Promoting content in a real-time messaging platform

Patent metadata

Field	Value
Publication number	US-10163133-B2
Application number	US-201615077847-A
Country	US
Kind code	B2
Filing date	Mar 22, 2016
Priority date	Mar 31, 2011
Publication date	Dec 25, 2018
Grant date	Dec 25, 2018

Abstract

A real-time messaging platform and method are disclosed which can be used to promote content in the messaging platform. In one embodiment, the promotion system is disclosed which performs initial candidate selection so as to narrow down the set of candidate promotions before applying more expensive processing. The candidate selection takes advantage of the connection graph information associated with accounts in the messaging platform to identify targeted accounts. In another embodiment, the promotion system uses a prediction model to predict a user's engagement with the promotion and utilizes the prediction to assist in ranking the candidate promotions. Promotions can be assigned metrics based, for example, on a weighted combination of user engagement rates, decayed with time to reflect an intuition that recent interactions by one or more users with the promotion will have a greater impact than older interactions with the promotion.

First claim

What is claimed is:

1. A method comprising:
receiving a request for a stream of messages to be provided to a requesting user of a real-time messaging platform that provides streams of messages authored by users of the platform;
filtering a set of available promotion items to identify a subset of candidate promotion items based on a fatigue map previously generated and stored in a memory object caching system, thereby removing those promotion items where prior presentations of promotion items to the requesting user in messages has exceeded a threshold;
for each of the subset of candidate promotion items remaining after filtering,
computing a level of engagement based on interactions with messages including the candidate promotion item by users of the platform who were previously presented the candidate promotion item in streams of messages, and
decaying the level of engagement based on an amount of time elapsed since the interactions; and

selecting at least one of the subset of candidate promotion items for inclusion in the message stream based on the decayed levels of engagement; and
providing the message stream in response to the request.

2. The method of claim 1 , wherein the threshold is a fatigue threshold, and wherein the filtering comprises:
comparing the fatigue threshold with a total number of promotion items in the set of available promotion items; and
selecting at least one candidate promotion items based on the comparison.

3. The method of claim 1 , wherein the threshold is a fatigue threshold, and wherein the filtering comprises:
determining a promoter associated with a first available promotion item;
comparing the fatigue threshold with a total number of promotion items in the set of available promotion items that are associated with the promoter; and
selecting at least one candidate promotion items based on the comparison.

4. The method of claim 1 , wherein the fatigue map indicates, for a first promotion item in the set of available promotion items, a time period within which the first promotion item was presented to the requesting user, wherein the threshold is a minimum time threshold, and wherein filtering comprises:
determining a promoter associated with a first available promotion item;
determining that the first promotion item is associated with the promoter; and
comparing the minimum time threshold set for the promoter with an elapsed time since the time period; and
selecting at least one candidate promotion items based on the comparison.

5. The method of claim 1 , wherein the threshold is a maximum promoter threshold, and wherein the filtering comprises:
determining a promoter associated with a first available promotion item;
determining a number of distinct promoters associated with the promotion items in the set of available promotion items, the promoter associated with the first available promotion item not included in the distinct promoters; and
filtering the first available promotion item from the candidate promotion items when the number of distinct promoters exceeds the maximum promoter threshold.

6. The method of claim 1 , wherein the filtering is further based on a fatigue model that indicates one or more preferences of the requesting user for promoted content in the set of available promotion items.

7. The method of claim 6 , wherein filtering further comprises:
determining a promotion type associated with a first available promotion item;
determining a first preference of the requesting user toward promoted content in the set of available promotion items having the promotion type; and
selecting at least one candidate promotion items based on the first preference.

8. The method of claim 1 , wherein, for a candidate promotion item, computing a level of engagement comprises:
identifying one or more user engagement events for the candidate promotion item, each user engagement event indicating (1) an interaction between a user of the real-time messaging platform and the candidate promotion item and (2) a time when the interaction occurred; and
processing the user engagement events to compute the level of engagement.

9. The method of claim 1 , wherein the fatigue map is generated by an asynchronous process in advance of the request related to the requesting user.

10. The method of claim 1 , wherein the fatigue map comprises rankings for at least one of candidate promotions items and promoters, and wherein the rankings are generated based on a fatigue model that takes into account the requesting user's engagement behavior towards at least one of the candidate promotion items and the promoters.

11. The method of claim 1 , wherein the fatigue map includes a stream frequency filter for at least one of a candidate promotion item and a promoter.

12. The method of claim 1 , wherein the fatigue map includes an impression limit filter for at least one of a candidate promotion item and a promoter.

13. The method of claim 1 , wherein the fatigue map includes a campaign filter associated with at least one of a candidate promotion item and a promoter.

14. The method of claim 1 , wherein the fatigue map includes a filter based on at least one characteristic of the requesting user.

15. A non-transitory computer readable storage medium storing instructions, the instructions when executed by a processor configured to cause the processor to:
receive a request for a stream of messages to be provided to a requesting user of a real-time messaging platform that provides streams of messages authored by users of the platform;
filter a set of available promotion items to identify a subset of candidate promotion items based on a fatigue map previously generated and stored in a memory object caching system, thereby removing those promotion items where prior presentations of promotion items to the requesting user in messages has exceeded a threshold;
for each of the subset of candidate promotion items remaining after filtering,
compute a level of engagement based on interactions with messages including the candidate promotion item by users of the platform who were Currently Amended the candidate promotion item in streams of messages, and
decay the level of engagement based on an amount of time elapsed since the interactions; and
select at least one of the subset of candidate promotion items for inclusion in the message stream based on the decayed levels of engagement; and
provide the message stream in response to the request.

16. The non-transitory computer readable storage medium of claim 15 , wherein the threshold is a fatigue threshold, and wherein the filtering comprises:
comparing the fatigue threshold with a total number of promotion items in the set of available promotion items; and
selecting at least one candidate promotion items based on the comparison.

17. The non-transitory computer readable storage medium of claim 15 , wherein the threshold is a fatigue threshold, and wherein the filtering comprises:
determining a promoter associated with a first available promotion item;
comparing the fatigue threshold with a total number of promotion items in the set of available promotion items that are associated with the promoter; and
selecting at least one candidate promotion items based on the comparison.

18. The non-transitory computer readable storage medium of claim 15 , wherein the fatigue map indicates, for a first promotion item in the set of available promotion items, a time period within which the first promotion item was presented to the requesting user, the threshold is a minimum time thresho

Assignees

• Twitter Inc

Inventors

• Srivastava Utkarsh
• Goel Ashish
• Rajgopal Srinivasan
• Banerji Anamitra

Classifications

• G06Q30/0242

  Determining effectiveness of advertisements · CPC title

• G06F17/3056

  Physics · mapped topic

• G06Q30/0269 Primary

  based on user profile or attribute · CPC title

• G06Q30/0275

  Auctions · CPC title

• G06F17/30528

  Physics · mapped topic