Data processing: database and file management or data structures – Database design – Data structure types
Reexamination Certificate
1998-06-23
2001-07-03
Black, Thomas (Department: 2171)
Data processing: database and file management or data structures
Database design
Data structure types
C707S793000, C707S793000, C707S793000
Reexamination Certificate
active
06256627
ABSTRACT:
BACKGROUND OF THE INVENTION
The present invention is directed to a method and system for maintaining a knowledge base and evidence set.
The field of search engines is known. Known search engines include those developed by Verity, Inc., Alta Vista, and Lycos. By implementing a search engine, a user can express with precision a focussed area of interest in order to retrieve needed information. Typically, a search engine retrieves documents satisfying the exact terms in a search query. For example, if the search query includes the term “PDA,” the search will not retrieve occurrences of “personal digital assistant,” “pocket device,” or other related terms. This produces under-inclusive results, meaning that documents containing relevant information are not retrieved. Often, however, it is difficult for a user to formulate a query capable of producing appropriately-inclusive results without existing knowledge of a subject area. This difficulty is especially prevalent when a lay user searches in subject areas containing technical terminology or jargon, which is unfamiliar to the lay user. For instance, when searching in the subject area of medical terminology, the lay user is more likely to employ everyday names for terms rather than the technical terms used by medical professionals. Even medical professionals may have difficulty in correctly spelling or recalling a proper medical term. Under-inclusive results also occur when relatively inexperienced users attempt to use search engines. For example, inexperienced users may fail to appreciate that certain search engines are case sensitive or require specific syntax.
Three approaches have been adopted to address under-inclusive results. The first approach employs manual query expansion. As noted above, if a search query is “PDA,” the search will not retrieve occurrences of “personal digital assistant,” “pocket device,” or other related terms. Users familiar with these related terms may manually expand the query by substituting “PDA” in the search query with “‘PDA’ OR ‘personal digital assistant’ OR ‘pocket device’”. This query uses the logical OR operator and would retrieve those documents containing at least one of these terms. Manual query expansion, however, requires user knowledge of related terms. In addition, manual query expansion requires excessive user input. For instance, if a user manually expands the same query term and wished to repeatedly conduct the search, the user must reenter the same related terms each time the query is submitted. Finally, users must have working knowledge of the search engine syntax and the vocabulary of the subject matter that is being searched.
The second approach to address under-inclusive results employs meta tagging. To implement meta tagging, the author of a document inserts metadata, also known as metainformation, into the contents of document itself or otherwise associates it with the document. Metadata is data that describes other data. For example, an author of a web page on the Internet's World Wide Web may insert meta tags into the source code of the web page. Typically, the meta tag is invisible to those viewing the web page with a traditional browser, such as Netscape Navigator, but is present in the source code and visible to search engines. Meta tags are usually words and phrases, which are related to the content of the web page, but do not exist in the text of the web page visible to the user. For example, when a search engine searches for “PDA” on the World Wide Web, the search engine retrieves documents containing “PDA,” if “PDA” is either in the meta tag or the contents of the document. One disadvantage to meta tagging, however, is the time investment required by authors to insert meta tags in each document. Moreover, once a document is created, it is time-consuming to modify the meta tags; each document must be reopened to edit the meta tags. Also, since meta tag information is inserted into each document there is an increased likelihood of a data entry error in the spelling or format of the meta tag information. In addition, the meta tag vocabulary might change, thus requiring a modification to all documents containing the meta tag information. Yet another disadvantage is that an individual inserting the meta tags must be taught the vocabulary relating to the content of the document. Finally, meta tagging requires knowledge of the content of the document. In many instances the author of a web page is a web page developer, who is developing the web page for others who are familiar with the content. Thus, meta tagging often requires coordination between a web page developer and those familiar with the content of the web page.
The third approach to address under-inclusive results employs evidence sets. An evidence set contains evidence, which constitute phrases or terms. The evidence is organized into topics. This knowledge is organized, typically in a hierarchical structure or taxonomy, and made available as a shared resource to users. An evidence set is employed by an application, such as a search engine, by incorporating knowledge about topics and associated phrases. One company, Sageware, Inc., has developed a number of Knowledge Sets, which are functionally similar to evidence sets, for specific subject areas. See SAGEWARE, INC.,
Our Products: Sageware KnowledgeSets
(accessed on Mar. 21, 1998; copyright 1997). One use of evidence sets is for query expansion. In contrast to manual query expansion, query expansion with evidence sets does not require a manual substitution of related terms for each query. Rather, the search engine may access the contents of the evidence set to automatically expand the search query.
Known methods for creating evidence sets require extensive user input. Other methods for learning evidence sets exist, however, it is known that evidence sets generated with learning algorithms on training data typically produce inferior quality evidence sets. In addition, known methods for creating evidence sets often produce evidence sets that are difficult to modify. Typically, methods for creating evidence sets include the use of either a standard text editor or a graphical user interface (GUI). An evidence set may be created with a text editor by inputting text and symbols in accordance with a known evidence set format. As evidence sets generally require a specific syntax, text editor creation has the disadvantage that minor inadvertent input errors may create an improperly formatted or non-working evidence set. For instance, a misplaced symbol or term may inadvertently change the relationship between evidences or topics in an evidence set. Because the syntax of evidence sets is often cumbersome, a user cannot readily apprehend when mistakes have occurred. Moreover, once an evidence set has been created with a text editor, can be relatively difficult to modify its structure. A text-edited modification requires reentry of evidences in the evidence set to comport with the newly-modified structure. Also, modifying an evidence set with a text editor requires a user with working knowledge of the syntax of the evidence set. In addition, a user may create an inconsistent evidence set. For instance, a user may create a text-edited evidence set with multiple occurrences of the same topic. Moreover, using a text editor to create an evidence set, each topic may have a different set of evidences. This could create an internal inconsistency in the evidence set and result in an evidence set that is non-functioning or, at the very least, capable of producing inconsistent results. Finally, when making changes to a text-edited evidence set, a regression test must often be performed to fully understand the impact of changes to the evidence set.
A second known method for creating evidence sets employs GUIs. Such a method, developed by Verity, Inc., is topicEditor. VERITY, INC.,
Introduction to Topics Guide V
2.0 (copyrighted Sep. 23, 1996; visited Mar. 21, 1998) discloses the use of topicEditor. In topicEditor users create topics and evidences in a hierarchical GUI environ
Beattie Thomas W.
Manning Harley
McGuinness Deborah L.
Resnick Lori Alperin
AT&T Corp.
Black Thomas
Kenyon & Kenyon
Mizrahi Diane D.
LandOfFree
System and method for maintaining a knowledge base and... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with System and method for maintaining a knowledge base and..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and System and method for maintaining a knowledge base and... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2512851