Keyword set and target audience profile generalization techniques

Abstract

A variety of techniques are described by which keyword sets and target audience profiles may be generalized in a systematic and effective way with reference to relationships between keywords, profiles, and the data of an underlying user population.

1 . A computer-implemented method for generalizing either or both of an initial keyword set and an initial target audience profile, comprising:
 receiving the initial keyword set and the initial target audience profile as input, the initial keyword set comprising a plurality of initial keywords, the initial target audience profile comprising an initial profile parameter value for each of a plurality of profile parameters;   identifying a first population comprising a plurality of users each corresponding to the initial target audience profile;   retrieving user data representing the first population of users and including both demographic data and online behavioral data for each of the users;   processing the initial keyword set and the initial target audience profile with reference to the user data including constructing a bipartite graph representing relationships among the initial keyword set, the initial target audience profile, and the user data, and identifying either or both of additional keywords and additional users by performing propagations within the bipartite graph; and   generating either or both of a generalized keyword set and a generalized target audience profile, wherein the generalized keyword set represents a modification of the initial keyword set including at least some of the initial keywords and one or more of the additional keywords, and wherein the generalized target audience profile represents a modification of the initial target audience profile representing a second population of users including at least some of the first population of users and one or more of the additional users.   
     
     
         2 . The method of  claim 1  wherein processing the initial keyword set and the initial target audience profile with reference to the user data further includes, in combination with constructing the bipartite graph, one or more of: (1) constructing a directed graph representing at least some of the relationships and employing a graph subset expansion technique, (2) constructing and using a probabilistic model representing at least some of the relationships, (3) generating a plurality of itemsets, each itemset grouping the initial keywords and additional keywords derived from the user data and using a data mining co-occurrence technique on the itemsets, (4) constructing a matrix relating each of the users to selected ones of the initial keywords and additional keywords derived from the user data and applying at least one similarity measure to the matrix, (5) forming keyword clusters and user clusters, each keyword cluster including selected ones of the initial keywords and additional keywords derived from the user data, each user cluster including first users corresponding to the initial target audience profile and second users having corresponding user profiles including at least some of the profile parameters of the initial target audience profile and at least one additional profile parameter not included in the target audience profile, and evaluating the keyword and user clusters, (6) identifying selected ones of additional keywords derived from the user data which are similar to at least one of the initial keywords using a similarity measure, or (7) constructing a keyword space with reference to search results corresponding to at least some of the initial keywords, mapping additional keywords derived from the user data to the keyword space, and identifying selected ones of the additional keywords for inclusion in the generalized keyword set with reference to the mapping and a similarity measure. 
     
     
         3 . The method of  claim 1  further comprising performing an additional iteration of the processing and generating using either or both of the generalized keyword set and the generalized target audience profile as input. 
     
     
         4 . The method of  claim 3  further comprising, before performing the additional iteration, employing either or both of the generalized keyword set and the generalized target audience profile to conduct an advertising campaign, and collecting statistics relating to effectiveness of the advertising campaign, wherein the additional iteration is performed with reference to the statistics. 
     
     
         5 . The method of  claim 3  further comprising determining whether to perform the additional iteration with reference to at least one constraint, wherein the at least one constraint comprises one or more of a budget, a predefined number of iterations, effectiveness of an advertising campaign using the generalized keyword set or the generalized target audience profile, relevance of the generalized keyword set or the generalized target audience profile to a market, a difference between the initial keyword set or the initial target audience profile and the corresponding one of the generalized keyword set or the generalized target audience profile. 
     
     
         6 . The method of  claim 1  further comprising employing the generalized keyword set or the generalized target audience profile to conduct an advertising campaign. 
     
     
         7 . The method of  claim 1  wherein generating either or both of the generalized keyword set and the generalized target audience profile comprises generating both of the generalized keyword set and the generalized target audience profile. 
     
     
         8 . A computer-implemented method for generalizing an initial keyword set, comprising:
 receiving the initial keyword set and an initial target audience profile as input, the initial keyword set comprising a plurality of initial keywords, the initial target audience profile comprising an initial profile parameter value for each of a plurality of profile parameters;   identifying a first population comprising a plurality of users each corresponding to the initial target audience profile;   retrieving user data representing the first population of users and including both demographic data and online behavioral data for each of the users;   processing the initial keyword set and the initial target audience profile with reference to the user data including constructing a directed graph representing relationships among the initial keyword set, the initial target audience profile, and the user data, each node in the directed graph representing one of the initial keywords or one of additional keywords derived from the user data, selected nodes in the directed graph being connected by edges which are generated with reference to the user data, and employing a graph subset expansion technique to identify selected ones of the additional keywords; and   generating a generalized keyword set that represents a modification of the initial keyword set including at least some of the initial keywords and one or more of the selected additional keywords.   
     
     
         9 . The method of  claim 8  wherein processing the initial keyword set and the initial target audience profile with reference to the user data further includes, in combination with constructing the directed graph, one or more of: (1) constructing a bipartite graph representing at least some of the relationships and performing propagations within the bipartite graph, (2) constructing and using a probabilistic model representing at least some of the relationships, (3) generating a plurality of itemsets, each itemset grouping the initial keywords and additional keywords derived from the user data and using a data mining co-occurrence technique on the itemsets, (4) constructing a matrix relating each of the users to selected ones of the initial keywords and additional keywords derived from the user data and applying at least one similarity measure to the matrix, (5) forming keyword clusters and user clusters, each keyword cluster including selected ones of the initial keywords and additional keywords derived from the user data, each user cluster including first users corresponding to the initial target audience profile and second users having corresponding user profiles including at least some of the profile parameters of the initial target audience profile and at least one additional profile parameter not included in the target audience profile, and evaluating the keyword and user clusters, (6) identifying selected ones of additional keywords derived from the user data which are similar to at least one of the initial keywords using a similarity measure, or (7) constructing a keyword space with reference to search results corresponding to at least some of the initial keywords, mapping additional keywords derived from the user data to the keyword space, and identifying selected ones of the additional keywords for inclusion in the generalized keyword set with reference to the mapping and a similarity measure. 
     
     
         10 . The method of  claim 8  further comprising performing an additional iteration of the processing and generating using the generalized keyword set as input. 
     
     
         11 . The method of  claim 10  further comprising, before performing the additional iteration, employing the generalized keyword set to conduct an advertising campaign, and collecting statistics relating to effectiveness of the advertising campaign, wherein the additional iteration is performed with reference to the statistics. 
     
     
         12 . The method of  claim 10  further comprising determining whether to perform the additional iteration with reference to at least one constraint, wherein the at least one constraint comprises one or more of a budget, a predefined number of iterations, effectiveness of an advertising campaign using the generalized keyword set, relevance of the generalized keyword set to a market, a difference between the initial keyword set and the generalized keyword set. 
     
     
         13 . The method of  claim 8  further comprising employing the generalized keyword set to conduct an advertising campaign. 
     
     
         14 . A computer-implemented method for generalizing either or both of an initial keyword set and an initial target audience profile, comprising:
 receiving the initial keyword set and the initial target audience profile as input, the initial keyword set comprising a plurality of initial keywords, the initial target audience profile comprising an initial profile parameter value for each of a plurality of profile parameters;   identifying a first population comprising a plurality of users each corresponding to the initial target audience profile;   retrieving user data representing the first population of users and including both demographic data and online behavioral data for each of the users;   processing the initial keyword set and the initial target audience profile with reference to the user data including constructing a probabilistic model representing relationships among the initial keyword set, the initial target audience profile, and the user data, and identifying either or both of additional keywords and additional users using the probabilistic model; and   generating either or both of a generalized keyword set and a generalized target audience profile, wherein the generalized keyword set represents a modification of the initial keyword set including at least some of the initial keywords and one or more of the additional keywords, and wherein the generalized target audience profile represents a modification of the initial target audience profile representing a second population of users including at least some of the first population of users and one or more of the additional users.   
     
     
         15 . The method of  claim 14  wherein processing the initial keyword set and the initial target audience profile with reference to the user data further includes, in combination with constructing the probabilistic model, one or more of: (1) constructing a bipartite graph representing at least some of the relationships and performing propagations within the bipartite graph, (2) constructing a directed graph representing at least some of the relationships and employing a graph subset expansion technique, (3) generating a plurality of itemsets, each itemset grouping the initial keywords and additional keywords derived from the user data and using a data mining co-occurrence technique on the itemsets, (4) constructing a matrix relating each of the users to selected ones of the initial keywords and additional keywords derived from the user data and applying at least one similarity measure to the matrix, (5) forming keyword clusters and user clusters, each keyword cluster including selected ones of the initial keywords and additional keywords derived from the user data, each user cluster including first users corresponding to the initial target audience profile and second users having corresponding user profiles including at least some of the profile parameters of the initial target audience profile and at least one additional profile parameter not included in the target audience profile, and evaluating the keyword and user clusters, (6) identifying selected ones of additional keywords derived from the user data which are similar to at least one of the initial keywords using a similarity measure, or (7) constructing a keyword space with reference to search results corresponding to at least some of the initial keywords, mapping additional keywords derived from the user data to the keyword space, and identifying selected ones of the additional keywords for inclusion in the generalized keyword set with reference to the mapping and a similarity measure. 
     
     
         16 . The method of  claim 14  further comprising performing an additional iteration of the processing and generating using either or both of the generalized keyword set and the generalized target audience profile as input. 
     
     
         17 . The method of  claim 16  further comprising, before performing the additional iteration, employing either or both of the generalized keyword set and the generalized target audience profile to conduct an advertising campaign, and collecting statistics relating to effectiveness of the advertising campaign, wherein the additional iteration is performed with reference to the statistics. 
     
     
         18 . The method of  claim 16  further comprising determining whether to perform the additional iteration with reference to at least one constraint, wherein the at least one constraint comprises one or more of a budget, a predefined number of iterations, effectiveness of an advertising campaign using the generalized keyword set or the generalized target audience profile, relevance of the generalized keyword set or the generalized target audience profile to a market, a difference between the initial keyword set or the initial target audience profile and the corresponding one of the generalized keyword set or the generalized target audience profile. 
     
     
         19 . The method of  claim 14  further comprising employing the generalized keyword set or the generalized target audience profile to conduct an advertising campaign. 
     
     
         20 . The method of  claim 14  wherein generating either or both of the generalized keyword set and the generalized target audience profile comprises generating both of the generalized keyword set and the generalized target audience profile.