Classifying search queries on online social networks

• US 10,726,022 B2
• Filed: 08/26/2016
• Issued: 07/28/2020
• Est. Priority Date: 08/26/2016
• Status: Expired due to Fees

• Current Assignee: Meta Platforms, Inc. (f/k/a Facebook, Inc.)
• Original Assignee: Meta Platforms, Inc. (f/k/a Facebook, Inc.)
• Inventors: Kuo, Shiun-Zu, Stoyanov, Veselin S., Philip, Rose Marie, Winstanley, Melissa Rose
• Primary Examiner(s): Fan, Shiow-Jy
• Application Number:
  US15/248,993
• Publication Number:
  US 20180060326A1
• Time in Patent Office: 1,432 Days
• Field of Search: 707748, 707E17014, 707722, 707706
• CPC Class Codes: G06F 16/24578 using ranking
• CPC Class Codes: G06F 16/353 into predefined classes
• CPC Class Codes: G06F 16/9535 Search customisation based on user profiles and personalisation
• CPC Class Codes: G06Q 50/01 Social networking

Abstract

In one embodiment, a method includes receiving a search query inputted by a first user, wherein the search query comprises one or more n-grams; calculating a needle-confidence score for the search query that is calculated by a needle-intent classifier based on at least the n-grams of the search query and a language model analysis of the n-grams, and wherein the needle-confidence score represents a probability that the search query was intended as a needle search; classifying the search query as a needle search if the calculated needle-confidence score is above a threshold confidence score; and generating a plurality of search-result modules, each search-result module comprising one or more search results matching the search query, wherein one of the search-result modules is a social module, and wherein the number of search results in the social module is based on the classification of the search query as a needle search.

Claims

1. A method comprising:

receiving, from a client system of a first user of an online social network, a search query inputted by the first user, wherein the search query comprises one or more n-grams;

calculating a needle-confidence score for the search query, wherein the needle-confidence score is calculated by a needle-intent classifier based on at least the n-grams of the search query and a language model analysis of the n-grams, and wherein the needle-confidence score represents a probability that the search query was intended as a needle search for specific content objects associated with the first user'"'"'s social network connections and previously viewed by the first user;

classifying the search query as a needle search if the calculated needle-confidence score is above a threshold confidence score;

identifying, responsive to the classification of the search query as a needle search, one or more content objects matching the search query, each identified content object being associated with the first user'"'"'s social network connections and previously viewed by the first user; and

generating a plurality of search-result modules, each search-result module comprising one or more search results referencing one or more content objects matching the search query, respectively, wherein one of the search-result modules is a social module, wherein a number of search results in the social module is based on the classification of the search query as a needle search, and wherein one or more of the search results in the social module reference one or more of the identified content objects associated with the first user'"'"'s social network connections and previously viewed by the first user.

2. The method of claim 1, further comprising:

calculating, for the search query;

a grammar-confidence score;

a person-confidence score; and

a public-confidence score; and

classifying the search query as one of;

a needle search, a grammar search, a person search, or a public search based on the relative values of the needle-confidence score, the grammar-confidence score, the person-confidence score, and the public-confidence score, respectively.

3. The method of claim 1, wherein, if the search query is classified as a needle search, the number of search results in the social module is greater than a number of search results in each other search-result module in the plurality of search-result modules.

4. The method of claim 1, further comprising:

calculating, for each search-result module, a leaf score for the search-result module, wherein the leaf score indicates a measure of closeness between the search query and the one or more search results in the search-result module; and

adjusting a number of search results in one or more of the search-result modules based on the leaf score for the respective search-result module.

5. The method of claim 1, wherein the language model is a query likelihood model wherein the query likelihood model is used to rank one or more of the search results.

6. The method of claim 1, wherein the needle-confidence score is further based on one or more contextual factors, wherein the one or more contextual factors comprise a time of day that the first user inputted the search query, one or more related searches previously inputted by the first user, or one or more content objects the first user has previously interacted with.

7. The method of claim 1, further comprising:

training the needle-intent classifier based on social-graph information of the first user and click-through data of the first user.

8. The method of claim 1, wherein the threshold confidence score has been predetermined by an administrator of the online social network.

9. The method of claim 1, wherein the threshold confidence score is dynamically determined based on the n-grams and one or more contextual factors.

10. The method of claim 1, further comprising:

accessing a social graph comprising a plurality of nodes and a plurality of edges connecting the nodes, each of the edges between two of the nodes representing a single degree of separation between them, the nodes comprising;

a first node corresponding to the first user; and

a plurality of second nodes corresponding to a plurality of second users of the online social network, respectively.

11. The method of claim 10, wherein the social module comprises one or more posts by one or more of the second users within a threshold degree of separation of the first user in the social graph.

12. The method of claim 11, wherein each search-result module comprises a selectable see-more element operable to retrieve one or more additional search results for the respective search-result module.

13. The method of claim 12, further comprising:

receiving an indication that the first user has selected the see-more element; and

using the indication as training data for the needle-intent classifier.

14. The method of claim 1, wherein the classification of the search query as a needle search indicates the search query is intended to retrieve a specific content object that the first user has previously interacted with.

15. The method of claim 1, wherein the needle-confidence score is defined as:

16. The method of claim 1, further comprising:

identifying, responsive to the search query, one or more content objects associated with the online social network that match the search query and were not previously viewed by the first user; and

generating one or more search result corresponding to one or more of the identified content objects, respectively,

wherein one of the search-result modules is a public module comprising one or more search results referencing one or more of the content objects matching the search query that were not previously viewed by the first user.

17. The method of claim 1, further comprising:

sending, to the client system in response to receiving the search query, a search-results interface comprising one or more of the search-result modules for display, wherein at least one of the displayed search-result modules comprises the social module.

18. One or more computer-readable non-transitory storage media embodying software that is operable when executed to:

receive, from a client system of a first user of an online social network, a search query inputted by the first user, wherein the search query comprises one or more n-grams;

calculate a needle-confidence score for the search query, wherein the needle-confidence score is calculated by a needle-intent classifier based on at least the n-grams of the search query and a language model analysis of the n-grams, and wherein the needle-confidence score represents a probability that the search query was intended as a needle search for specific content objects associated with the first user'"'"'s social network connections and previously viewed by the first user;

classify the search query as a needle search if the calculated needle-confidence score is above a threshold confidence score;

identify, responsive to the classification of the search query as a needle search, one or more content objects matching the search query, each identified content object being associated with the first user'"'"'s social network connections and previously viewed by the first user; and

generate a plurality of search-result modules, each search-result module comprising one or more search results referencing one or more content objects matching the search query, respectively, wherein one of the search-result modules is a social module, wherein a number of search results in the social module is based on the classification of the search query as a needle search, and wherein one or more of the search results in the social module reference one or more of the identified content objects associated with the first user'"'"'s social network connections and previously viewed by the first user.

19. A system comprising:

one or more processors; and

a non-transitory memory coupled to the processors comprising instructions executable by the processors, the processors operable when executing the instructions to;

receive, from a client system of a first user of an online social network, a search query inputted by the first user, wherein the search query comprises one or more n-grams;

calculate a needle-confidence score for the search query, wherein the needle-confidence score is calculated by a needle-intent classifier based on at least the n-grams of the search query and a language model analysis of the n-grams, and wherein the needle-confidence score represents a probability that the search query was intended as a needle search for specific content objects associated with the first user'"'"'s social network connections and previously viewed by the first user;

classify the search query as a needle search if the calculated needle-confidence score is above a threshold confidence score;

identify, responsive to the classification of the search query as a needle search, one or more content objects matching the search query, each identified content object being associated with the first user'"'"'s social network connections and previously viewed by the first user; and

generate a plurality of search-result modules, each search-result module comprising one or more search results referencing one or more content objects matching the search query, respectively, wherein one of the search-result modules is a social module, wherein a number of search results in the social module is based on the classification of the search query as a needle search, and wherein one or more of the search results in the social module reference one or more of the identified content objects associated with the first user'"'"'s social network connections and previously viewed by the first user.