Intent-oriented internet browsing

Patent metadata

Field	Value
Publication number	US-11625449-B2
Application number	US-201916686811-A
Country	US
Kind code	B2
Filing date	Nov 18, 2019
Priority date	Nov 16, 2018
Publication date	Apr 11, 2023
Grant date	Apr 11, 2023

Abstract

A system that provides intent-oriented browsing powered by machine learning and crowdsourcing. The system allows users to enter their intents, which are then assigned to target pages via supervised learning models based on hyperlinks and contributions made by other users. The system has a prediction server that is programmed to receive hyperlinks from a website and return target hyperlinks based on known intent, a user interface for inputting user intent, and a browser programmed to connect to the intent repository and to the prediction server via a user script. The list of supported intents can grow over time based on correct page marks for intent-page mappings as well as via continuous training of machine learning models.

First claim

What is claimed is:

1. A system for providing users with intent based web page selection, comprising the steps of
an intent repository containing a plurality of known intents mapped to a plurality of target pages and a plurality of hyperlink hashes;
a website server supporting a website and programmed to transmit a plurality of hyperlinks representing pages of the website;
a prediction server programmed to receive the plurality of hyperlinks from the website server and to send a plurality of target hyperlinks that the prediction server has determined correspond to the plurality of known intents back to the website server; and
a browser having a user script programmed to connect to the intent repository and to the prediction server, wherein the user script is programmed to allow a user to enter a user selected intent, to display any of the plurality of target hyperlinks received from the prediction server that correspond to the user selected intent.

2. The system of claim 1 , wherein the user script is further programmed to allow the user to select one of the target hyperlinks.

3. The system of claim 2 , wherein the user script is further programmed to redirect the browser to the selected one of the hyperlinks.

4. The system of claim 3 , wherein the user script is further programmed to allow the user to confirm whether any displayed plurality of target hyperlinks actually correspond to the user selected intent.

5. The system of claim 4 , wherein the user script is further programmed to allow the user to manually select one of the hyperlinks.

6. The system of claim 5 , wherein the prediction server is programmed to determine whether a target page corresponds to a known user intent using a machine learning model trained with hyperlink features.

7. The system of claim 6 , wherein the hyperlink features are at least one feature selected from the group consisting of text, content, position, length, and url words.

8. The system of claim 7 , wherein the machine learning model comprises a Support Vector Machine classifier with linear kernel trained via distant supervision and manual labeling.

9. A method for providing users with intent based web page selection, comprising the steps of:
providing an intent repository containing a plurality of known intents mapped to a plurality of target pages and a plurality of hyperlink hashes;
providing a website server supporting a website and programmed to transmit a plurality of hyperlinks representing pages of the website;
providing a prediction server programmed to receive the plurality of hyperlinks from the website server and to send a plurality of target hyperlinks that the prediction server has determined correspond to the plurality of known intents back to the website server; and
providing a browser having a user script programmed to connect to the intent repository and to the prediction server;
allowing a user to enter a user selected intent with the user script; and
displaying any of the plurality of target hyperlinks received from the prediction server that correspond to the user selected intent with the user script.

10. The method of claim 9 , further comprising the step of allowing the user to select one of the target hyperlinks.

11. The method of claim 10 , further comprising the step of redirecting the browser to the selected one of the hyperlinks.

12. The method of claim 11 , further comprising the step of allowing the user to confirm whether any displayed plurality of target hyperlinks actually correspond to the user selected intent.

13. The method of claim 12 , further comprising the step of allowing the user to manually select one of the hyperlinks.

14. The method of claim 13 , wherein the prediction server is programmed to determine whether a target page corresponds to a known user intent using a machine learning model trained with hyperlink features.

15. The method of claim 14 , wherein the hyperlink features are at least one feature selected from the group consisting of text, content, position, length, and url words.

16. The method of claim 15 , wherein the machine learning model comprises a Support Vector Machine classifier with linear kernel trained via distant supervision and manual labeling.

Assignees

• Barbosa Nata
• Wang Yang
• Univ Syracuse

Inventors

• Barbosa Natá
• Wang Yang

Classifications

• G06F16/953

  Querying, e.g. by the use of web search engines · CPC title

• G06F16/9558 Primary

  Details of hyperlinks; Management of linked annotations · CPC title

• G06F16/957 Primary

  Browsing optimisation, e.g. caching or content distillation · CPC title

• G06N20/10

  using kernel methods, e.g. support vector machines [SVM] · CPC title

• G06N5/04

  Inference or reasoning models · CPC title