Data processing: database and file management or data structures – Database design – Data structure types
Reexamination Certificate
2001-07-06
2004-09-28
Le, Uyen (Department: 2171)
Data processing: database and file management or data structures
Database design
Data structure types
C715S252000
Reexamination Certificate
active
06799176
ABSTRACT:
FIELD OF THE INVENTION
This invention relates generally to techniques for analyzing linked databases. More particularly, it relates to methods for assigning ranks to nodes in a linked database, such as any database of documents containing citations, the world wide web or any other hypermedia database.
BACKGROUND OF THE INVENTION
Due to the developments in computer technology and its increase in popularity, large numbers of people have recently started to frequently search huge databases. For example, internet search engines are frequently used to search the entire world wide web. Currently, a popular search engine might execute over 30 million searches per day of the indexable part of the web, which has a size in excess of 500 Gigabytes. Information retrieval systems are traditionally judged by their precision and recall. What is often neglected, however, is the quality of the results produced by these search engines. Large databases of documents such as the web contain many low quality documents. As a result, searches typically return hundreds of irrelevant or unwanted documents which camouflage the few relevant ones. In order to improve the selectivity of the results, common techniques allow the user to constrain the scope of the search to a specified subset of the database, or to provide additional search terms. These techniques are most effective in cases where the database is homogeneous and already classified into subsets, or in cases where the user is searching for well known and specific information. In other cases, however, these techniques are often not effective because each constraint introduced by the user increases the chances that the desired information will be inadvertently eliminated from the search results.
Search engines presently use various techniques that attempt to present more relevant documents. Typically, documents are ranked according to variations of a standard vector space model. These variations could include (a) how recently the document was updated, and/or (b) how close the search terms are to the beginning of the document. Although this strategy provides search results that are better than with no ranking at all, the results still have relatively low quality. Moreover, when searching the highly competitive web, this measure of relevancy is vulnerable to “spamming” techniques that authors can use to artificially inflate their document's relevance in order to draw attention to it or its advertisements. For this reason search results often contain commercial appeals that should not be considered a match to the query. Although search engines are designed to avoid such ruses, poorly conceived mechanisms can result in disappointing failures to retrieve desired information.
Hyperlink Search Engine, developed by IDD Information Services, (http://rankdex.gari.com/) uses backlink information (i.e., information from pages that contain links to the current page) to assist in identifying relevant web documents. Rather than using the content of a document to determine relevance, the technique uses the anchor text of links to the document to characterize the relevance of a document. The idea of associating anchor text with the page the text points to was first implemented in the World Wide Web Worm (Oliver A. McBryan, GENVL and WWWW: Tools for Taming the Web, First International Conference on the World Wide Web, CERN, Geneva, May 25-27, 1994). The Hyperlink Search Engine has applied this idea to assist in determining document relevance in a search. In particular, search query terms are compared to a collection of anchor text descriptions that point to the page, rather than to a keyword index of the page content. A rank is then assigned to a document based on the degree to which the search terms match the anchor descriptions in its backlink documents.
The well known idea of citation counting is a simple method for determining the importance of a document by counting its number of citations, or backlinks. The citation rank r(A) of a document which has n backlink pages is simply
r
(
A
)=
n.
In the case of databases whose content is of relatively uniform quality and importance it is valid to assume that a highly cited document should be of greater interest than a document with only one or two citations. Many databases, however, have extreme variations in the quality and importance of documents. In these cases, citation ranking is overly simplistic. For example, citation ranking will give the same rank to a document that is cited once on an obscure page as to a similar document that is cited once on a well-known and highly respected page.
SUMMARY OF THE INVENTION
Various aspects of the present invention provide systems and methods for ranking documents in a linked database. One aspect provides an objective ranking based on the relationship between documents. Another aspect of the invention is directed to a technique for ranking documents within a database whose content has a large variation in quality and importance. Another aspect of the present invention is to provide a document ranking method that is scalable and can be applied to extremely large databases such as the world wide web. Additional aspects of the invention will become apparent in view of the following description and associated figures.
One aspect of the present invention is directed to taking advantage of the linked structure of a database to assign a rank to each document in the database, where the document rank is a measure of the importance of a document. Rather than determining relevance only from the intrinsic content of a document, or from the anchor text of backlinks to the document, a method consistent with the invention determines importance from the extrinsic relationships between documents. Intuitively, a document should be important (regardless of its content) if it is highly cited by other documents. Not all citations, however, are necessarily of equal significance. A citation from an important document is more important than a citation from a relatively unimportant document. Thus, the importance of a page, and hence the rank assigned to it, should depend not just on the number of citations it has, but on the importance of the citing documents as well. This implies a recursive definition of rank: the rank of a document is a function of the ranks of the documents which cite it. The ranks of documents may be calculated by an iterative procedure on a linked database.
Because citations, or links, are ways of directing attention, the important documents correspond to those documents to which the most attention is directed. Thus, a high rank indicates that a document is considered valuable by many people or by important people. Most likely, these are the pages to which someone performing a search would like to direct his or her attention. Looked at another way, the importance of a page is directly related to the steady-state probability that a random web surfer ends up at the page after following a large number of links. Because there is a larger probability that a surfer will end up at an important page than at an unimportant page, this method of ranking pages assigns higher ranks to the more important pages.
In one aspect of the invention, a computer implemented method is provided for scoring linked documents. The method includes identifying links from linking documents to linked documents in the network and determining an importance of the identified links. The method further includes weighting the identified links based on the determined importance and scoring the linked documents based on the weighted links.
In accordance with another implementation consistent with the present invention, a method for scoring documents, where at least some of the documents contain links to other ones of the documents, includes determining a probability that a searcher will access each of the documents after following a number of the links; and scoring each of the documents based on the determined probability.
In accordance with yet another implementation consistent with the present inventi
Harrity & Snyder LLP
Le Uyen
The Board of Trustees of the Leland Stanford Junior University
LandOfFree
Method for scoring documents in a linked database does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method for scoring documents in a linked database, we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method for scoring documents in a linked database will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-3201340