Method for ranking web page search results
AssigneeOverture Services, Inc.
InventorsAndrei Z. Broder, Farzin Maghoul
Life of this patent
Estimating the term…
Abstract
A system for indexing a document including: receiving a document to be processed for inclusion in an index of documents; locating a set of documents that include hyperlinks to the document; retrieving anchortext associated with each hyperlink; parsing the anchortext into one or more tokens. Then for each token the following acts are performed: determining a weight for the token, determining whether the weight assigned to the token exceeds a threshold token weight; and indexing the document under the token, if the token weight assigned to the token exceeds the threshold token weight.
Claims14
A method for including a document in an index in a hyperlinked environment, comprising the acts of:
- receiving a document to be processed;
- locating a set of documents that include hyperlinks to the document;
- retrieving anchortext associated with at least one of the hyperlinks;
- parsing the anchortext into one or more tokens;
- for each token:
- determining a weight for the token,
- determining whether the weight assigned to the token exceeds a threshold token weight;
- and
- indexing the document under the token, if the token weight assigned to the token exceeds the threshold token weight, wherein the indexing act includes assigning to the token a location within the index corresponding to part of the page being indexed that is allocated for tokens having a higher degree of importance than other tokens in the same page.
The method of 1, wherein the act of determining a weight comprises:
- determining a first frequency at which the anchortext appears in the index;
- determining a second frequency at which each token derived from the anchortext appears in the index;
- and assigning a weight to the token, wherein the weight is a function of the first and second frequencies.
The method of 5 further comprising determining a normalized weight for each token.
A program product embedded in a computerized machine-readable medium for including a document in an index in a hyperlinked environment, comprising the instructions for:
- receiving a document to be processed;
- locating a set of documents that include hyperlinks to the document;
- retrieving anchortext associated with each hyperlink;
- parsing the anchortext into one or more tokens;
- and program instructions for each token comprising instructions for:
- determining a weight for the token,
- determining whether the weight assigned to the token exceeds a threshold token weight;
- and
- indexing the document under the token, if the token weight assigned to the token exceeds the threshold token weight, wherein the indexing includes assigning to the token a location within the index corresponding to part of the page being indexed that is allocated for tokens having a higher degree of importance than other tokens in the same page.
A computer system for indexing a document in a hyperlinked environment, comprising:
- a receiver for receiving a document to be processed for inclusion in an index of documents;
- a module for locating a set of documents that include hyperlinks to the document;
- a module for retrieving anchortext associated with each hyperlink;
- a parsing module for parsing the anchortext into one or more tokens;
- a module for:
- determining a weight for the token,
- determining whether the weight assigned to the token exceeds a threshold token weight;
- and
- indexing the document under the token, if the token weight assigned to the token exceeds the threshold token weight, wherein indexing includes assigning to the token a location within the index corresponding to part of the page being indexed that is allocated for tokens having a higher degree of importance than other tokens in the same page.
Dependencies are inferred from the claim text.