Temporal features in a messaging platform

• US 9,117,227 B1
• Filed: 03/14/2014
• Issued: 08/25/2015
• Est. Priority Date: 03/31/2011
• Status: Expired due to Fees

• Current Assignee: X Corp. (f/k/a Twitter, Inc.) (X Holdings Corp.)
• Original Assignee: X Corp. (f/k/a Twitter, Inc.) (X Holdings Corp.)
• Inventors: Ma, Qicheng, Agrawal, Parag, Srivastava, Utkarsh, Chellapilla, Kumar
• Primary Examiner(s): Lu, Kuen
• Application Number:
  US14/214,523
• Time in Patent Office: 529 Days
• Field of Search: 707/732
• US Class Current: 1/1
• CPC Class Codes: G06F 16/24578 using ranking
• CPC Class Codes: G06F 16/335 Filtering based on additional data, e.g. user or group profiles filtering in web context G06F16/9535, G06F16/9536
• CPC Class Codes: G06F 16/951 Indexing; Web crawling techniques
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation
• CPC Class Codes: G06F 16/9538 Presentation of query results
• CPC Class Codes: G06Q 30/0242 Determining effectiveness of advertisements
• CPC Class Codes: G06Q 30/0254 based on statistics
• CPC Class Codes: G06Q 30/0275 Auctions
• CPC Class Codes: H04L 51/04 Real-time or near real-time messaging, e.g. instant messaging [IM]

Abstract

A real-time messaging platform allows advertiser accounts to pay to insert candidate messages into the message streams requested by account holders. To accommodate multiple advertisers, the messaging platform controls an auction process that determines which candidate messages are selected for inclusion in a requested account holder'"'"'s message stream. Selection is based on a bid for the candidate message, the message stream that is requested, and a variety of other factors that vary depending upon the implementation. The process for selection of candidate messages generally includes the following steps, though any given step may be omitted or combined into another step in a different implementation: targeting, filtering, prediction, ranking, and selection.

Claims

1. A computer-implemented method comprising:

identifying a requested account holder of a messaging platform associated with one or more temporal account features, each temporal account feature associated with a temporal account weight calculated based on engagement data representing past engagements by account holders with messages present in the message streams of other account holders, the engagement data associated with timestamps, the temporal account features determined based in part on the timestamps;

identifying a set of candidate messages each associated with one or more temporal message features;

for each candidate message of the set of candidate messages;

accessing temporal message weights for the temporal message features associated with the candidate message, the temporal message weights calculated based on the engagement data representing past engagements by account holders with messages, the temporal message features determined based in part on the timestamps of the engagement data; and

determining, by a processor, a value representing a likelihood of engagement with the candidate message by the requested account holder, the likelihood of engagement based at least on the temporal message weights and the temporal account weights; and

selecting at least one of the set of candidate messages for inclusion in a message stream of the requested account holder based on the determined values.

2. The method of claim 1, wherein the temporal message features comprise a fatigue feature, the fatigue feature based on a fatigue score between the requested account holder and the candidate message.

3. The method of claim 2, wherein the fatigue score is determined based on a weighted combination of exponential decay factors, each exponential decay factor corresponding to an impression between the requested account and the candidate message, each exponential decay factor based on time elapsed since the impression between the requested account and the candidate message.

4. The method of claim 2, wherein the fatigue score is determined based on a weighted combination of impressions, each impression weighted based on at least one of:

an engagement between the requested account holder and the candidate message and a lack of engagement between the requested account holder and the candidate message.

5. The method of claim 2, wherein the fatigue score is determined based on impressions between the requested account and other messages authored by the account that authored the candidate message.

6. The method of claim 1, wherein at least one of the temporal account features represents an inferred attribute of the requested account holder, the inferred attribute based on engagement data meeting temporal criteria.

7. The method of claim 6, wherein the engagement data meeting temporal criteria comprise engagement data drawn from at least one of a recurring time period and a recent time period.

8. The method of claim 6, wherein the inferred attribute of the requested account holder is at least one of:

a location, an interest, a device type used to access the messaging platform, and a software type used to access the messaging platform.

9. The method of claim 1, wherein at least one of the temporal account features represents an inferred attribute of the requested account holder, the inferred attribute based on engagement data meeting temporal criteria.

10. The method of claim 1, wherein determining the value representing the likelihood of engagement comprises:

accessing cross temporal weights for cross temporal features between the requested account holder and the candidate message, the cross temporal weights calculated based on the engagement data representing past engagements by account holders with messages and based on timestamps of the engagement data; and

determining the value representing the likelihood of engagement based in part on the cross weights.

11. The method of claim 10, wherein at least one of the temporal cross features has a value based on a current time and a temporal account feature associated with a recurring time period comprising the current time.

12. The method of claim 1, wherein the past engagements include engagements performed by subscribing accounts subscribed to receive messages from the requested account holder.

13. The method of claim 1, wherein the past engagements include engagements performed by publishing accounts that the requested account holder is subscribed to receive messages from.

14. The method of claim 1, wherein the past engagements include engagements performed by subscribing accounts subscribed to receive messages from publishing accounts that the requested account holder is subscribed to receive messages from.

15. A non-transitory computer-readable storage medium comprising instructions executable by a processor, the instructions for:

identifying a requested account holder of a messaging platform associated with one or more temporal account features, each temporal account feature associated with a temporal account weight calculated based on engagement data representing past engagements by account holders with messages present in the message streams of other account holders, the engagement data associated with timestamps, the temporal account features determined based in part on the timestamps;

identifying a set of candidate messages each associated with one or more temporal message features;

for each candidate message of the set of candidate messages;

accessing temporal message weights for the temporal message features associated with the candidate message, the temporal message weights calculated based on the engagement data representing past engagements by account holders with messages, the temporal message features determined based in part on the timestamps of the engagement data; and

determining a value representing a likelihood of engagement with the candidate message by the requested account holder, the likelihood of engagement based at least on the temporal message weights and the temporal account weights; and

selecting at least one of the set of candidate messages for inclusion in a message stream of the requested account holder based on the determined values.

16. The computer-readable medium of claim 15, wherein the temporal message features comprise a fatigue feature, the fatigue feature based on a fatigue score between the requested account holder and the candidate message.

17. The computer-readable medium of claim 15, wherein at least one of the temporal account features represents an inferred attribute of the requested account holder, the inferred attribute based on engagement data meeting temporal criteria.

18. The computer-readable medium of claim 15, wherein at least one of the temporal account features represents an inferred attribute of the requested account holder, the inferred attribute based on engagement data meeting temporal criteria.

19. The computer-readable medium of claim 15, wherein instructions for determining the value representing the likelihood of engagement comprise instructions for:

accessing cross temporal weights for cross temporal features between the requested account holder and the candidate message, the cross temporal weights calculated based on the engagement data representing past engagements by account holders with messages and based on timestamps of the engagement data; and

determining the value representing the likelihood of engagement based in part on the cross weights.

20. The computer-readable medium of claim 15, wherein the past engagements include engagements performed by subscribing accounts subscribed to receive messages from publishing accounts that the requested account holder is subscribed to receive messages from.