Problem isolation through weighted search of knowledge bases

• US 8,010,528 B2
• Filed: 03/03/2008
• Issued: 08/30/2011
• Est. Priority Date: 03/03/2008
• Status: Active Grant

• Current Assignee: International Business Machines Corporation
• Original Assignee: International Business Machines Corporation
• Inventors: Trotter, Richard, Nastacio, Denilson, Wamboldt, Michael L.
• Primary Examiner(s): Ali; Mohammad
• Assistant Examiner(s): Shmatov; Alexey
• Application Number:
  US12/041,520
• Publication Number:
  US 20090222436A1
• Time in Patent Office: 1,275 Days
• Field of Search: 707/3, 707/706, 707/722, 707/999.003
• US Class Current: 707/728
• CPC Class Codes: G06F 16/334 Query execution G06F16/335 takes precedence

Abstract

A computer program product for problem isolation through a weighted search of knowledge bases includes computer useable program code that generates an aggregate relevance index which ranks the search results. The aggregate relevance index is calculated using a measure of relevance of each of said pertinent documents across all keyword searches. A method for problem isolation through the weighted search of knowledge bases comprises searching knowledge bases using extracted keywords to identify pertinent documents contained within said knowledge databases; and generating a global rank associated with each of the pertinent documents, the global rank being calculated using a measure of the relevance of each of the pertinent documents across all keyword searches and a measure of the relevance of each of the keywords to the records as a whole.

Claims

1. A computer program product for problem isolation through a weighted search of knowledge bases, the computer program product comprised of:

a computer usable storage medium having computer usable program code embodied therewith, the computer usable program code comprising;

computer useable program code configured to perform multiple searches of said knowledge bases using keywords extracted from records, said records being related to a technical problem within a computer system, said keywords being used to identify pertinent documents contained within said knowledge databases, each of said pertinent documents containing a subset of said keywords;

computer useable program code configured to generate an aggregate relevance index associated with each of said pertinent documents, said aggregate relevance index being a composite of a keyword count which captures significance of searched keywords to said records as a whole and a search rank index which captures overall relevance of each of said pertinent documents over all searches for said searched key words;

said search rank index comprising a ratio which includes a search rank of a said pertinent document among all documents returned in a given search and a total number of documents returned in said given search;

wherein said aggregate relevance index comprises a sum of the search rank index for a said pertinent document across each search for each of said keywords, said aggregate relevance index being equal to the product of said search rank index and said keyword count summed over all keyword searches.

2. The computer program product of claim 1, wherein each of said pertinent documents are identified by a uniform resource locator or a file path.

3. The computer program product of claim 1, wherein the search rank of each pertinent document during each search and total number of search results are recorded, said search rank and said total number of search results for each pertinent document being incorporated into an aggregate relevance index for each pertinent document.

4. The computer program product of claim 1, wherein said pertinent documents are displayed in descending order of said aggregate relevance index.

5. The computer program product of claim 1, in which said search rank index comprises:

6. The computer program product of claim 5, in which said aggregate relevance index further comprises a relevance index, said relevance index comprising a composite measure of relevance of a keyword to said records as a whole and said search rank index, said relevance index being defined by:

7. The computer program product of claim 6, in which said aggregate relevance index further comprises an aggregate relevance index value for each URL, said aggregate relevance index value comprising a sum of said relevance index over all searches for keywords k₁ through k_m;

said aggregate relevance index value for each URL being defined by;

8. The computer program product of claim 1, in which said aggregate relevance index comprises a measure of relevance of a particular URL across multiple related searches;

each URL returned by said multiple related searches being assigned an aggregate relevance index value, wherein at least a portion of URLs returned by said multiple related searches are displayed to a user by descending aggregate relevance index value.

9. A method for problem isolation through a weighted search of knowledge bases comprising:

obtaining keywords, said keywords being extracted from records, said records being related to a technical problem within a computer system;

searching said knowledge bases using said keywords to identify pertinent documents contained within said knowledge databases, each of said pertinent documents containing a subset of said keywords;

generating a global rank associated with each of said pertinent documents, said global rank comprising a measure of relevance of each of said pertinent documents across all keyword searches and a measure of relevance of each of said keywords to said records as a whole, the global rank comprising an aggregate relevance index generated by;

summing a rank for each of said pertinent documents across each search for each of said keywords to produce a search rank index, the search rank index comprising a ratio which includes a search rank of a said pertinent document among all documents returned in a given search and a total number of documents returned in said given search, the search rank capturing the relevance of each of said pertinent documents across all keyword searches;

multiplying said search rank index by a keyword count to produce a weighted search rank, the keyword count capturing relevance of each of said keywords to said records as a whole; and

summing said weighted search rank index over all keyword searches to produce the aggregate relevance index.

10. The method of claim 9, wherein said measure of relevance of each of said keywords to said records as a whole comprises generating a keyword count associated with each of said keywords, said keyword count being equal to a total number of incidences of a said keyword within said records;

and said measure of relevance of each of said pertinent documents across all keyword searches comprises forming a search rank index by measuring a rank of a said pertinent document within all other documents returned by a search engine for a said keyword.

11. The method of claim 9, further comprising displaying said pertinent documents in descending order of said aggregate relevance index.