Method and apparatus for intranet searching

• US 7,779,012 B2
• Filed: 11/12/2004
• Issued: 08/17/2010
• Est. Priority Date: 10/15/2004
• Status: Expired due to Fees

• Current Assignee: Microsoft Technology Licensing LLC (Microsoft Corporation)
• Original Assignee: Microsoft Corporation
• Inventors: D'Urso, Mark S.
• Primary Examiner(s): Pham; Khanh B
• Assistant Examiner(s): Oberly; Van H
• Application Number:
  US10/987,283
• Publication Number:
  US 20060085447A1
• Time in Patent Office: 2,104 Days
• Field of Search: 707/748, 707/723
• US Class Current: 707/748
• CPC Class Codes: G06F 16/24578 using ranking
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques

Abstract

A search system for finding information relevant on an intranet in response to a user query. Separate from the query, the system crawls the intranet to select at least some of the pages in the intranet. The system determines, for items in a selected page, an indication of the specificity or generality of that information based on structural information, such as display formatting or a number of links in a shortest path from a root page to the identified page. The system can also determine the level of specificity of a search query. When a user provides a query to request information over the intranet, pages matching the query can be ranked by comparing the level of specificity or generality of the query to the level of specificity or generality of the terms that match the query.

Claims

1. A method for creating data for use in ranking results of a search of a plurality of pages in an intranet, each page comprising a plurality of terms arranged in a hierarchical structure, the method comprising acts of:

(A) parsing the plurality of pages to identify the plurality of terms, wherein at least one page of the plurality of pages comprises a plurality of instances of a term of the plurality of terms;

(B) automatically computing, for each instance of each identified term and for each page in the plurality of pages in which the identified term appears, a specificity value of the instance, the specificity value being computed, for each instance of the identified term on the page, based upon a combination of a tag parameter associated with the instance and a term level of the instance within the hierarchical structure of the page, the term level being measured from at least one root level of the page;

(C) the act of combining, for each identified term and for each page in the plurality of pages in which the identified term appears, the computed specificity values of the instances of the identified term within the page to produce a combined computed specificity value of the identified term;

(D) storing, for each identified term and for each page in the plurality of pages in which the identified term appears, the combined computed specificity value of the identified term and an identifier for the page; and

(E) ranking a plurality of matching pages in the plurality of pages, the matching pages returned in response to a search query performed on the plurality of pages, the ranking based on a comparison, for each matching page, between the combined computed specificity value stored for an identified term matching a term of the query and a degree of specificity associated with the query, wherein the matching pages are identified as including an identified term matching a term of the query based on the identifier for the page stored for the identified term.

2. The method of claim 1, wherein the hierarchical structure comprises a plurality of levels, and wherein automatically computing the specificity value of the instance of the identified term further comprises determining a number of levels in a shortest path from a level of the at least one root level to the term level of the instance of the identified term.

3. The method of claim 2, wherein at least one page in the plurality of pages is arranged according to HTML and the at least one root level is associated with an HTML root tag.

4. The method of claim 3, wherein the levels in the hierarchical structure of the at least one page are associated with levels of nodes in the HTML structure of the page.

5. The method of claim 4, wherein the levels of nodes in the HTML structure are associated, at least in part, with tag parameters describing formatting characteristics.

6. The method of claim 5, wherein the formatting characteristics comprise at least one tag type selected from the group of tag types comprising a distinguishing font, a distinguishing color, a hypertext link, a title, a coding tag, and a meta tag.

7. The method of claim 6, wherein automatically computing the specificity value for each instance of each identified term is performed in association, at least in part, with a specificity weight factor of a tag type associated with the identified term.

8. The method of claim 7, wherein the specificity weight factor of the tag type is based, at least in part, on a frequency of the tag type in the intranet.

10. The method of claim 1, wherein the specificity value is greater for an instance of an identified term having a greater distance from a root level of the hierarchical structure than a specificity value for an instance of an identified term having a relatively shorter distance from the root level.

19. The method of claim 1, wherein act (B) comprises computing a normalized overall term specificity value (NTW) for a term in a page, relating to all appearances of the term in the page, according to the following formulae:

NLW=NL−

TTL;

TW=TagW+NLW;

NTW=log(STW/ASTW); and

wherein;

NL is a raw term level of the term within the hierarchy of the page, as measured from a root tag of the page;

TTL is a threshold term level defining a normalized minimum nesting level of the term for all pages of the intranet including the term;

TagW is a tag specificity parameter ascribed to the term based on an HTML tag type of the immediate parent of the term;

STW is a sum of all the TW values for all appearances of a term in a particular page; and

ASTW is an average of all STW parameter values for a term for all pages in which the term appears in the intranet.

9. The method of 1, wherein the act (B) further comprises computing the specificity value of the instance of the identified term based, at least in part, on a specificity value determined for each matching page.

11. A method for creating a term index for use in ranking results of a search of a plurality of pages in an intranet, each page comprising a plurality of terms arranged in a hierarchical structure, the method comprising acts of:

(A) parsing the plurality of pages to identify the plurality of terms, wherein at least one page of the plurality of pages comprises a plurality of instances of a term of the plurality of terms;

(B) for each of the plurality of pages;

(i) automatically computing, for each instance of each identified term, a specificity value of the instance based upon a combination of a tag parameter associated with the instance and a term level of the instance within the hierarchical structure of the page, the term level being measured from a root tag of the page;

(ii) combining, for each identified term, the computed specificity value of the instances of the identified term within the page to produce a combined specificity value for the identified term; and

(iii) storing in the term index for each identified term, the page in which the identified term appears, and information identifying the identified term in conjunction with information indicating the combined specificity value for the identified term.

12. The method of claim 11, wherein the tag parameter of an identified term is associated with a tag type of a tag associated with the identified term.

13. The method of claim 12, wherein the tag parameter is associated with a frequency of appearance in the intranet of the tag type.

16. The method of claim 11, further comprising, for each of the plurality of pages and for each identified term, normalizing the combined specificity value for the identified term by the length of the page.

17. The method of claim 11, wherein combining the computed specificity values of the instances of the identified term within the page comprises summing the computed specificity related of the instances of the identified term within the page.

18. The method of claim 11, wherein combining the computed specificity values of the instances of the identified term within the page comprises averaging the computed specificity values of the instances of the identified term within the page.

14. A computer storage medium encoded with a program for execution on at least one processor, the program, when executed on the at least one processor, performing a method for creating data for use in ranking results of a search of a plurality of pages in an intranet, each page comprising a plurality of terms arranged in a hierarchical structure, the method comprising:

(A) the act of parsing the plurality of pages to identify the plurality of terms, wherein at least one page of the plurality of pages comprises a plurality of instances of a term of the plurality of terms;

(B) the act of computing, for each instance of each identified term within the plurality of pages, a specificity value of the instance based on a combination of a tag parameter associated with the instance and a term level of the instance within the hierarchical structure of the page of the instance, the term level being measured from a root tag of the page of the instance;

(C) the act of combining, for each identified term and for each page in the plurality of pages in which the identified term appears, the computed specificity value of each instance of the identified term within the page to produce a combined computed specificity value of the identified term; and

(D) the act of storing information, for each identified term and for each page in the plurality of pages in which the identified term appears, the combined computed specificity value of the identified term and an identifier for the page.

15. The computer storage medium of claim 14, wherein the hierarchical structure comprises a plurality of levels comprising at least one root level, and wherein the act (B) further comprises determining a number of levels in a shortest path from a level of the at least one root level to the term level of the instance of the identified term.