Systems and methods for improving content recommendations using a trained model
Systems and methods are disclosed herein for a recommendations engine that generates content recommendations using a trained model that is personalized based on the information corresponding to content consumption. The disclosed techniques herein provide a trained model to provide content recommendations. The trained model may have been trained using a predefined set of training data agnostic of a particular user profile. A system receives information corresponding to content consumption. The system may associate the information corresponding to content consumption with a profile. The system generates a personalized model based on the information corresponding to content consumption and on the trained model. The personalized model may be associated with the user profile. The system generates the content recommendations using the personalized model. The system then causes to be provided the content recommendations.
1. A method of personalizing a model for recommending content, the method comprising:
accessing a first model for generating content recommendations, the first model trained using a predefined training data set;
retrieving content consumption information from a user profile, wherein the content consumption information comprises information that was collected subsequent to defining the predefined training data set, wherein the predefined training data set does not include the information that was collected subsequent to defining the predefined training data set;
generating a second model based on the first model and the content consumption information, wherein generating the second model comprises:
determining one or more states, starting with a plurality of weights associated with the first model and iteratively updating the plurality of weights based on the content consumption information;
determining a plurality of optimized weights corresponding to the user profile; and
generating the second model based on the plurality of optimized weights and on the one or more states; and
generating, for display and using the second model, the content recommendations.
2. The method of claim 1 , wherein the predefined training data set is agnostic of the user profile.
3. The method of claim 1 , wherein the content consumption information is a first content consumption information, the method further comprising:
receiving a second content consumption information comprising information not included in the first content consumption information;
generating a third model based on the second content consumption information and on the second model; and
generating, using the third model, second content recommendations.
4. The method of claim 1 , wherein the content consumption information comprises activity data collected during content consumption.
5. The method of claim 4 , wherein the activity data comprises a control function selection made during content consumption.
6. The method of claim 1 , wherein the content consumption information comprises one or more of a time of consumption, a location of consumption, a genre of content consumed, a type of content consumed.
7. The method of claim 6 , further comprising:
ranking content genres in the content consumption information; and
wherein generating the content recommendations further comprises ordering the content recommendations based on the ranking.
8. The method of claim 1 , wherein the content consumption information is based on at least one of full consumption of content, partial consumption of content, or frequency of consumption of content.
9. The method of claim 1 , wherein generating the content recommendations comprises generating recommendations comprising one or more portions of a content item using the second model.
10. A system for personalizing a model for recommending content, the system comprising:
control circuitry configured to access a first model for generating content recommendations, the first model trained using a predefined training data set; and
communication circuitry configured to retrieve content consumption information from a user profile, wherein the content consumption information comprises information that was collected subsequent to defining the predefined training data set, wherein the predefined training data set does not include the information that was collected subsequent to defining the predefined training data set;
wherein the control circuitry is further configured to:
generate a second model based on the first model and the content consumption information, wherein the control circuitry is configured to:
determine one or more states, starting with a plurality of weights associated with the first model and iteratively updating the plurality of weights based on the content consumption information;
determine a plurality of optimized weights corresponding to the user profile; and
generate the second model based on the plurality of optimized weights and on the one or more states; and
generate, for display and using the second model, the content recommendations.
11. The system of claim 10 , wherein the predefined training data set is agnostic of the user profile.
12. The system of claim 10 , wherein the content consumption information is a first content consumption information, and wherein the control circuitry is further configured to:
receive a second content consumption information comprising information not included in the first content consumption information;
generate a third model based on the second content consumption information and on the second model; and
generate, using the third model, second content recommendations.
13. The system of claim 10 , wherein the content consumption information comprises activity data collected during content consumption.
14. The system of claim 13 , wherein the activity data comprises a control function selection made during content consumption.
15. The system of claim 10 , wherein the content consumption information comprises one or more of a time of consumption, a location of consumption, a genre of content consumed, a type of content consumed.
16. The system of claim 15 , wherein the control circuitry is further configured to:
rank content genres in the content consumption information; and
order the content recommendations based on the ranking.
17. The system of claim 10 , wherein the content consumption information is based on at least one of full consumption of content, partial consumption of content, or frequency of consumption of content.
18. The system of claim 10 , wherein the control circuitry is configured to generate recommendations comprising one or more portions of a content item using the second model.