Method to compute the prominence score to phone numbers on web pages and automatically annotate/attach it to ads

Abstract

Systems and methods for automatically extracting a plurality of contact information from a resource, calculating prominence scores of each contact information, and associating a selected contact information with a content item are provided. A content item and a uniform resource locator are received from a content provider. A resource identified by the uniform resource locator is loaded. A plurality of contact information is detected from the loaded resource. For each of the detected contact information, a prominence score is calculated. One of the plurality of contact information is selected based on the calculated prominence scores. The selected contact information is associated with the content item.

What is claimed is: 
     
         1 . A computer-implemented method for extracting contact information from a resource and associating it with a content item, comprising:
 receiving a content item and uniform resource locator (URL) from a content provider, the URL identifying a resource;   loading the resource identified by the received URL;   detecting a plurality of contact information from the loaded resource;   calculating a prominence score for each contact information of the plurality of contact information;   selecting one of the plurality of contact information based on the calculated prominence scores; and   associating the selected contact information with the content item.   
     
     
         2 . The method of  claim 1 , further comprising:
 determining the domain of the received URL;   finding a plurality of URLs under the domain, each plurality of URL identifying a resource that is less than a predetermined number of arcs from the received URL; and   wherein loading the resource comprises loading a plurality of resources identified by the plurality of URLs.   
     
     
         3 . The method of  claim 1 , wherein loading the resource comprises:
 parsing the resource into an object tree;   rendering the resource using the object tree; and wherein detecting a plurality of contact information further comprises detecting a plurality of contact information using optical character recognition from the rendered resource.   
     
     
         4 . The method of  claim 1 , further comprising:
 filtering a portion of the loaded resource; and   wherein detecting a plurality of contact information comprises detecting a plurality of contact information from an unfiltered portion of the loaded resource.   
     
     
         5 . The method of  claim 1 , wherein calculating the prominence score comprises
 calculating a prominence score for each of the plurality of contact information, a prominence score comprising a weighted sum of a plurality of signals associated with a respective contact information.   
     
     
         6 . The method of  claim 5 , wherein a signal is computed using one of:
 a number of times the respective contact information is shown on the loaded resource;   a number of other contact information detected on the loaded resource;   a typographic emphasis of a respective contact information;   a location of the contact information within the loaded resource;   a text located near the respective contact information within the loaded resource; or   a prominence of an image associated with the respective contact information referenced by the loaded resource.   
     
     
         7 . The method of  claim 1 , wherein selecting one of the plurality of contact information further comprises culling contact information below a predetermine minimum threshold value. 
     
     
         8 . The method of  claim 1 , wherein selecting one of the plurality of contact information comprises:
 electing a subset of contact information from the plurality of contact information; and   receiving, from the content provider, a selected contact information out of the subset of contact information.   
     
     
         9 . The method of  claim 1 , wherein selecting one of the plurality of contact information comprises:
 receiving, from the content provider, that the selected contact information is accurate responsive to the respective prominence score of the selected contact information falling below a confidence threshold value.   
     
     
         10 . The method of  claim 1 , wherein the contact information comprises one of a phone number, mailing address, email address, fax number, instant messaging handle, social media profile, payment information, or bank routing number. 
     
     
         11 . A system for extracting contact information from a resource and associating it with a content item, comprising:
 a receiver, receiving a content item and uniform resource locator (URL) from a content provider, the URL identifying a resource;   a loader, loading the resource identified by the received URL;   a detector, detecting a plurality of contact information from the loaded resource;   a scorer, calculating a prominence score for each of the plurality of contact information;   a selector, selecting one of the plurality of contact information based on the calculated prominence scores; and   a linker, associating the selected contact information with the content item.   
     
     
         12 . The system of  claim 11 , wherein the loader further comprises:
 determining the domain of the received URL;   finding a plurality of URLs under the domain, each plurality of URL identifying a resource that is less than a predetermined number of arcs from the received URL; and   loading a plurality of resources identified by the plurality of URLs.   
     
     
         13 . The system of  claim 11 , wherein the loader further comprises:
 parsing the resource into an object tree;   rendering the resource using the object tree; and   wherein the detector further comprises detecting a plurality of contact information using optical character recognition from the rendered resource.   
     
     
         14 . The system of  claim 11 , wherein the loader further comprises:
 filtering a portion of the loaded resource; and   wherein the detector further comprises detecting a plurality of contact information from an unfiltered portion of the loaded resource.   
     
     
         15 . The system of  claim 11 , wherein the scorer further comprises calculating a prominence score for each of the plurality of contact information, a prominence score comprising a weighted sum of a plurality of signals associated with a respective contact information. 
     
     
         16 . The system of  claim 15 , wherein the scorer further comprises computing a signal using one of:
 a number of times the respective contact information is shown on the loaded resource;   a number of other contact information detected on the loaded resource;   a typographic emphasis of a respective contact information;   a location of the contact information within the loaded resource;   a text located near the respective contact information within the loaded resource; or   a prominence of an image associated with the respective contact information referenced by the loaded resource.   
     
     
         17 . The system of  claim 11 , wherein the selector further comprises culling contact information below a predetermine minimum threshold value. 
     
     
         18 . The system of  claim 11 , wherein the selector further comprises:
 electing a subset of contact information from the plurality of contact information; and   receiving, from the content provider, a selected contact information out of the subset of contact information.   
     
     
         19 . The system of  claim 11 , wherein the selector further comprises:
 receiving, from the content provider, that the selected contact information is accurate responsive to the respective prominence score of the selected contact information falling below a confidence threshold value.   
     
     
         20 . The system of  claim 11 , wherein the contact information comprises one of a phone number, mailing address, email address, fax number, instant messaging handle, social media profile, payment information, or bank routing number.