Machine learning techniques for advanced frequency management
Systems and methods for frequency management, including: an online media service configured to receive a request for a media item, the request comprising a recipient identifier of a recipient, and identify a set of candidate media items ranked based at least partially on relevance to the recipient; and a frequency management service configured to: (i) identify a quantity of impressions associated with a first candidate media item of the set of candidate media items and the recipient identifier over a preceding duration of time, (ii) identify a maximum frequency threshold, (iii) determine, based on the quantity of impressions, that the maximum frequency threshold is exceeded, (iv) exclude the first candidate media item from a result set based on the frequency threshold being exceeded, and (v) provide the result set comprising an identifier of a second candidate media item in response to the request.
1. A system for frequency management, comprising:
a computer processor;
an online media service configured to:
receive a request for a media item, the request comprising a recipient identifier of a recipient;
identify a set of candidate media items ranked based at least partially on relevance to the recipient; and
a frequency management service executing on the computer processor and configured to enable the computer processor to:
identify a quantity of impressions associated with a first candidate media item of the set of candidate media items and the recipient identifier over a preceding duration of time;
identify a maximum frequency threshold;
determine, based on the quantity of impressions, that the maximum frequency threshold is exceeded;
exclude the first candidate media item from a result set based on the frequency threshold being exceeded;
provide the result set comprising an identifier of a response candidate media item of the set of candidate media items in response to the request; and
identify an industry identifier associated with the first candidate media item, wherein the quantity of impressions is weighted according to an aggregate quantifier representing impressions corresponding to the industry identifier and the recipient identifier.
2. The system of claim 1 , wherein the online media service is further configured to:
obtain a set of cross-device identifiers associated with the recipient identifier, wherein the quantity of impressions is an aggregate of impressions corresponding to the set of cross-device identifiers and the recipient identifier.
3. The system of claim 2 , wherein the cross-device identifiers and the recipient identifier correspond to a household.
4. The system of claim 1 , wherein the online media service is further configured to:
receive a second request for a second media item, the second request comprising a second recipient identifier of a second recipient;
identify a second set of candidate media items ranked based at least partially on relevance to the second recipient;
wherein the frequency management service is further configured to enable the computer processor to:
identify a second quantity of impressions associated with a second candidate media item of the second set of candidate media items and the second recipient identifier over a second preceding duration of time;
identify a minimum frequency threshold;
determine, based at least on the second quantity of impressions, that the minimum frequency threshold is not met;
include the second candidate media item in a second result set based on the minimum frequency threshold not being met; and
provide the second result set comprising an identifier of the second candidate media item in response to the second request.
5. The system of claim 4 , wherein the minimum frequency threshold and the maximum frequency threshold form a lower and upper bound for a range of target impressions of any single entity for each recipient.
6. The system of claim 1 , wherein the identifier of the response candidate media item is associated, in a lookup cache, with a probability generated by an artificial intelligence model configured to perform entity detection on video frames of the response candidate media item, and wherein a prior request comprising an identifier of the response candidate media item triggered an ingestion process resulting in the entity identifier and the probability being stored in the lookup cache.
7. The system of claim 1 , wherein the quantity of impressions is decayed over time using a decay formula, and wherein the decay formula is configured to reduce the quantity of impressions over each of a predefined unit of time.
8. The system of claim 1 , wherein the online media service is further configured to:
receive the result set including the identifier of the response candidate media item; and
utilize the quantity of impressions as an input to a realtime bidding service to determine whether to serve the response candidate media item in response to the request.
9. A method for frequency management, comprising:
receiving a request for a media item, the request comprising a recipient identifier of a recipient;
identifying a set of candidate media items ranked based at least partially on relevance to the recipient;
identifying, using at least one computer processor, a quantity of impressions associated with a first candidate media item of the set of candidate media items and the recipient identifier over a preceding duration of time;
identifying a maximum frequency threshold;
determining, using the at least one computer processor, that the maximum frequency threshold is exceeded based on the quantity of impressions;
excluding the first candidate media item from a result set based on the frequency threshold being exceeded;
providing the result set comprising an identifier of a response candidate media item of the set of candidate media items in response to the request; and
identifying an industry identifier associated with the first candidate media item, wherein the quantity of impressions is weighted according to an aggregate quantifier representing impressions corresponding to the industry identifier and the recipient identifier.
10. The method of claim 9 , further comprising:
obtaining a set of cross-device identifiers associated with the recipient identifier, wherein the quantity of impressions is an aggregate of impressions corresponding to the set of cross-device identifiers and the recipient identifier.
11. The method of claim 10 , wherein the cross-device identifiers and the recipient identifier correspond to a household.
12. The method of claim 9 , further comprising:
receiving a second request for a second media item, the second request comprising a second recipient identifier of a second recipient;
identifying a second set of candidate media items ranked based at least partially on relevance to the second recipient;
identifying a second quantity of impressions associated with a second candidate media item of the second set of candidate media items and the second recipient identifier over a second preceding duration of time;
identifying a minimum frequency threshold;
determining, based at least on the second quantity of impressions, that the minimum frequency threshold is not met;
including the second candidate media item in a second result set based on the minimum frequency threshold not being met; and
providing the second result set comprising an identifier of the second candidate media item in response to the second request.
13. The method of claim 12 , wherein the minimum frequency threshold and the maximum frequency threshold form a lower and upper bound for a range of target impressions of any single entity for each recipient.
14. The method of claim 9 , wherein the identifier of the response candidate media item is associated, in a lookup cache, with a probability generated by an artificial intelligence model configured to perform entity detection on video frames of the response candidate media item, and wherein a prior request comprising an identifier of the response candidate media item triggered an ingestion process resulting in the entity identifier and the probability being stored in the lookup cache.
15. The method of claim 9 , wherein the quantity of impressions is decayed over time using a decay formula, and wherein the decay formula is configured to reduce the quantity of impressions over each of a predefined unit of time.
16. The method of claim 9 , further comprising:
receiving the result set including the identifier of the response candidate media item; and
utilizing the quantity of impressions as an input to a realtime bidding service to determine whether to serve the response candidate media item in response to the request.
17. A non-transitory computer-readable storage medium comprising a plurality of instructions for frequency management, the plurality of instructions configured to execute on at least one computer processor to enable the at least one computer processor to:
receive a request for a media item, the request comprising a recipient identifier of a recipient;
identify a set of candidate media items ranked based at least partially on relevance to the recipient;
identify a quantity of impressions associated with a first candidate media item of the set of candidate media items and the recipient identifier over a preceding duration of time;
identify a maximum frequency threshold;
determine, based on the quantity of impressions, that the maximum frequency threshold is exceeded;
exclude the first candidate media item from a result set based on the frequency threshold being exceeded;
provide the result set comprising an identifier of a response candidate media item of the set of candidate media items in response to the request; and
identify an industry identifier associated with the first candidate media item, wherein the quantity of impressions is weighted according to an aggregate quantifier representing impressions corresponding to the industry identifier and the recipient identifier.
18. The non-transitory computer-readable storage medium of claim 17 , the plurality of instructions further configured to enable the at least one computer processor to:
obtain a set of cross-device identifiers associated with the recipient identifier, wherein the quantity of impressions is an aggregate of impressions corresponding to the set of cross-device identifiers and the recipient identifier.