Analyzing uniform resource locators
AssigneeHewlett-Packard Development Company, L.P.
InventorGeorgia Koutrika
Life of this patent
Estimating the term…
Abstract
Methods for analyzing a Uniform Resource Locator (URL) and apparatus for performing such methods. The methods include parsing the URL into text segments and generating n-grams from the text segments. The methods further include generating annotations, each annotation corresponding to one of the n-grams and comprising a match value for its corresponding n-gram, a description of its match value, and a score. The methods still further include selecting a subset of the annotations.
Claims15
A method of analyzing a Uniform Resource Locator (URL), comprising:
- obtaining the URL from an Internet browser instantiated and displayed on a computer;
- and
- performing, by a processor:
- parsing the URL into text segments;
- generating n-grams from the text segments of the URL;
- comparing the n-grams generated from the text segments of the URL to at least one knowledge base;
- generating annotations, each annotation corresponding to one of the n-grams generated from the text segments of the URL and comprising a match value for its corresponding n-gram determined from the comparison to the at least one knowledge base, a description of its match value from the at least one knowledge base, and a score indicating a relative confidence level that an association between the corresponding n-gram and the description of its match value is correct;
- and
- selecting a subset of the annotations.
A method of analyzing a Uniform Resource Locator (URL), comprising:
- obtaining the URL from an Internet browser instantiated and displayed on a computer:
- and performing, by a processor:
- parsing the URL into text segments;
- generating n-grams from the text segments;
- generating annotations, each annotation corresponding to one of the n-grams and comprising a match value for its corresponding n-gram, a description of its match value, and a score indicating a relative confidence level that an association between the corresponding n-gram and the description of its match value is correct;
- and
- selecting a subset of the annotations;
- wherein selecting a subset of the annotations comprises a selection policy selected from a group consisting of selecting a particular number of the annotations having the highest scores of all the annotations, selecting a particular percentage of the annotations having the highest scores of all the annotations, selecting those annotations having scores above a particular threshold value, and selecting annotations in response to which descriptions occur most frequently among the annotations.
Dependencies are inferred from the claim text.