Recommending contents using a base profile
A method and system for recommending content to a user whose interest(s) has not been identified is disclosed. A base user profile may be created for association with the user. The base user profile may be created by generating a list of ranked interests of a set of representative users. The list of ranked interests may be generated based on activity information obtained for the set of representative users. Content may be recommended to the user based on the base user profile.
1. A method for recommending content to a user, the method being implemented on a computing device having at least one processor, storage, and a communication interface connected to a network, the method comprising:
creating a base user profile for a user in an event that interest information identifying one or more interests of the user is not available, the base user profile including information indicating a plurality of ranked interests of a set of other users, wherein creating the base user profile comprises:
selecting the set of other users based on selection criteria which include a frequency with which each of the other users interacts with a content recommendation system, wherein each of the selected set of other users is assigned a bias factor determined based on an amount of time since a user account was created for the selected other user with the content recommendation system;
obtaining activity information of the selected set of other users, the activity information indicating activities engaged by the other users; and
analyzing the obtained activity information of the selected set of other users to determine the plurality of ranked interests of the set of other users based on weights associated with each of the selected set of other users, wherein the weights are determined based on the bias factor of each of the selected set of other users, features extracted from content items with which the activity information is associated, and user engagement information associated with the selected set of other users; and
recommending content to the user based on the base user profile.
2. The method of claim 1 , further comprising:
building a content pool comprising a plurality of content items, wherein each of the plurality of content items is selected for inclusion in the content pool based on a measure of its popularity with respect to one or more of the other users; and
retrieving a set of candidate content items from the content pool based on scores associated with the plurality of ranked interests and on a relevancy score associated with each of one or more key terms extracted from each of the candidate content items, wherein:
the content recommended to the user comprises the retrieved set of candidate content items, and
the content comprises the content items with which the activity information is associated.
3. The method of claim 2 , wherein the retrieval of the set of candidate content items is further based on discovered interests of the user.
4. The method of claim 2 , further comprising:
dynamically determining a pace at which content in the content pool is updated based on at least one of entered search terms, declared interests, or user activity with respect to content.
5. The method of claim 2 , further comprising:
updating the content pool by purging content items based on a performance characteristic of the to-be-purged content items or on a time stamp associated with the purged content item satisfying a criterion.
6. The method of claim 2 , further comprising:
responsive to a new content source becoming available for crawling, updating the content pool.
7. The method of claim 2 , wherein the content pool is organized as a tiered system by comprising both a general content pool and a personalized content pool for each of different users, and wherein each personalized content pool comprises links, pointers, or indices that provide references to where a given content item is stored in the general content pool.
8. The method of claim 2 , wherein the set of candidate content items are retrieved from the content pool based on a dot product of the scores associated with the plurality of ranked interests with a relevancy score associated with each of the one or more extracted key terms.
9. The method of claim 1 , wherein the user engagement information comprises user metrics including at least one of user experience, activity level, user reputation, user power, user ranking, user age, or user education level.
10. The method of claim 1 , wherein the set of other users are selected from a plurality of other users that interact with the content recommendation system based on the frequency with which each of the other users interact with the content recommendation system meeting a threshold log-in frequency of the content recommendation system.
11. A non-transitory, computer-readable medium having information recorded thereon for recommending content to a user wherein the information, when read by at least one processor, effectuate operations comprising:
creating a base user profile for a user in an event that interest information identifying one or more interests of the user is not available, the base user profile including information indicating a plurality of ranked interests of a set of other users, wherein creating the base user profile comprises:
selecting the set of other users based on selection criteria which include a frequency with which each of the other users interacts with a content recommendation system, wherein each of the selected set of other users is assigned a bias factor determined based on an amount of time since a user account was created for the selected other user with the content recommendation system;
obtaining activity information of the selected set of other users, the activity information indicating activities engaged by the other users; and
analyzing the obtained activity information of the selected set of other users to determine the plurality of ranked interests of the set of other users, based on weights associated with each of the selected set of other users, wherein the weights are determined based on the bias factor of each of the selected set of other users, features extracted from content items with which the activity information is associated, and user engagement information associated with the selected set of other users; and
recommending content to the user based on the base user profile.
12. A system for recommending content to a user, the system comprising:
storage;
a communication platform connected to a network; and
one or more processors programmed to execute computer program instructions that, when executed, cause the one or more processors to:
create a base user profile for a user in an event that interest information identifying one or more interests of the user is not available, the base user profile including information indicating a plurality of ranked interests of a set of other users, wherein creating the base user profile comprises:
selecting the set of other users based on selection criteria which include a frequency with which each of the other users interacts with a content recommendation system, wherein each of the selected set of other users is assigned a bias factor determined based on an amount of time since a user account was created for the selected other user with the content recommendation system;
obtaining activity information of the selected set of other users, the activity information indicating activities engaged by the other users; and
analyzing the obtained activity information of the selected set of other users to determine the plurality of ranked interests of the set of other users, based on weights associated with each of the selected set of other users, wherein the weights are determined based on the bias factor of each of the selected set of other users, features extracted from content items with which the activity information is associated, and user engagement information associated with the selected set of other users; and
recommend content to the user based on the base user profile.
13. The system of claim 12 , wherein the one or more processors are further caused to:
build a content pool comprising a plurality of content items, wherein each of the plurality of content items is selected for inclusion in the content pool based on a measure of its popularity with respect to one or more of the other users; and
retrieve a set of candidate content items from the content pool based on scores associated with the plurality of ranked interests and on a relevancy score associated with each of one or more key terms extracted from each of the candidate content items, wherein:
the content recommended to the user comprises the retrieved set of candidate content items, and
the content comprises the content items with which the activity information is associated.
14. The system of claim 13 , wherein the retrieval of the set of candidate content items is further based on discovered interests of the user.
15. The system of claim 13 , wherein the one or more processors are further caused to:
dynamically determine a pace at which content in the content pool is updated based on at least one of entered search terms, declared interests, or user activity with respect to content.
16. The system of claim 13 , wherein the one or more processors are further caused to:
update the content pool by purging content items based on a performance characteristic of the to-be-purged content items or on a time stamp associated with the purged content item satisfying a criterion.
17. The system of claim 13 , wherein the one or more processors are further caused to:
responsive to a new content source becoming available for crawling, update the content pool.
18. The system of claim 13 , wherein the content pool is organized as a tiered system by comprising both a general content pool and a personalized content pool for each of different users, and wherein each personalized content pool comprises links, pointers, or indices that provide references to where a given content item is stored in the general content pool.
19. The system of claim 13 , wherein the set of candidate content items are retrieved from the content pool based on a dot product of the scores associated with the plurality of ranked interests with a relevancy score associated with each of the one or more extracted key terms.