Granular data for behavioral targeting
A method of targeting receives several granular events and preprocesses the received granular events thereby generating preprocessed data to facilitate construction of a model based on the granular events. The method generates a predictive model by using the preprocessed data. The predictive model is for determining a likelihood of a user action. The method trains the predictive model. A system for targeting includes granular events, a preprocessor for receiving the granular events, a model generator, and a model. The preprocessor has one or more modules for at least one of pruning, aggregation, clustering, and/or filtering. The model generator is for constructing a model based on the granular events, and the model is for determining a likelihood of a user action. The system of some embodiments further includes several users, a selector for selecting a particular set of users from among the several users, a trained model, and a scoring module.
1. A method comprising:
receiving, via a processor, a plurality of events, wherein an event comprises an instance of online activity between a user and at least one website;
processing, via the processor, the received events by analyzing each event and determining, based on said analysis, content of each event, said processing further comprising identifying, based on the determined content of each event, a type of predictive model;
generating, via the processor, a plurality of input features for use by the type of predictive model by grouping the events into one or more clusters based on the content of each event, wherein each cluster corresponds to a unique input feature;
generating, via the processor, the predictive model, based on a machine-learning model, in accordance with said generated clusters, event content and the type of predictive model;
applying, via the processor, said generated predictive model to said received plurality of events; and
targeting, via the processor, an advertisement to the user based on the applied predictive model.
2. The method as set forth in claim 1 , wherein the processing of the events comprises processing of events to remove events that are associated with fewer than a threshold amount of users.
3. The method as set forth in claim 1 , wherein an event comprises a search query that comprises a search term, and wherein the grouping of the events comprises grouping the events into clusters based on the search term of the search query.
4. The method as set forth in claim 1 , wherein the generating of the predictive model comprises generating a first predictive model based on data types such that the first predictive model is generated for each data type.
5. The method as set forth in claim 4 , wherein the data types comprise a number of advertisement clicks the user is expected to generate and a number of advertisement views that the user is expected to generate.
6. The method as set forth in claim 1 , wherein the predictive model comprises a Poisson type model.
7. The method as set forth in claim 6 , further comprising using the Poisson type model to score the user based on a ratio of a predicted number of advertisement clicks associated with the user divided by a predicted number of advertisement views associated with the user.
8. A non-transitory computer readable storage medium storing thereon computer-executable instructions, that when executed by a processor, cause the processor to perform the steps of:
receiving, via the processor, a plurality of events, wherein an event comprises an instance of online activity between a user and a website;
processing, via the processor, the received events by analyzing each event and determining, based on said analysis, content of each event, said processing further comprising identifying, based on the determined content of each event, a type of predictive model;
generating, via the processor, a plurality of input features for use by the type of predictive model by grouping the events into one or more clusters based on the content of each event, wherein each cluster corresponds to a unique input feature;
generating, via the processor, predictive model, based on a machine-learning model, in accordance with said generated clusters, event content and the type of predictive model;
applying, via the processor, said generated predictive model to said received plurality of events; and
targeting, via the processor, an advertisement to the user based on the applied predictive model.
9. The non-transitory computer readable storage medium as set forth in claim 8 , wherein the processing of the events comprises processing of events to remove events that are associated with fewer than a threshold amount of users.
10. The non-transitory computer readable storage medium as set forth in claim 8 , wherein an event comprises a search query that comprises a search term, and wherein the grouping of the events comprises grouping the events into clusters based on the search term of the search query.
11. The non-transitory computer readable storage medium as set forth in claim 8 , wherein the generating of the predictive model comprises generating a first predictive model based on data types such that the first predictive model is generated for each data type.
12. The non-transitory computer readable storage medium as set forth in claim 11 , wherein the data types comprise a number of advertisement clicks the user is expected to generate and a number of advertisement views that the user is expected to generate.
13. The non-transitory computer readable storage medium as set forth in claim 8 , wherein the predictive model comprises a Poisson type model.
14. The non-transitory computer readable storage medium as set forth in claim 13 , further comprising using the Poisson type model to score the user based on a ratio of a predicted number of advertisement clicks associated with the user divided by a predicted number of advertisement views associated with the user.
15. A system comprising:
a processor; and
a non-transitory computer-readable storage medium for tangibly storing therein program logic for execution by the processor, the stored program logic comprising:
event receiving logic executed by the processor for receiving a plurality of events, wherein an event comprises an instance of online activity between a user and at least one website;
the event logic executed by the processor for processing the received events by analyzing each event and determining, based on said analysis, content of each event, said processing further comprising identifying, based on the determined content of each event, a type of predictive model;
cluster logic executed by the processor for generating a plurality of input features for use by the type of predictive model by grouping the events into one or more clusters based on the content of each event, wherein each cluster corresponds to a unique input feature;
model generating logic executed by the processor for generating the predictive model, based on a machine-learning model, in accordance with said generated clusters, event content and the type of predictive model;
the model generating logic executed by the processor for applying said generated predictive model to said received plurality of events; and
advertisement logic executed by the processor for targeting an advertisement to the user based on the applied predictive model.
16. The system as set forth in claim 15 , wherein the event processing logic further comprises event removing logic executed by the processor for removing events that are associated with fewer than a threshold amount of users.
17. The system as set forth in claim 15 , wherein an event comprises a search query comprising a search term, and wherein the cluster grouping logic for grouping the events comprises search term cluster grouping logic executed by the processor for grouping the events into clusters based on the search term of the search query.
18. The system as set forth in claim 15 , wherein the generating of the predictive model is further based on data types such that a first predictive model is generated for each data type.
19. The system as set forth in claim 18 , wherein the data types comprise a number of advertisement clicks the user is expected to generate and a number of advertisement views that the user is expected to generate.
20. The system as set forth in claim 15 , wherein the predictive model comprises a Poisson type model.