Baseline interest profile for recommendations using a geographic location
Software for a content-aggregation website generates a first representation of interests for a geographical location. The representation includes a plurality of entities that are derived from a corpus of documents. Each of the plurality of entities is associated with an expected value that is based on engagement signals from users in the geographical location and that is weighted using a sparse-polarity approach to be discriminative with respect to other entities. Each of the ingested articles is represented by the second representation that associates an aboutness score with each of the plurality of entities. The software uses the first representation, a similarity measure, and a second representation to create rankings of a plurality of ingested articles received. Then the software receives a request for access to the content-aggregation service from a new user from the geographical location and serves the new or infrequent user a content stream based on the rankings.
1. A method, comprising:
generating a first representation of interests for a geographical location, wherein the representation includes a plurality of entities that are derived from at least one corpus of documents and wherein each of the plurality of entities is associated with an expected value that is based at least in part on engagement signals received by a content-aggregation service from users in the geographical location and that is weighted using a sparse-polarity approach to be discriminative with respect to other entities, wherein each of said entities are labels derived from the at least one corpus;
using the first representation, a similarity measure, and a second representation to create rankings of a plurality of ingested articles, wherein each of the ingested articles is represented by the second representation that associates an about score from each of the plurality of entities;
receiving a request for access to the content-aggregation service from a new or infrequent user from the geographical location; and
serving the new or infrequent user a content stream based at least in part on the rankings, wherein each operation of the method is executed by one or more processors.
2. The method of claim 1 , wherein the geographical location is a city.
3. The method of claim 1 , wherein the first representation and the second representation are vectors and the similarity measure is cosine similarity.
4. The method of claim 1 , further comprising an operation of updating the plurality of entities with new signals in real-time or near-real-time, using stream processing.
5. The method of claim 1 , wherein the labels are nodes in a taxonomy created at least in part from documents previously ingested by the content-aggregation service.
6. The method of claim 1 , wherein the expected value reflects a conditional probability of a positive user engagement with the entity, given the geographical location.
7. The method of claim 1 , wherein the geolocation of the new or infrequent user is determined from an internet-protocol (IP) address associated with the new or infrequent user.
8. The method of claim 1 , wherein the sparse-polarity approach involves application of a threshold to a z-statistic.
9. One or more computer-readable media that are non-transitory and that store a program, wherein the program, when executed, instructs a processor to perform the following operations:
generate a first representation of interests for a geographical location, wherein the representation includes a plurality of entities that are derived from at least one corpus of documents and wherein each of the plurality of entities is associated with a probability score that is based at least in part on engagement signals received from users in the geographical location and that is weighted using a sparse-polarity approach to be discriminative with respect to other entities, wherein each of said entities are labels derived from the at least one corpus;
use the first representation, a similarity measure, and a second representation to create rankings of a plurality of ingested articles received by the content-aggregation service, wherein each of the ingested articles is represented by the second representation that associates an about score from each of the plurality of entities;
receive a request for access to the content-aggregation service from a new or infrequent user from the geographical location; and
serve the new or infrequent user a content stream based at least in part on the rankings.
10. The computer-readable media of claim 9 , wherein the geographical location is a city.
11. The computer-readable media of claim 9 , wherein the first representation and the second representation are vectors and the similarity measure is cosine similarity.
12. The computer-readable media of claim 9 , further comprising an operation of updating the plurality of entities with new signals in real-time or near-real-time, using stream processing.
13. The computer-readable media of claim 9 , wherein the labels are nodes in a taxonomy created at least in part from documents previously, ingested by the content-aggregation service.
14. The computer-readable media of claim 9 , wherein the expected value reflects a conditional probability of a positive user engagement with the entity, given the geographical location.
15. The computer-readable media of claim 9 , wherein the geolocation of the new or infrequent user is determined from an internet-protocol (IP) address associated with the new or infrequent user.
16. The computer-readable media of claim 9 , wherein the sparse-polarity approach involves application of a threshold to a z-statistic.
17. A method, comprising the operations of:
generating a first representation of interests for a city, wherein the representation includes a plurality of entities that are derived from at least one corpus of documents and wherein each of the plurality of entities is associated with an expected value that is based at least in part on engagement signals received by a content-aggregation service from users in the city and that is weighted using a sparse-polarity approach to be discriminative with respect to other entities, wherein each of said entities are labels derived from the at least one corpus;
using the first representation, a similarity measure, and a second representation to create rankings of a plurality of ingested articles, wherein each of the ingested articles is represented by the second representation that associates an about score from each of the plurality of entities and wherein the first representation and the second representation are vectors and the similarity measure is cosine similarity;
receiving a request for access to the content-aggregation service from a new or infrequent user from the city; and
serving a user a content stream based at least in part on the rankings, wherein each operation of the method is executed by one or more processors.
18. The method of claim 17 , wherein the sparse-polarity approach involves application of a threshold to a z-statistic.