System and method for retrieving justifiably relevant cases from a case library

• US 5,224,206 A
• Filed: 04/13/1992
• Issued: 06/29/1993
• Est. Priority Date: 12/01/1989
• Status: Expired due to Term

• Current Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Original Assignee: Digital Equipment Corporation (HP Inc.)
• Inventors: Simoudis, Evangelos
• Primary Examiner(s): Fleming, Michael R.
• Assistant Examiner(s): Downs, Robert W.
• Application Number:
  US07/869,203
• Time in Patent Office: 442 Days
• Field of Search: 395/77, 395/50, 395/51
• US Class Current: 706/61
• CPC Class Codes: G06F 11/2257 using expert systems
• CPC Class Codes: G06N 20/00 Machine learning

Abstract

A case-based retrieval system and method for using same for retrieving justifiably relevant cases from a case library in which past problem solving experiences (cases) are stored. The system obtains the symptoms of a new problem. A set of cases are initially retrieved from the case library based on the symptoms of the new problem. It is then determined for each initially retrieved case using a casual explanation for the case whether the case is justifiably relevant to the new problem, with only justifiably relevant cases being returned to the user of the system.

Claims

1. A method of performing case retrieval with a computer that incorporates a case-based reasoning system that has a case library, a case retriever and a case reasoner, comprising the steps of:

(a) obtaining symptoms of a problem;

(b) retrieving a set of cases stored in a case library based on the symptoms of said problem, each said case comprising symptoms, a causal explanation, and a solution;

(c) determining for a case in said set of cases whether that case is justifiably relevant to said problem based on the causal explanation of the case, if the case is determined to be justifiably relevant then step (d) and if the case is determined not to be justifiably relevant then step (e);

(d) automatically accepting as justifiably relevant certain other cases of the retrieved set of cases that are similarly relevant based on information learned during determining step c, then step (f);

(e) automatically rejecting as not justifiably relevant certain other cases of the retrieved set of cases that are similarly not relevant based on the information learned during determining step (c);

(f) repeating step (c) through (e) for the remaining cases of the retrieved set of cases; and

(g) returning to a user of said case-based reasoning system those cases that have been determined to be justifiably relevant to said problem.

2. The method of claim 1, wherein the automatically accepting and rejecting steps further comprise learning whether said cases in said set of cases are justifiably relevant to said problem based upon information obtained during said determining step c, and accepting a first class of cases learned to be justifiably relevant and rejecting a second classes of cases learned to be not justifiably relevant without explicitly matching their casual explanation to said problem.

3. The method of claim 2, further comprising storing the symptoms of said problem, a causal explanation, and a solution of said problem in said case library as an additional case if said problem is solved.

4. The method of claim 3, wherein said determining step includes attempting to match a casual explanation of a retrieved case to said problem.

5. The method of claim 4, wherein the step of attempting to match a causal explanation further comprises using a causal model with the causal explanation.

6. An arrangement for providing justifiably relevant solutions to problems of a computer operating system of a computer, comprising:

(a) a system dump analyzer for coupling to a computer, said system dump analyzer for obtaining information from the computer related to an operating system crash; and

(b) a case-based reasoning system coupled to said system dump analyzer and including a case library and a case retriever, said case retriever operating to retrieve a set of relevant cases from the case library and to determine whether a case of the relevant cases is justifiably relevant based on information obtained by the system dump analyzer, said case retriever automatically accepts as justifiably relevant certain ones of the retrieved set of cases determined to be similarly relevant based on information learned during justification of said case or automatically rejects as not justifiably relevant certain ones of the retrieved set of cases determined to be similarly no relevant based on information learned during justification of said case, and said case retriever accepts as relevant and rejects as not relevant the remaining cases in a same manner.

7. The arrangement of claim 6, wherein the case-based reasoning system further includes a semantic knowledge database coupled to the case retriever, a component causal model database coupled to the case retriever, and a case reasoner coupled to the case retriever.