Informative advertisements on hobby and strong interests feature space

Abstract

Information for an advertising campaign is received that includes one or more advertisements for presentation to one or more users of an online system. A strong interest feature domain is generated for each user of the online system. A set of related items is obtained, wherein each item in the set is associated with an action conducted by the one or more users. It can be determined that a viewing user of the online system has interacted with at least one item in the set of related items, and the generated interest feature domain for the viewing user includes an interest feature domain specified in the advertising campaign. Responsive to the determination, an advertisement is selected for presentation to the viewing user.

1 - 18 . (canceled) 
     
     
         19 . A system, comprising:
 a processor;   a memory storing instructions, which when executed by the processor, cause the processor to:
 generate, by a trained machine learning model, a set of values for interests associated with a viewing user, wherein each value represents a measure of the viewing user's affinity for other interests from a pool of different interests; 
 determine, based on the set of values and a history of interactions by a viewing user with content items of a set of content items, an order that the viewing user interacted with a first content item and a second content item of the set; 
 compare the determined order with one or more orders of interactions described by one or more graphs for sets of related content items comprising the first content item and the second content item; 
 match the determined order with one or more orders based on the comparison; 
 identify digital content associated with a campaign corresponding to the one or more graphs based on the matching; and 
 select the identified digital content for presenting to the viewing user. 
   
     
     
         20 . The system of  claim 19 , wherein generating the set of values for interests associated with the viewing user comprises:
 inferring one or more strong interests of the viewing user based on actions performed by the viewing user, wherein the actions performed by the viewing user comprises at least one of the following:
 purchasing an item associated with a set of related items; or 
 viewing or attending one or more events associated with the set of related items. 
   
     
     
         21 . The system of  claim 19 , wherein the trained machine learning model is continuously trained based on attribution data associated with the viewing user. 
     
     
         22 . The system of  claim 19 , further comprising:
 receiving a set of related items, wherein each item in the set of related items is associated with an action conducted by the viewing user, and wherein the set of related items comprises at least one of: a plurality of complementary items, or a plurality of related digital media items.   
     
     
         23 . The system of  claim 22 , further comprising:
 determining that the viewing user has interacted with at least one item in the set of related items;   receiving indication of time periods of the interaction of the viewing user with the at least one item in the set of related items; and   tracking the interaction of the viewing user with the at least one item in the set of related items during the identified time periods.   
     
     
         24 . The system of  claim 22 , further comprising:
 selecting the digital content for presentation to the viewing user based on determining that the digital content is related to an item that completes the set of related items.   
     
     
         25 . The system of  claim 22 , further comprising:
 receiving indication about an order of interaction of the viewing user with at least one item in the set of related items based on a cyclic graph of the related items; and   selecting the digital content for presentation to the viewing user based on the order of the interaction of the viewing user with the at least one item in the set of related items.   
     
     
         26 . A method, comprising:
 generating, by a trained machine learning model, a set of values for interests associated with a viewing user, wherein each value represents a measure of the viewing user's affinity for other interests from a pool of different interests;   determining, based on the set of values and a history of interactions by a viewing user with content items of a set of content items, an order that the viewing user interacted with a first content item and a second content item of the set;   comparing the determined order with one or more orders of interactions described by one or more graphs for sets of related content items comprising the first content item and the second content item;   matching the determined order with one or more orders based on the comparison;   identifying digital content associated with a campaign corresponding to the one or more graphs based on the matching; and   selecting the identified digital content for presenting to the viewing user.   
     
     
         27 . The method of  claim 26 , wherein generating the set of values for interests associated with the viewing user comprises:
 inferring one or more strong interests of the viewing user based on actions performed by the viewing user, wherein the actions performed by the viewing user comprises at least one of the following:
 purchasing an item associated with a set of related items; or 
 viewing or attending one or more events associated with the set of related items. 
   
     
     
         28 . The method of  claim 26 , wherein the trained machine learning model is continuously trained based on attribution data associated with the viewing user. 
     
     
         29 . The method of  claim 26 , further comprising:
 receiving a set of related items, wherein each item in the set of related items is associated with an action conducted by the viewing user, and wherein the set of related items comprises at least one of: a plurality of complementary items, or a plurality of related digital media items.   
     
     
         30 . The method of  claim 29 , further comprising:
 determining that the viewing user has interacted with at least one item in the set of related items;   receiving indication of time periods of the interaction of the viewing user with the at least one item in the set of related items; and   tracking the interaction of the viewing user with the at least one item in the set of related items during the identified time periods.   
     
     
         31 . The method of  claim 29 , further comprising:
 selecting the digital content for presentation to the viewing user based on determining that the digital content is related to an item that completes the set of related items.   
     
     
         32 . The method of  claim 29 , further comprising:
 receiving indication about an order of interaction of the viewing user with at least one item in the set of related items based on a cyclic graph of the related items; and   selecting the digital content for presentation to the viewing user based on the order of the interaction of the viewing user with the at least one item in the set of related items.   
     
     
         33 . A non-transitory computer-readable storage medium, which when executed, causes a processor to:
 generate, by a trained machine learning model, a set of values for interests associated with a viewing user, wherein each value represents a measure of the viewing user's affinity for other interests from a pool of different interests;   determine, based on the set of values and a history of interactions by a viewing user with content items of a set of content items, an order that the viewing user interacted with a first content item and a second content item of the set;   compare the determined order with one or more orders of interactions described by one or more graphs for sets of related content items comprising the first content item and the second content item;   match the determined order with one or more orders based on the comparison;   identify digital content associated with a campaign corresponding to the one or more graphs based on the matching; and   select the identified digital content for presenting to the viewing user.   
     
     
         34 . The non-transitory computer-readable storage medium of  claim 33 , wherein generating the set of values for interests associated with the viewing user comprises:
 inferring one or more strong interests of the viewing user based on actions performed by the viewing user, wherein the actions performed by the viewing user comprises at least one of the following:
 purchasing an item associated with a set of related items; or 
 viewing or attending one or more events associated with the set of related items. 
   
     
     
         35 . The non-transitory computer-readable storage medium of  claim 33 , wherein the trained machine learning model is continuously trained based on attribution data associated with the viewing user. 
     
     
         36 . The non-transitory computer-readable storage medium of  claim 33 , further comprising:
 receiving a set of related items, wherein each item in the set of related items is associated with an action conducted by the viewing user, and wherein the set of related items comprises at least one of: a plurality of complementary items, or a plurality of related digital media items.   
     
     
         37 . The non-transitory computer-readable storage medium of  claim 36 , further comprising:
 determining that the viewing user has interacted with at least one item in the set of related items;   receiving indication of time periods of the interaction of the viewing user with the at least one item in the set of related items; and   tracking the interaction of the viewing user with the at least one item in the set of related items during the identified time periods.   
     
     
         38 . The non-transitory computer-readable storage medium of  claim 36 , further comprising:
 selecting the digital content for presentation to the viewing user based on determining that the digital content is related to an item that completes the set of related items.