Mobile content attribute recommendation engine
At least one analytical agent extracts a plurality of attributes from each of a plurality of member input vectors. Each member input vector includes raw data characterizing contextual aspects about an associated and different user. Thereafter, a content search vector is generated for each user by the at least one analytical agent that includes the attributes extracted from the member input vector associated with such user and weights corresponding to each attribute that are particular to such user. A search engine, accessing a content library, then matches each content search vector with one of a plurality of content workflows based on both the attributes and weights within such content search vector. A context engine then initiates execution of each matching content workflow which results in tailored messages specified by the matching content workflow being sent to the user associated with the matching content workflow.
1. A method for implementation by one or more data processors forming part of at least one computing device, the method comprising:
extracting, by at least one analytical agent, a plurality of attributes from each of a plurality of member input vectors, each member input vector comprising raw data characterizing contextual aspects about an associated and different user;
generating, for each user by the at least one analytical agent, a content search vector comprising the attributes extracted from the member input vector associated with such user and weights corresponding to each attribute that are particular to such user;
matching, by a search engine accessing a content library, each content search vector with one of a plurality of content workflows based on both the attributes and weights within such content search vector; and
initiating, by a context engine, execution of each matching content workflow which results in tailored messages specified by the matching content workflow being sent to the user associated with the matching content workflow;
wherein:
each content workflow specifies a different sequence, modality, and delivery flow of the tailored messages;
the modality of delivery of the tailored messages includes at least one of: short messaging service, multimedia messaging service, application notification, or e-mail message.
2. The method of claim 1 , further comprising generating the member input vector for each user.
3. The method of claim 2 , wherein attributes of the member input vector are subject to change and the extracting, generating, matching, and initiating are updated to reflect changes in the member input vector.
4. The method of claim 3 , wherein a dimensionality of the member input vector is automatically expanded upon addition of one or more data sources without refactoring other data sources.
5. The method of claim 1 , further comprising:
receiving, for each user, a plurality of responses to tailored messages previously sent to such user;
performing computer-implemented natural language processing on the plurality of responses to generate at least a portion of the attributes of each member input vector.
6. The method of claim 1 , wherein there are a plurality of analytical agents and at least one analytical agent uses an output of at least one other analytical agent in connection with the extracting and generating.
7. The method of claim 1 , wherein there are a plurality of analytical agents that each evaluate only a subset of dimensions of the member input vector and which generate only a different subset of the attributes.
8. The method of claim 1 , wherein the at least one analytical agent comprises a natural language processing agent to extract key topics from a user-generated response and which uses a machine learning model.
9. The method of claim 1 , wherein the at least one analytical agent comprises a mapping agent that runs data mapping rules to map data falling within a range into attribute.
10. The method of claim 1 , wherein the at least one analytical agent comprises a classification agent using random forests to classify continuous feature vectors with a finite set of classes.
11. The method of claim 1 , wherein the at least one analytical agent comprises an emotion recognition agent that takes individual messages and generates an emotional profile of the messages.
12. The method of claim 1 , wherein the at least one analytical agent comprises a psychographic monitoring agent that translates user generated self-reports or provided outcomes obtained from external sources to generate a psychographic profile for a user.
13. The method of claim 1 further comprising:
generating a forward index to store a list corresponding to all of the attributes;
inverting the forward index to result in an inverted index, the inverted index being used by the search engine to match attributes of the content search vector with attributes of the content workflows.
14. The method of claim 1 further comprising:
generating an activation score for each attribute in each content search vector;
wherein the matching by the search engine utilizes the generated activation scores to identify a best matching content workflow.
15. The method of claim 14 , wherein the tailored messages pertain to a healthcare and/or wellness regimen.
16. The method of claim 15 , wherein the activation score is based on a self efficacy score derived from responses by the respective user of self-efficacy assessments.
17. The method of claim 15 , wherein the activation score is based on a state of change of the respective user.
18. The method of claim 15 , wherein the activation score is based on a current internal change of the respective user that characterizes actions performed by the respective user in relation to overall behavior change goals.
19. The method of claim 15 , wherein the activation score is based on a current behavior change of the respective user that characterizes actions performed by the respective user in relation to overall behavior change goals.
20. The method of claim 15 , wherein the activation score is based on a current engagement level of the respective user characterizing how the user responds to the tailored messages.
21. The method of claim 14 , wherein each content workflow comprises a content matching algorithm that uses a content search vector that has weights that are determined by a genetic algorithm wherein a fitness measure for optimization is the resulting activation score.
22. The method of claim 1 further comprising:
determining, for each content search vector, a matching attribute score between the content search vector and each of a plurality of content workflows;
wherein the matching by the search engine is based on the content workflow having a highest matching attribute scores relative to the corresponding content search vector.
23. The method of claim 1 , wherein each content workflow comprises a plurality of messages that are each tagged with respective attributes characterizing content of such messages.
24. A system comprising:
at least one programmable data processor; and
memory storing instructions which, when executed by the at least one programmable data processor, result in operations comprising:
extracting, by at least one analytical agent, a plurality of attributes from each of a plurality of member input vectors, each member input vector comprising raw data characterizing contextual aspects about an associated and different user in connection with a healthcare or wellness regimen;
generating, for each user by the at least one analytical agent, a content search vector comprising the attributes extracted from the member input vector associated with such user and weights corresponding to each attribute that are particular to such user;
matching, by a search engine accessing a content library, each content search vector with one of a plurality of content workflows based on both the attributes and weights within such content search vector, the content workflows specifying messaging sequences to send to the corresponding user in order to adhere to the healthcare or wellness regimen; and
initiating, by a context engine, execution of each matching content workflow which results in tailored messages specified by the matching content workflow being sent to the user associated with the matching content workflow.
25. A method for implementation by one or more data processors forming part of at least one computing device, the method comprising:
extracting, by at least one analytical agent, a plurality of attributes from each of a plurality of member input vectors, each member input vector comprising raw data characterizing contextual aspects about an associated and different user in connection with a healthcare or wellness regimen;
generating, for each user by the at least one analytical agent, a content search vector comprising the attributes extracted from the member input vector associated with such user and weights corresponding to each attribute that are particular to such user;
matching, by a search engine accessing a content library, each content search vector with one of a plurality of content workflows based on both the attributes and weights within such content search vector, the content workflows specifying messaging sequences to send to the corresponding user in order to adhere to the healthcare or wellness regimen; and
initiating, by a context engine, execution of each matching content workflow which results in tailored messages specified by the matching content workflow being sent to the user associated with the matching content workflow;
wherein:
the weighting for the weights within the content search vector are determined by a genetic algorithm in which an activation score can be used as a fitness measure;
the matching content workflow is selected to maximize a value of the activation score;
each content workflow specifies a different sequence, modality, and delivery flow of the tailored messages;
the modality of delivery of the tailored messages includes at least one of: short messaging service, multimedia messaging service, application notification, or e-mail message.