Providing user-specific previews within text
Examples described herein include systems and methods for providing user-specific previews for terms within text. An example method can include receiving tracked user behavior reflecting terms selected by a user and entered into a search. A representation of known words can be created based on the tracked user behavior. By training machine-learning models for each individual user, personalized previews can be presented when each user encounters a new body of text, such as in a webpage or email. The preview can apply to a term not previously known to the user but likely to be searched by the user, relying on content gathered from a search on a search medium that the user was likely to use. The content can be presented to the user in a graphical user interface allowing for interaction and feedback.
1. A method for providing user-specific previews for terms within text, comprising:
sending, from a user device to a management server, tracked user search behavior, wherein the user search behavior indicates a term selected by a user from of a body of text and entered into a search;
receiving, at the user device from the management server:
a representation of known terms previously searched by the user or frequently appearing to the user;
term categories that categorize terms selected by the user;
source categories that categorize sources of texts containing the terms selected by the user;
search-medium categories that categorize the search mediums utilized by the user to search the terms selected by the user; and
a machine-learning model trained at the management server using the tracked user search behavior;
parsing, by the user device, a new body of text viewed by the user;
identifying a candidate term within the new body of text by excluding stop words and terms matching the known terms in the stored representation;
predicting a search medium most likely to be used by the user to search for the candidate term, wherein the predicting is performed at the user device by implementing the machine-learning model using the term categories, source categories, and search-medium categories;
performing a backend search for the candidate term using a search medium selected based on the tracked user search behavior;
extracting information from the backend search and converting the extracted information into an image or preview card;
injecting the image or preview card into the candidate term within the new body of text; and
displaying the image or preview card for the candidate term upon the user selecting or hovering over the candidate term.
2. The method of claim 1 , wherein the machine-learning model outputs an identification of the search medium most likely to be utilized by the user to search the candidate term.
3. The method of claim 1 , further comprising displaying multiple images or preview cards for the candidate term, wherein each of the multiple images or preview cards is retrieved from a different search medium.
4. The method of claim 1 , further comprising determining that the tracked user search behavior indicates that the user searches for terms with at least a threshold frequency.
5. The method of claim 1 , wherein identifying a candidate term further comprises ranking a plurality of potential candidate terms and selecting the candidate term based on the ranking.
6. The method of claim 1 , wherein displaying the image or preview card further comprises displaying a feedback element allowing a user to indicate whether the displayed image or preview card is helpful to the user.
7. The method of claim 1 , wherein the new body of text is at least a portion of an email displayed within an email client.
8. A non-transitory, computer-readable medium containing instructions that, when executed by a hardware-based processor, performs stages for providing user-specific previews for terms within text, the stages comprising:
sending, from a user device to a management server, tracked user search behavior, wherein the user search behavior indicates a term selected by a user from of a body of text and entered into a search;
receiving, at the user device from the management server:
a representation of known terms previously searched by the user or frequently appearing to the user;
term categories that categorize terms selected by the user;
source categories that categorize sources of texts containing the terms selected by the user;
search-medium categories that categorize the search mediums utilized by the user to search the terms selected by the user; and
a machine-learning model trained at the management server using the tracked user search behavior;
parsing, by the user device, a new body of text viewed by the user;
identifying a candidate term within the new body of text by excluding stop words and terms matching the known terms in the stored representation;
predicting a search medium most likely to be used by the user to search for the candidate term, wherein the predicting is performed at the user device by implementing the machine-learning model using the term categories, source categories, and search-medium categories;
performing a backend search for the candidate term using a search medium selected based on the tracked user search behavior;
extracting information from the backend search and converting the extracted information into an image or preview card;
injecting the image or preview card into the candidate term within the new body of text; and
displaying the image or preview card for the candidate term upon the user selecting or hovering over the candidate term.
9. The non-transitory, computer-readable medium of claim 8 , wherein the machine-learning model outputs an identification of the search medium most likely to be utilized by the user to search the candidate term.
10. The non-transitory, computer-readable medium of claim 8 , the stages further comprising displaying multiple images or preview cards for the candidate term, wherein each of the multiple images or preview cards is retrieved from a different search medium.
11. The non-transitory, computer-readable medium of claim 8 , the stages further comprising determining that the tracked user search behavior indicates that the user searches for terms with at least a threshold frequency.
12. The non-transitory, computer-readable medium of claim 8 , wherein identifying a candidate term further comprises ranking a plurality of potential candidate terms and selecting the candidate term based on the ranking.
13. The non-transitory, computer-readable medium of claim 8 , wherein displaying the image or preview card further comprises displaying a feedback element allowing a user to indicate whether the displayed image or preview card is helpful to the user.
14. The non-transitory, computer-readable medium of claim 8 , wherein the new body of text is at least a portion of an email displayed within an email client.
15. A system for providing user-specific previews for terms within text, comprising:
a memory storage including a non-transitory, computer-readable medium comprising instructions; and
a computing device including a hardware-based processor that executes the instructions to carry out stages comprising:
sending, from the computing device to a management server, tracked user search behavior, wherein the user search behavior indicates a term selected by a user from of a body of text and entered into a search;
receiving, at the computing device from the management server:
a representation of known terms previously searched by the user or frequently appearing to the user;
term categories that categorize terms selected by the user;
source categories that categorize sources of texts containing the terms selected by the user;
search-medium categories that categorize the search mediums utilized by the user to search the terms selected by the user; and
a machine-learning model trained at the management server using the tracked user search behavior;
parsing, by the computing device, a new body of text viewed by the user;
identifying a candidate term within the new body of text by excluding stop words and terms matching the known terms in the stored representation;
predicting a search medium most likely to be used by the user to search for the candidate term, wherein the predicting is performed at the computing device by implementing the machine-learning model using the term categories, source categories, and search-medium categories;
performing a backend search for the candidate term using a search medium selected based on the tracked user search behavior;
extracting information from the backend search and converting the extracted information into an image or preview card;
injecting the image or preview card into the candidate term within the new body of text; and
displaying the image or preview card for the candidate term upon the user selecting or hovering over the candidate term.
16. The system of claim 15 , wherein the machine-learning model outputs an identification of the search medium most likely to be utilized by the user to search the candidate term.
17. The system of claim 15 , the stages further comprising displaying multiple images or preview cards for the candidate term, wherein each of the multiple images or preview cards is retrieved from a different search medium.
18. The system of claim 15 , the stages further comprising determining that the tracked user search behavior indicates that the user searches for terms with at least a threshold frequency.
19. The system of claim 15 , wherein identifying a candidate term further comprises ranking a plurality of potential candidate terms and selecting the candidate term based on the ranking.
20. The system of claim 15 , wherein displaying the image or preview card further comprises displaying a feedback element allowing a user to indicate whether the displayed image or preview card is helpful to the user.