Content suggestions for content digests for assistant systems

Patent metadata

Field	Value
Publication number	US-11887359-B2
Application number	US-202117375473-A
Country	US
Kind code	B2
Filing date	Jul 14, 2021
Priority date	Apr 20, 2018
Publication date	Jan 30, 2024
Grant date	Jan 30, 2024

Abstract

In one embodiment, a method includes, by one or more computing systems, receiving, from a client system via an assistant xbot, a first audio input from a user, parsing the first audio input to identify a request for a content digest from an online social network, retrieving a plurality of content objects from the online social network, each of the content objects being accessible by the user, generating a customized newsfeed summary based on a determined semantical meaning of each of the plurality of content objects, and sending, to the client system via the assistant xbot, instructions for presenting an audio clip to the user responsive to the request for the content digest, wherein the audio clip comprises the customized newsfeed summary.

First claim

What is claimed is:

1. A method comprising:
receiving, from a client system via an assistant xbot, a first audio input from a user;
parsing the first audio input to identify a request for a content digest from an online social network;
retrieving a plurality of content objects from the online social network responsive to the content digest, each of the content objects being accessible by the user;
generating a customized newsfeed summary responsive to the content digest, wherein the customized newsfeed summary comprises a plurality of content summaries corresponding to the plurality of content objects, and wherein each content summary is based on a determined semantical meaning of one of the content objects of the plurality of content objects;
generating one or more content suggestions that are each interactable to respond to one or more of the retrieved content objects used to generate the customized newsfeed summary, wherein the one or more content suggestions are generated based on the determined semantical meaning of the respective content object; and
sending, to the client system via the assistant xbot, instructions for presenting an audio clip to the user responsive to the request for the content digest, wherein the audio clip comprises (1) the customized newsfeed summary comprising the plurality of content summaries corresponding to the plurality of content objects and (2) the one or more content suggestions that are each interactable to respond to one or more of the retrieved content objects used to generate the customized newsfeed summary.

2. The method of claim 1 , wherein the content suggestions comprise one or more of:
a comment;
a like;
an audio clip;
an image;
a video;
a link;
a message; or
a review.

3. The method of claim 1 , wherein the content suggestions comprises a prompt to the user to generate a comment to be associated with the corresponding content object.

4. The method of claim 1 , further comprising:
determining, based on a query model, a semantical-embedding for each retrieved content, wherein the semantical-embedding for each content object represents a semantical meaning of the respective content object;
determining one or more categories for each retrieved content object, wherein each category is selected from a plurality of predefined categories;
generating, based on a response model, a set of content suggestions for one or more of the retrieved content objects based on the one or more categories associated with the content object and the semantical-embedding of the content object, wherein each the set of content suggestions comprises one or more content suggestions responsive to the respective content object; and
ranking, for each set of content suggestions, the one or more content suggestions in the respective set based on a comparison of a semantical-embedding associated with each content suggestion to the semantical-embedding of the content object, wherein the content suggestions are presented based on their respective rankings.

5. The method of claim 4 , further comprising training the query model by:
accessing a plurality of prior content objects, wherein each of the prior content objects is associated with a predefined category from the plurality of predefined categories;
determining a plurality of semantical-embeddings for the plurality of prior content objects, respectively; and
generating a plurality of clusters of prior content objects based on the semantical-embeddings of the plurality of prior content objects.

6. The method of claim 4 , further comprising training the response model by:
accessing a plurality of prior content responses, wherein each prior content response is associated with a prior content object from the plurality of prior content objects;
determining a plurality of semantical-embeddings for the plurality of prior content responses, respectively; and
generating, for the prior content objects within each predefined category, a plurality of clusters of prior content responses based on the semantical-embeddings of the plurality of prior content responses.

7. The method of claim 6 , wherein training the response model further comprises training a pair NN model by:
selecting a prior content response associated with a prior content object as a positive input for the pair NN model; and
selecting a random prior content response as a negative input for the pair NN model.

8. The method of claim 6 , further comprising:
assigning each of the prior content responses of the plurality prior content responses to its corresponding prior content object, wherein the semantical-embedding of the prior content response is associated with the semantical-embedding of the corresponding prior content object; and
assigning the one or more clusters of prior content responses to one or more clusters of prior content objects.

9. The method of claim 8 , wherein generating the set of content suggestions for each retrieved content object further comprises:
identifying, for each of the retrieved content objects, which cluster of prior content objects the retrieved content object is associated with based on a comparison of the semantical-embedding of the retrieved content object with the semantical-embeddings of the prior content objects in the respective cluster; and
selecting, for each of the retrieved content objects, one or more of the prior content responses from the cluster of prior content responses assigned to the identified cluster of prior content objects as content suggestions for the set of content suggestions for the retrieved content object.

10. The method of claim 1 , further comprising:
determining, for each retrieved content object, whether the content object is accessible by the user based on a privacy setting associated with content object.

11. The method of claim 1 , wherein the first audio input comprises a voice command to invoke the retrieval of the plurality of objects.

12. The method of claim 1 , wherein the instructions for presenting the customized newsfeed summary comprises presenting a summary of each of the retrieved content objects based on its determined semantical meaning.

13. The method of claim 1 , further comprising:
receiving, from the client system via the assistant xbot, a second audio input from the user;
parsing the second audio input to identify a selection of one of the retrieve content objects; and
to the client system via the assistant xbot, instructions for presenting an audio clip to the user comprising the selected content object.

14. The method of claim 1 , further comprising:
accessing a user profile of the user from a user context engine; and
modifying customized newsfeed summary based on the user profile of the user.

15. The method of claim 1 , further comprising:
receiving, from the client system via the assistant xbot, a second audio input from the user;
parsing the second audio input to identify a selection of one of the one or more content suggestions to respond to the one or more of the retrieved content objects used to generate the customized newsfeed summary; and
performing, responsive to the selection, the one of the one or more content suggestions to respond to the one or more of the retrieved content objects used to generate the customized newsfeed summary.

16. One or more computer-readable non-transitory storage media embodying software that is operable when executed to:
receive, from a client system via an assistant xbot, a first audio input from a user;
parse the first audio input to identify a request for a content digest from an online social ne

Assignees

• Meta Platforms Inc

Inventors

• Zhou Zheng
• Han Kun
• Peng Fuchun

Classifications

• G06Q10/40

  Business processes related to social networking or social networking services · CPC title

• G06N3/0442

  characterised by memory or gating, e.g. long short-term memory [LSTM] or gated recurrent units [GRU] · CPC title

• G06N3/0464

  Convolutional networks [CNN, ConvNet] · CPC title

• G06N3/09

  Supervised learning · CPC title

• G06Q10/42

  Determination of affinities or common interests between users · CPC title