Artificial intelligence and/or machine learning systems and methods for evaluating audiences in an embedding space based on keywords
In some embodiments, a method includes determining a position for a search query and a position for each audience record from multiple audience records in an embedding space. The method further includes receiving multiple device records, each associated with an audience record. The method further includes determining multiple keywords, each associated with an audience record and determining a position for each keyword in the embedding space. The method further includes calculating a first distance between the position of the search query in the embedding space and the position of each audience record in the embedding space. The method further includes calculating a second distance between the position of the search query in the embedding space and the position of each keyword in the embedding space. The method further includes ranking each audience record based on the first distance and the second distance.
1. A non-transitory, processor-readable medium storing code representing instructions to be executed by a processor, the code comprising code to cause the processor to:
receive an indication of an audience;
access at least one device record associated with the audience;
identify a keyword associated with the audience based on the at least one device record;
determine a position of the audience in an embedding space based on the keyword, the embedding space generated based on a natural language corpus;
receive a search query associated with an item of targeted content;
determine a position of the search query in the embedding space;
calculate a distance between the position of the audience and the position of the search query in the embedding space; and
send a signal to cause delivery of the item of targeted content to the audience based on the distance.
2. The non-transitory, processor-readable medium of claim 1 , wherein the audience includes one audience device.
3. The non-transitory, processor-readable medium of claim 1 , wherein the audience includes a plurality of audience devices.
4. The non-transitory, processor-readable medium of claim 1 , wherein the device record includes an indication of one or more websites visited by an audience device belonging to the audience.
5. The non-transitory, processor-readable medium of claim 1 , wherein the keyword is associated with at least one website represented in the at least one device record.
6. The non-transitory, processor-readable medium of claim 1 , wherein the keyword is scraped from a website represented in the at least one device record.
7. The non-transitory, processor-readable medium of claim 1 , wherein the keyword is at least one of a summary characteristic or a descriptive title for the audience.
8. The non-transitory, processor-readable medium of claim 1 , the code further comprising code to cause the processor to generate the embedding space.
9. The non-transitory, processor-readable medium of claim 1 , wherein:
the audience is a first audience from a plurality of audiences;
the code to cause the processor to identify the keyword for the first audience includes code to cause the processor to identify a keyword for each audience from the plurality of audiences;
the code to cause the processor to determine the position of the first audience includes code to cause the processor to determine a position of each audience from the plurality of audiences in the embedding space based on the keyword for that audience;
the code to cause the processor to calculate the distance between the position of the first audience and the search query includes code to cause the processor to calculate, for each audience from the plurality of audiences, a distance between the position of that audience and the position of the search query in the embedding space; and
rank at least a subset of audiences from the plurality of audiences based on the distance between the position of each audience from the subset of audiences and the position of the search query in the embedding space, the first audience being highest ranked based on a shortest distance between the position of first audience and the position of the search query in the embedding space.
10. The non-transitory, processor-readable medium of claim 1 , wherein:
the audience is a first audience from a plurality of audiences; and
the code to cause the processor to identify the keyword for the first audience includes code to cause the processor to identify, for each audience from the plurality of audiences, a keyword associated with that audience based on content of a website disproportionately visited by at least one audience device associated with that audience,
the position of each audience from the plurality of audiences in the embedding space being based on the keyword associated with that audience.
11. The non-transitory, processor-readable medium of claim 1 , wherein:
the audience is associated with a plurality of audience devices; and
the at least one device record includes aggregated web browsing history data for the plurality of audience devices.
12. The non-transitory, processor-readable medium of claim 1 , wherein the distance between the position of the audience and the position of the search query in the embedding space represents a quantitative semantic similarity between the keyword and the search query.
13. The non-transitory, processor-readable medium of claim 1 , wherein:
the at least one device record includes an indication of a website visited by the audience;
the keyword is from a plurality of keywords associated with the website; and
the position of the audience in the embedding space is determined based on the plurality of keywords.
14. A non-transitory, processor-readable medium storing code representing instructions to be executed by a processor, the code comprising code to cause the processor to:
receive web browsing history data associated with an audience;
identify a plurality of keywords associated with the audience, each keyword from the plurality of keywords associated with a website indicated in the web browsing history data;
determine a position of the audience in an embedding space based on the plurality of keywords, the embedding space generated based on a natural language corpus;
receive a search query associated with an item of targeted content;
determine a position of the search query in the embedding space;
calculate a distance between the position of the audience and the position of the search query in the embedding space.
15. The non-transitory, processor-readable medium of claim 14 , wherein:
determining the position of the audience includes determining a plurality of positions of the audience in the embedding space, each position of the audience associated with a keyword from the plurality of keywords; and
calculating the distance between the position of the audience and the position of the search query in the embedding space includes calculating a weighted average of distances between the search query and the plurality of positions of the audience in the embedding space.
16. The non-transitory, processor-readable medium of claim 14 , wherein the plurality of keywords is scraped from the website.
17. The non-transitory, processor-readable medium of claim 14 , wherein a keyword from the plurality of keywords is obtained from metadata associated with the website.
18. The non-transitory, processor-readable medium of claim 14 , wherein the web browsing history data is for a plurality of audience devices associated with the audience.
19. The non-transitory, processor-readable medium of claim 14 , wherein the web browsing history data is aggregated and for a plurality of audience devices associated with the audience.
20. The non-transitory, processor-readable medium of claim 14 , the code further comprising code to cause the processor to facilitate delivery of the item of targeted content to the audience based on the distance.
21. The non-transitory, processor-readable medium of claim 14 , the code further comprising code to cause the processor to send a signal to cause the item of targeted content to be delivered to the audience based on the distance.
22. A computer-implemented method, comprising:
accessing an embedding space generated based on a natural language corpus;
receiving an indication of an audience;
accessing at least one device record associated with the audience;
identifying a keyword associated with the audience based on the at least one device record;
determining a position of the audience in the embedding space based on the keyword;
receiving a search query associated with an item of targeted content;
determining a position of the search query in the embedding space;
calculating a distance between the position of the audience and the position of the search query in the embedding space; and
facilitating delivery of the item of targeted content to the audience based on the distance.