Intuitive content search results suggestion system
System and methods for intuitive search operation results using machine learning including: identifying a first candidate content item matching a content item search request; identifying a first content item model corresponding to the first candidate content item including word vector collaborative filtering representations of the first candidate content item; identifying a set of content item models where each: is associated with at least one corresponding available content item, and includes word vector collaborative filtering representations; applying deep neural learning to compare the first content item model with the set of content item models to generate a subset of the content item models most relevant to the first content item model; generating a result set of available content items corresponding to the subset of the content item models most relevant to the first content item model; and providing the result set of available content items.
1. A system for intuitive search operation results using machine learning, the system comprising:
a computer processor;
a content comprehension engine executing on the computer processor and configured to:
receive a content item search request identifying a requested content item, wherein the requested content item has insufficient platform velocity to be classified as a known content item;
search a repository comprising a plurality of known content item models to determine that no known content item model comprising word vector collaborative filtering representations of user engagement with the requested content item is available to the content comprehension engine, wherein each of the plurality of known content item models has sufficient platform velocity to be classified as a known content item;
generate, based on determining that no known content item model is available, an unknown content item model corresponding to the requested content item using a set of input embeddings of an unknown content embedding space generated based on content metadata and contextual content data associated with the requested content item;
execute a deep neural machine learning model configured to project from the unknown content embedding space to a known content embedding space by:
providing the set of input embeddings as inputs to the deep neural machine learning model;
generating a set of output embeddings representing collaborative filtering representations of anticipated user engagement with the requested content item in the known content embedding space;
perform a relevance search using the set of output embeddings to identify a subset of the plurality of known content item models based on a match strength of each of the plurality of known content item models with the content item search request, wherein each of the subset of the plurality of known content item models:
is associated with at least one corresponding available content item, and
includes word vector collaborative filtering representations of sufficient real user engagement with the associated at least one corresponding available content item in the known content embedding space; and
provide a result set comprising the corresponding available content items of the identified subset of the plurality of known content item models in response to the content item search request, wherein the result set is displayed in a user interface element of a client device ordered according to the match strength of each of the subset of the plurality of known content item models.
2. The system of claim 1 , wherein performing the relevance search further comprises:
querying the repository comprising the plurality of known content item models;
ranking, based on the output embeddings, the plurality of known content item models by relevance using a ranking criteria; and
selecting the subset of the plurality of known content item models as being most relevant to the unavailable content item model based on the ranking.
3. The system of claim 2 , wherein the ranking criteria comprises at least one selected from a group consisting of relevancy, temporal relevancy, similarity, popularity, value, promotional status, requesting account demographics, requesting account preferences, and requesting account past behavior.
4. The system of claim 1 , wherein:
generating the result set comprises including content tier information corresponding to each available content item in the result set for use in arranging a display of the available content items according to content tier.
5. The system of claim 1 , wherein the content comprehension engine is further configured to:
utilize the output embeddings to determine a set of content tiers most relevant to the unknown content item model, wherein providing the result set comprises including the set of content tiers for display.
6. The system of claim 1 , wherein the result set of available content items comprises at least one content item of a different type from a type of the requested content item.
7. A method for intuitive search operation results using machine learning, the method comprising:
receiving a content item search request identifying a requested content item, wherein the requested content item has insufficient platform velocity to be classified as a known content item;
searching a repository comprising a plurality of known content item models to determine that no known content item model comprising word vector collaborative filtering representations of user engagement with the requested content item is available to the content comprehension engine, wherein each of the plurality of known content item models has sufficient platform velocity to be classified as a known content item;
generating, by at least one computer processor and based on determining that no known content item model is available, an unknown content item model corresponding to the requested content item using a set of input embeddings of an unknown content embedding space generated based on content metadata and contextual content data associated with the requested content item;
executing, by the at least one computer processor, a deep neural machine learning model configured to project from the unknown content embedding space to a known content embedding space by:
providing the set of input embeddings as inputs to the deep neural machine learning model;
generating a set of output embeddings representing collaborative filtering representations of anticipated user engagement with the requested content item in the known content embedding space;
performing a relevance search using the set of output embeddings to identify a subset of the plurality of known content item models based on a match strength of each of the plurality of known content item models with the content item search request, wherein each of the subset of the plurality of known content item models:
is associated with at least one corresponding available content item, and
includes word vector collaborative filtering representations of sufficient real user engagement with the associated at least one corresponding available content item in the known content embedding space; and
providing a result set comprising the corresponding available content items of the identified subset of the plurality of known content item models in response to the content item search request, wherein the result set is displayed in a user interface element of a client device ordered according to the match strength of each of the subset of the plurality of known content item models.
8. The method of claim 7 , wherein performing the relevance search further comprises:
querying the repository comprising the plurality of known content item models;
ranking, based on the output embeddings, the plurality of known content item models by relevance using a ranking criteria; and
selecting the subset of the plurality of known content item models as being most relevant to the unavailable content item model based on the ranking.
9. The method of claim 8 , wherein the ranking criteria comprises at least one selected from a group consisting of relevancy, temporal relevancy, similarity, popularity, value, promotional status, requesting account demographics, requesting account preferences, and requesting account past behavior.
10. The method of claim 7 , wherein:
generating the result set comprises including content tier information corresponding to each available content item in the result set for use in arranging a display of the available content items according to content tier.
11. The method of claim 7 , further comprising:
utilizing the output embeddings to determine a set of content tiers most relevant to the unknown content item model, wherein providing the result set comprises including the set of content tiers for display.
12. The method of claim 7 , wherein the result set of available content items comprises at least one content item of a different type from a type of the requested content item.
13. A non-transitory computer-readable storage medium comprising a plurality of instructions for intuitive search operation results using machine learning, the plurality of instructions configured to execute on at least one computer processor to enable the at least one computer processor to:
receive a content item search request identifying a requested content item, wherein the requested content item has insufficient platform velocity to be classified as a known content item;
search a repository comprising a plurality of known content item models to determine that no known content item model comprising word vector collaborative filtering representations of user engagement with the requested content item is available to the content comprehension engine, wherein each of the plurality of known content item models has sufficient platform velocity to be classified as a known content item;
generate, based on determining that no known content item model is available, an unknown content item model corresponding to the requested content item using a set of input embeddings of a unknown content embedding space generated based on content metadata and contextual content data associated with the requested content item;
execute a deep neural machine learning model configured to project from the unknown content embedding space to a known content embedding space by:
providing the set of input embeddings as inputs to the deep neural machine learning model;
generating a set of output embeddings representing collaborative filtering representations of anticipated user engagement with the requested content item in the known content embedding space;
perform a relevance search using the set of output embeddings to identify a subset of the plurality of known content item models based on a match strength of each of the plurality of known content item models with the content item search request, wherein each of the subset of the plurality of known content item models:
is associated with at least one corresponding available content item, and
includes word vector collaborative filtering representations of sufficient real user engagement with the associated at least one corresponding available content item in the known content embedding space; and
provide a result set comprising the corresponding available content items of the identified subset of the plurality of known content item models in response to the content item search request, wherein the result set is displayed in a user interface element of a client device ordered according to the match strength of each of the subset of the plurality of known content item models.
14. The non-transitory computer-readable storage medium of claim 13 , wherein performing the relevance search further comprises:
querying the repository comprising the plurality of known content item models;
ranking, based on the output embeddings, the plurality of known content item models by relevance using a ranking criteria; and
selecting the subset of the plurality of known content item models as being most relevant to the unavailable content item model based on the ranking.
15. The system of claim 1 , wherein content comprehension engine is further configured to:
identify, after providing the result set, real user engagement with the requested content item;
determine that the sufficient platform velocity of user engagement is met based on the real user engagement;
generate a set of warm-start characteristics using the real user engagement;
receive a second content item search request identifying the requested content item; and
provide (i) the set of warm-start characteristics and (ii) the set of input embeddings as inputs to a second execution of the deep neural machine learning model to generate a second set of output embeddings, wherein the second set of output embeddings are used to perform a second relevance search.
16. The method of claim 7 , further comprising:
identifying, after providing the result set, real user engagement with the requested content item;
determining that the sufficient platform velocity of user engagement is met based on the real user engagement;
generating a set of warm-start characteristics using the real user engagement;
receiving a second content item search request identifying the requested content item; and
providing (i) the set of warm-start characteristics and (ii) the set of input embeddings as inputs to a second execution of the deep neural machine learning model to generate a second set of output embeddings, wherein the second set of output embeddings are used to perform a second relevance search.
17. The non-transitory computer-readable storage medium of claim 13 , wherein the plurality of instructions are further configured to execute on the at least one computer processor to enable the at least one computer processor to:
identify, after providing the result set, real user engagement with the requested content item;
determine that the sufficient platform velocity of user engagement is met based on the real user engagement;
generate a set of warm-start characteristics using the real user engagement; and
receive a second content item search request identifying the requested content item;
provide (i) the set of warm-start characteristics and (ii) the set of input embeddings as inputs to a second execution of the deep neural machine learning model to generate a second set of output embeddings, wherein the second set of output embeddings are used to perform a second relevance search.