates the query by evaluating the rules and generates an answer;

e) within a second inference unit comprised in an evaluation unit allocated to the first inference unit an explanation of the answer to the query is generated by evaluating predefined explanation rules;

f) wherein inference protocols are read into the evaluation unit, which state, which rules were evaluated for answering the query, and which contain the facts, which were used for instantiation of the rules for answering the query; and

g) wherein for the evaluation of the explanation rules for generating an explanation of the answer the inference protocols form instances of the explanation rules.

11. A computer program, wherein, when it runs on a computer or on a multiplicity of computers of a computer network, it executes the method as claimed in claim 10.

12. A computer program with program code means for performing a method as claimed in claim 10, when the computer program is executed on a computer or on a multiplicity of computers of a computer network.

13. The computer program with program code means as claimed in claim 12, which are stored on a computer-readable data medium.

14. A data medium on which a data structure is stored which, after being loaded into an internal and/or main memory of a computer or of a multiplicity of computers of a computer network, executes the method as claimed in claim 10.

15. A computer program product with program code means stored on a machine-readable medium for performing all steps as claimed in claim 10 when the program is executed on a computer or on a multiplicity of computers of a computer network.

16. A modulated data signal which contains instructions for executing a method as claimed in claim 10, which can be executed by a computer system or by a multiplicity of computers of a computer network.