Learning personalized actionable domain models

1. A method for learning personalized actionable domain models by a processor, comprising:

generating a domain model according to data representative of a plurality of actions, extracted from one or more online data sources, of a plurality of cluster representatives, wherein the plurality of actions achieve a goal;

dividing the data extracted from the one or more online data sources into one or more training datasets and one or more testing datasets;

training, in a machine learning operation, a hierarchical action model using the one or more training datasets, wherein the training includes identifying probabilities of relationships established between those of the plurality of actions that are interdependent and transitions between those of the plurality of actions;

generating the hierarchical action model based on the probabilities of the domain model and the plurality of actions, wherein the hierarchical action model comprises a sequence of actions of the plurality of actions for achieving the goal;

filtering the hierarchical action model to a selected set of actions according to weighted actions of the plurality of actions; and

validating the filtered hierarchical action model, in the machine learning operation, using the one or more testing datasets.

2. The method of claim 1 , wherein the generating the domain model further includes grouping one or more actions from the plurality of actions into a cluster based on similarity of the plurality of cluster representatives.

3. The method of claim 1 , further including:

analyzing semantic text of the one or more online data sources using text analysis; and

extracting the plurality of actions from the analyzed semantic text of the one or more online data sources.

4. The method of claim 1 , further including determining a social media distance in social media networks between each one of the plurality of cluster representatives, wherein the one or more online data sources include the social media networks.

5. The method of claim 4 , further including assigning a weighted value to the plurality of actions according to the social media distance of each one of the plurality of cluster representatives.

6. The method of claim 4 , further including:

dividing those of the plurality of cluster representatives into a first tier having the social media distance less than a defined social media distance threshold to the one of the plurality of cluster representatives;

dividing those of the plurality of cluster representatives into a second tier having the social media distance greater than a defined social media distance threshold to the one of the plurality of cluster representatives; and

ranking the weighted actions of the plurality of actions of those in the first tier greater than the weighted actions of the plurality of actions of those in the second tier.

7. The method of claim 1 , further including:

learning a plurality of preferences in social media networks and the plurality of actions relating to each one of the plurality of cluster representatives; and

updating probabilities of the hierarchical action model according to the plurality of preferences in social media networks and the plurality of actions.

8. The method of claim 1 , further including:

determining a quality of the hierarchical action model; and

determining an efficiency of the hierarchical action model, wherein the hierarchical action model is probabilistic.

9. A system for learning actionable domain models, comprising:

one or more computers with executable instructions that when executed cause the system to:

generate a domain model according to data representative of a plurality of actions, extracted from one or more online data sources, of a plurality of cluster representatives, wherein the plurality of actions achieve a goal;

divide the data extracted from the one or more online data sources into one or more training datasets and one or more testing datasets;

train, in a machine learning operation, a hierarchical action model using the one or more training datasets, wherein the training includes identifying probabilities of relationships established between those of the plurality of actions that are interdependent and transitions between those of the plurality of actions;

generate the hierarchical action model based on the probabilities of the domain model and the plurality of actions, wherein the hierarchical action model comprises a sequence of actions of the plurality of actions for achieving the goal;

filter the hierarchical action model to a selected set of actions according to weighted actions of the plurality of actions; and

validate the filtered hierarchical action model, in the machine learning operation, using the one or more testing datasets.

10. The system of claim 9 , wherein the generating the domain model further includes grouping one or more actions from the plurality of actions into a cluster based on similarity of the plurality of cluster representatives.

11. The system of claim 9 , wherein the executable instructions:

analyze semantic text of the one or more online data sources using text analysis; and

extract the plurality of actions from the analyzed semantic text of the one or more online data sources.

12. The system of claim 9 , wherein the executable instructions determine a social media distance in social media networks between each one of the plurality of cluster representatives, wherein the one or more online data sources include the social media networks.

13. The system of claim 12 , wherein the executable instructions assign a weighted value to the plurality of actions according to the social media distance of each one of the plurality of cluster representatives.

14. The system of claim 12 , wherein the executable instructions:

divide those of the plurality of cluster representatives into a first tier having the social media distance less than a defined social media distance threshold to the one of the plurality of cluster representatives;

divide those of the plurality of cluster representatives into a second tier having the social media distance greater than a defined social media distance threshold to the one of the plurality of cluster representatives; and

rank the weighted actions of the plurality of actions of those in the first tier greater than the weighted actions of the plurality of actions of those in the second tier.

15. The system of claim 9 , wherein the executable instructions:

learn a plurality of preferences in social media networks and the plurality of actions relating to each one of the plurality of cluster representatives;

update probabilities of the hierarchical action model according to the plurality of preferences in social media networks and the plurality of actions;

determine a quality of the hierarchical action model; and

determine an efficiency of the hierarchical action model, wherein the hierarchical action model is probabilistic.

16. A computer program product for, by a processor, learning actionable domain models, the computer program product comprising a non-transitory computer-readable storage medium having computer-readable program code portions stored therein, the computer-readable program code portions comprising:

an executable portion that generates a domain model according to data representative of a plurality of actions, extracted from one or more online data sources, of a plurality of cluster representatives, wherein the plurality of actions achieve a goal;

an executable portion that divides the data extracted from the one or more online data sources into one or more training datasets and one or more testing datasets;

an executable portion that trains, in a machine learning operation, a hierarchical action model using the one or more training datasets, wherein the training includes identifying probabilities of relationships established between those of the plurality of actions that are interdependent and transitions between those of the plurality of actions;

an executable portion that generates the hierarchical action model based on the probabilities of the domain model and the plurality of actions, wherein the hierarchical action model comprises a sequence of actions of the plurality of actions for achieving the goal;

an executable portion that filters the hierarchical action model to a selected set of actions according to weighted actions of the plurality of actions; and

an executable portion that validates the filtered hierarchical action model, in the machine learning operation, using the one or more testing datasets.

17. The computer program product of claim 16 , wherein the generating the domain model further includes grouping one or more actions from the plurality of actions into a cluster based on similarity of the plurality of cluster representatives.

18. The computer program product of claim 16 , further including an executable portion that:

analyzes semantic text of the one or more online data sources using text analysis; and

extracts the plurality of actions from the analyzed semantic text of the one or more online data sources.

19. The computer program product of claim 16 , further including an executable portion that:

determines a social media distance in social media networks between each one of the plurality of cluster representatives, wherein the one or more online data sources include the social media networks;

assigns a weighted value to the plurality of actions according to the social media distance of each one of the plurality of cluster representatives;

divides those of the plurality of cluster representatives into a first tier having the social media distance less than a defined social media distance threshold to the one of the plurality of cluster representatives;

divides those of the plurality of cluster representatives into a second tier having the social media distance greater than a defined social media distance threshold to the one of the plurality of cluster representatives; and

ranks the weighted actions of the plurality of actions of those in the first tier greater than the weighted actions of the plurality of actions of those in the second tier.

20. The computer program product of claim 16 , further including an executable portion that:

learns a plurality of preferences in social media networks and the plurality of actions relating to each one of the plurality of cluster representatives;

updates probabilities of the hierarchical action model according to the plurality of preferences in social media networks and the plurality of actions;

determines a quality of the hierarchical action model; and

determines an efficiency of the hierarchical action model, wherein the hierarchical action model is probabilistic.