Automatic Item Placement Recommendations Based on Entity Similarity

Abstract

Automatic item placement recommendation is described. An item placement configuration system receives an item for which a recommended placement is to be generated and identifies an entity associated with the item. The item placement configuration system then identifies a multi-domain taxonomy that describes relationships between different entities based on items associated with the different entities published among different domains. A representation of the entity associated with the item to be placed is then identified within the multi-domain taxonomy, along with a representation of at least one similar entity. Upon identifying a similar entity, historic item placement metrics for the similar entity are leveraged to generate a placement recommendation for the received item. In some implementations, the placement recommendation is output with a visual indication of a similar entity and associated performance metrics that were considered in generating the recommended placement.

What is claimed is: 
     
         1 . A method comprising:
 receiving, by a processing device, a request for a placement recommendation for a digital content item;   ascertaining, by the processing device, an item type of the digital content item and an entity associated with the digital content item;   obtaining, by the processing device, a multi-domain taxonomy for the item type that describes how different entities relate to one another based on digital content items associated with the different entities;   identifying, by the processing device, a multi-dimensional vector that represents the entity within the multi-domain taxonomy;   identifying, by the processing device, at least one of the different entities as a similar entity by:
 generating a mapping of the multi-domain taxonomy by embedding each of the different entities as a point in two-dimensional or three-dimensional space using a t-distributed stochastic neighbor embedding method; and 
 embedding the multi-dimensional vector into the mapping of the multi-domain taxonomy using the t-distributed stochastic neighbor embedding method; and 
   outputting, by the processing device, the placement recommendation for the digital content item based on historic item placement configurations associated with the similar entity.   
     
     
         2 . The method of  claim 1 , wherein each dimension of the multi-dimensional vector corresponds to a grouping of digital content items and is represented by a value indicating a degree of relevance between the entity and the grouping of digital content items. 
     
     
         3 . The method of  claim 1 , wherein the outputting the placement recommendation for the digital content item is performed independent of user input specifying a cost threshold or an intended performance metric associated with placement of the digital content item. 
     
     
         4 . The method of  claim 1 , wherein the identifying the at least one of the different entities as the similar entity is performed by computing a cosine distance between the multi-dimensional vector for the entity and a multi-dimensional vector representing the similar entity. 
     
     
         5 . The method of  claim 1 , wherein the identifying the multi-dimensional vector that represents the entity in the multi-domain taxonomy comprises identifying a plurality of different digital content items associated with the entity and generating the multi-dimensional vector based on the digital content item and the plurality of different digital content items associated with the entity. 
     
     
         6 . The method of  claim 1 , wherein each of the historic item placement configurations have an associated performance metric and the placement recommendation is determined based on a historic item placement configuration having a performance metric that satisfies a performance metric threshold. 
     
     
         7 . The method of  claim 1 , wherein the placement recommendation identifies a particular user interface in which the digital content item should be placed. 
     
     
         8 . The method of  claim 7 , wherein the particular user interface is defined for output based on one or more of a geographic region, a demographic audience, a display time, or a language. 
     
     
         9 . The method of  claim 1 , wherein the placement recommendation identifies one or more domains in which the digital content item should be placed. 
     
     
         10 . The method of  claim 1 , wherein the placement recommendation identifies one or more domains in which the digital content item should be excluded from placement. 
     
     
         11 . The method of  claim 1 , wherein the placement recommendation specifies a particular portion of a user interface in which the digital content item is to be placed. 
     
     
         12 . The method of  claim 1 , further comprising generating a configured user interface that presents the digital content item according to the placement recommendation and outputting the configured user interface. 
     
     
         13 . The method of  claim 1 , wherein the outputting the placement recommendation is performed independent of information specifying an intended target audience for the digital content item. 
     
     
         14 . The method of  claim 1 , wherein the outputting the placement recommendation for the digital content item further comprises outputting a forecasted value of using the placement recommendation, the forecasted value comprising at least one of:
 a cost-per-impression;   an overall number of impressions;   a percentage confidence associated with the forecasted value; or   a degree of variability for the forecasted value.   
     
     
         15 . A non-transitory computer-readable medium storing executable instructions, which when executed by a processing device, cause the processing device to perform operations comprising:
 receiving a request for a placement recommendation for a digital content item;   ascertaining an item type of the digital content item and an entity associated with the digital content item;   obtaining a multi-domain taxonomy for the item type that describes how different entities relate to one another;   identifying a multi-dimensional vector that represents the entity within the multi-domain taxonomy;   identifying at least one of the different entities as a similar entity by:
 generating a mapping of the multi-domain taxonomy by embedding each of the different entities as a point in a multi-dimensional space using a t-distributed stochastic neighbor embedding method; and 
 embedding the multi-dimensional vector into the mapping of the multi-domain taxonomy using the t-distributed stochastic neighbor embedding method; and 
   outputting the placement recommendation for the digital content item based on historic item placement configurations associated with the similar entity.   
     
     
         16 . The non-transitory computer-readable medium of  claim 15 , wherein each dimension of the multi-dimensional vector corresponds to a grouping of digital content items and is represented by a value indicating a degree of relevance between the entity and the grouping of digital content items. 
     
     
         17 . The non-transitory computer-readable medium of  claim 15 , wherein the outputting the placement recommendation for the digital content item is performed independent of user input specifying a cost threshold or an intended performance metric associated with placement of the digital content item. 
     
     
         18 . A system comprising:
 a memory component; and   a processing device coupled to the memory component, the processing device to perform operations comprising:
 receiving a request for a placement recommendation for a digital content item; 
 determining an item type of the digital content item and an entity associated with the digital content item; 
 obtaining a multi-domain taxonomy for the item type that describes how different entities relate to one another based on digital content items associated with the different entities; 
 identifying a multi-dimensional vector that represents the entity within the multi-domain taxonomy; 
 identifying at least one of the different entities as a similar entity by:
 generating a mapping of the multi-domain taxonomy by embedding each of the different entities as a point in two-dimensional or three-dimensional space using a t-distributed stochastic neighbor embedding method; and 
 embedding the multi-dimensional vector into the mapping of the multi-domain taxonomy using the t-distributed stochastic neighbor embedding method; and 
 
 outputting the placement recommendation for the digital content item based on historic item placement configurations associated with the similar entity. 
   
     
     
         19 . The system of  claim 18 , wherein each dimension of the multi-dimensional vector corresponds to a grouping of digital content items and is represented by a value indicating a degree of relevance between the entity and the grouping of digital content items. 
     
     
         20 . The system of  claim 18 , wherein the outputting the placement recommendation for the digital content item is performed independent of user input specifying a cost threshold or an intended performance metric associated with placement of the digital content item.