Methods and systems for displaying web pages based on a user-specific browser history analysis

• US 9,310,879 B2
• Filed: 11/09/2011
• Issued: 04/12/2016
• Est. Priority Date: 11/09/2011
• Status: Expired due to Fees

• Current Assignee: Xerox Corporation (Xerox Holdings Corp.)
• Original Assignee: Xerox Corporation (Xerox Holdings Corp.)
• Inventors: St. Jacques, Robert J. Jr.
• Primary Examiner(s): Baderman, Scott
• Assistant Examiner(s): Cortes, Howard
• Application Number:
  US13/292,302
• Publication Number:
  US 20130117677A1
• Time in Patent Office: 1,616 Days
• Field of Search: 715/738
• US Class Current: 1/1
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation
• CPC Class Codes: G06F 3/00 Input arrangements for transferring data to be processed into a form capable of being handled by the computer; Output arrangements for transferring data from processing unit to output unit, e.g. interface arrangements
• CPC Class Codes: G06F 40/143 Markup, e.g. Standard Generalized Markup Language [SGML] or Document Type Definition [DTD]

Abstract

Methods and systems for automatically parsing the content of a user'"'"'s browser history to extract information about that user'"'"'s browsing habits, interests, likes, and dislikes utilizing text analytics and unsupervised machine learning. A topic model can be utilized to filter the display of web content and employ visualization techniques to highlight specific content (e.g. keywords) that correspond with the users specific tastes. The analysis can be as coarse or fine as the user desires and all of the analysis can be done locally on the user'"'"'s own data processing device.

Claims

1. A method for displaying user specific content, said method comprising:

determining a size of a random sample of a plurality of links in a browser history;

selecting a predetermined number of links from said plurality of links in said browser history for said random sample of said plurality of links;

fetching content for each of said plurality of links in said random sample;

extracting a plurality of topics comprising a variable number of topics from said content including text from HTML, said extracting of said plurality of topics executed in a background as a low priority process;

generating a topic model from said plurality of topics extracted from said content, said topic model to be used to alter a manner in which selected web content is displayed;

evaluating content of a newly selected web page against said topic model after generating said topic model to determine a probability score for each of said plurality of topics based on how accurately each of said plurality of topics describes said content of the newly selected web page and by providing every topic in said topic model with said probability score within a certain threshold of all other topics;

selecting a subset of said plurality of topics based on said probability score;

evaluating said content from said newly selected web page for keywords associated with said subset of topics and which represent significant trends across said subset of tonics; and

highlighting said keywords within said newly selected web page based on said topic model in order to display user specific content.

2. The method of claim 1, wherein each of said plurality of topics comprises plurality of keywords that describe said content.

3. The method, of claim 2 further comprising extracting said plurality of topics utilizing Latent Dirichlet Allocation (LDA).

4. The method of claim 3 wherein highlighting specific words within a web page based on said topic model, further comprises:

extracting content of said newly selected web page;

determining a probability score for each of said plurality of topics in said topic model for how well said each of said plurality of topics describes said content;

selecting a set of said plurality of topics that meet a predefined criteria

analyzing said content utilizing said set of topics;

determining a weight for each of a plurality of keywords associated with said set of topics, wherein different weights are determined for different keywords among said plurality of keywords depending how significant said each of said plurality of keywords are determined to be; and

differentially highlighting said plurality of keywords associated with said set of topics within said content of said newly selected web page according to said weight.

5. The method of claim 4 further comprising limiting said browsing history to history from a particular browsing session and to history from a specific tab within a particular browsing session.

6. The method of claim 4 evaluating said content from said newly selected web page for keywords associated with said subset of topics and which represent significant trends across said subset of topics, further comprises utilizing natural language processing to map different forms of said keywords and synonyms to specific stem-words within said topic model.

7. A computer-readable non-transitory storage media comprising instructions operable for displaying user specific content to cause a computer to:

determine a size of a random sample of a plurality of links in a browser history;

select a predetermined number of links from said plurality of links in said browser history for said random sample of said plurality of links;

fetch content for each of said plurality of links in said random sample;

extract a plurality of topics comprising a variable number of topics from said content including text from HTML, said extracting of said plurality of topics executed in a background as a low priority process;

generate a topic model from said plurality of topics extracted from said content, said topic model to be used to alter a manner in which selected web content is displayed;

evaluate content of a newly selected web page against said topic model after generating said topic model to determine a probability score for each of said plurality of topics based on how accurately each of said plurality of topics describes said content of the newly selected web page and by providing every topic in said topic model with said probability score within a certain threshold of all other topics;

select a subset of said plurality of topics based on said probability score;

evaluate said content from said newly selected web page for keywords associated with said subset of topics and which represent significant trends across said subset of topics; and

highlight said keywords within said newly selected web page based on said topic model in order to display user specific content.

8. The computer-readable non-transitory storage media of claim 7 wherein each of said plurality of topics comprises a plurality of keywords that describe said content.

9. The computer-readable non transitory storage media of claim 8 wherein said plurality of topics is extracted utilizing Latent Dirichlet Allocation (LDA).

10. The computer-readable non-transitory storage media of claim 9 wherein said highlight specific words within said web page based on said topic model, further comprises:

extracting content of said newly selected web page;

determining a probability score for each of said plurality of topics in said topic model for how well said each of said plurality of topics describes said content;

selecting a set of said plurality of topics that meet a predefined criteria

analyzing said content utilizing said set of topics;

determining a weight for each of a plurality of keywords associated with said set of topics, wherein different weights are determined for different keywords among said plurality of keywords depending how significant said each of said plurality of keywords are determined to be; and

differentially highlighting said plurality of keywords associated with said set of topics within said content of said newly selected web page according to said weight.

11. The computer-readable non-transitory storage media of claim 10 wherein said browsing history is limited to history from a particular browsing session and from a specific tab within said particular browsing session.

12. The computer-readable non-transitory storage media of claim 10 wherein said evaluate said content from said newly selected web page for keywords associated with said subset of topics and which represent significant trends across said subset of topics, further comprises utilizing natural language processing to map different forms of said keywords and synonyms to specific stem-words within said topic model.

13. A system for displaying user specific content, said system comprising:

a processor; and

a non-transitory computer-usable medium embodying computer program code, said non-transitory computer-usable medium capable of communicating with the processor, said computer program code comprising instructions executable by said processor and configured for;

determining a size of a random sample of a plurality of links in a browser history;

selecting a predetermined number of links from said plurality of links in said browser history for said random sample of said plurality of links;

fetching content for each of said plurality of links in said random sample;

extracting a plurality of topics comprising a variable number of topics from said content including text from HTML, said extracting of said plurality of topics executed in a background as a low priority process;

generating a topic model from said plurality of topics extracted from said content, said topic model to be used to alter a manner in which selected web content is displayed;

evaluating content of a newly selected web page against said topic model after generating said topic model to determine a probability score for each of said plurality of topics based on how accurately each of said plurality of topics describes said content of the newly selected web page and by providing every topic in said topic model with said probability score within a certain threshold of all other topics;

selecting a subset of said plurality of topics based on said probability score;

evaluating said content from said newly selected web page for keywords associated with said subset of topics and which represent significant trends across said subset of topics; and

highlighting said keywords within said newly selected web page based on said topic model in order to display user specific content.

14. The system of claim 13 wherein each of said plurality of topics comprises a plurality of keywords that describe said content.

15. The system of claim 14 wherein said instructions are further configured for extracting said plurality of topics utilizing Latent Dirichlet Allocation (LDA).

16. The system of claim 15 wherein said instructions for highlighting specific words within a web page based on said topic model, further comprise instructions further configured for:

extracting content of said newly selected web page;

determining a probability score for each of said plurality of topics in said topic model for how well each of said plurality of topics describes said content;

selecting a set of said plurality of topics that meet a predefined criteria;

analyzing said content utilizing said set of topics;

determining a weight for each of a plurality of keywords associated with said set of topics, wherein different weights are determined for different keywords among said plurality of keywords depending how significant said each of said plurality of keywords are determined to be; and

differentially highlighting said plurality of keywords associated with said set of topics within said content of said newly selected web page according to said weight.

17. The system of claim 16 wherein said instructions are further configured for limiting said browsing history to history from a particular browsing session and from a specific tab within a particular browsing session.

18. The system of claim 16 wherein said instructions are further configured for utilizing natural language processing to map different forms of said keywords and synonyms to specific stem-words within said topic model.