Indexing semantic user profiles for targeted advertising
Embodiments facilitate greater flexibility in definition of user segments for targeted advertising, by employing indexed semantic user profiles. Semantic user profiles are built through extraction of online user behavior data such as user search queries and page views, and include user interest information that is inferred based on user behavior. Semantic user profiles are then indexed to facilitate search for a set of users that fit specified semantic search terms. Search results for semantic profiles are ranked according to a ranking model developed through machine learning. In some embodiments, building and indexing of semantic profiles and learning of the ranking model is performed offline to facilitate more efficient online processing of queries.
1. A computer-implemented method for facilitating online advertising, comprising:
generating a semantic user profile for each of a plurality of users, based at least on user behavior data collected for each user;
indexing the semantic user profiles;
receiving at least one semantic query;
performing a search over the indexed semantic user profiles based on the at least one semantic query, to enable selection of a target user segment from the plurality of users;
determining a set of the indexed semantic user profiles resulting from the search;
ranking the set of the indexed semantic user profiles based at least on an index of each of the indexed semantic user profiles of the set; and
providing the ranked set of the indexed semantic user profiles, to enable the selection of the target user segment.
2. The method of claim 1 , wherein the user behavior data includes at least one of search query data and page view data for each user.
3. The method of claim 1 , wherein the semantic user profile includes at least one of an interest domain, a user intent, and a user preference.
4. The method of claim 1 , wherein the semantic user profile includes at least one habit that is inferred based at least on temporal information in the user behavior data.
5. The method of claim 1 , further comprising retrieving a static user profile for each user, and wherein the semantic user profile is further based on analyzing the static user profile.
6. The method of claim 1 , wherein the generating and the indexing of the semantic user profiles are performed offline.
7. The method of claim 1 , wherein the ranking of the set of indexed semantic user profiles is further based on a correspondence between the at least one semantic query and each of the set of indexed semantic user profiles.
8. The method of claim 1 , wherein ranking the set of indexed semantic user profiles employs a ranking model developed using machine learning.
9. The method of claim 8 , wherein the machine learning is supervised.
10. The method of claim 8 , further comprising developing the ranking model offline.
11. The method of claim 1 , wherein the at least one semantic query includes at least one of a user interest domain and a user intent.
12. A system for facilitating online advertising, comprising:
one or more processors;
an extraction component, executed by at least one of the one or more processors, that extracts a dynamic user profile from a user behavior data stream;
a semantic build component, executed by at least one of the one or more processors, that correlates information from the dynamic user profile to build a semantic user profile;
an indexing component, executed by at least one of the one or more processors, that indexes the semantic user profile;
a modeling component, executed by at least one of the one or more processors, that generates a user ranking model based at least on machine learning; and
a query processing component, executed by at least one of the one or more processors, that retrieves a set of indexed semantic user profiles in response to a received semantic query, wherein the query processing component ranks the set of indexed semantic user profiles based on the user ranking model.
13. The system of claim 12 , wherein the machine learning incorporates a support vector machine technique.
14. The system of claim 12 , wherein the extraction component extracts the dynamic user profile based on a predefined, moving time window of the user behavior data stream.
15. The system of claim 12 , wherein the semantic build component further incorporates a static user profile to build the semantic user profile.
16. The system of claim 12 , wherein the semantic build component further incorporates social network information to build the semantic user profile.
17. The system of claim 12 , wherein the indexing component executes in an offline mode, and wherein the query processing component executes in an online mode.
18. One or more computer-readable storage media storing computer-executable instructions that, when executed by a computer, cause the computer to perform acts comprising:
generating a semantic user profile for each of a plurality of users, based at least on user behavior data collected for each user;
indexing the semantic user profiles;
receiving a semantic query;
performing a search over the indexed semantic user profiles based on the semantic query, to enable selection of a target user segment from the plurality of users;
determining a set of the indexed semantic user profiles resulting from the search;
ranking the set of the indexed semantic user profiles based at least on an index of each of the indexed semantic user profiles of the set; and
providing the ranked set of the indexed semantic user profiles, to enable the selection of the target user segment.
19. The one or more computer-readable storage media of claim 18 , wherein the semantic query includes at least one of a user interest domain and a user intent.
20. The one or more computer-readable storage media of claim 18 , wherein the ranking employs a ranking model developed through supervised machine learning.