Prediction and automatic performance of computer-related activities
A method comprises constructing, for a specific activity, a decision model based on action data of one or more actions, each action having an identifier of an activity performed with a computer application, use data being operated on by the computer application, and a timestamp, the decision model including a rule specifying a first activity as a next activity or a probabilistic classifier that accepts use data being operated on via the specific activity and outputs an identifier of an activity to be performed as a next activity with a probability. The method comprises detecting performance of the specific activity, when the decision model for the specific activity includes the rule, automatically applying the rule, and when the decision model for the specific activity includes the probabilistic classifier and no applicable rule: executing the probabilistic classifier to obtain a list of candidate next activities and an associated list of probabilities.
1. A computer-implemented method of predicting and automatically performing activities related to computer applications, comprising:
constructing, for a specific activity of a plurality of activities, a decision model based on action data of one or more actions,
each action of the one or more actions having an identifier of an activity of the plurality of activities performed with a computer application, use data being operated on by the computer application via the activity, and a timestamp,
the decision model including a rule specifying a first activity as a next activity or a probabilistic classifier that accepts use data being operated on via the specific activity and outputs an identifier of an activity to be performed as a next activity with a probability;
detecting, in real time, performance of the specific activity with specific use data;
when the decision model for the specific activity includes the rule, automatically applying the rule; and
when the decision model for the specific activity includes the probabilistic classifier and no applicable rule:
executing the probabilistic classifier to obtain a list of candidate next activities and an associated list of probabilities; and
sending the list of candidate next activities, indicating a recommended next activity, and the associated list of probabilities to a device,
wherein one or more steps is performed using a computer.
2. The computer-implemented method of claim 1 , further comprising:
receiving event data indicating one or more use events related to user interactions with one or more computer applications;
creating the action data from the event data based on a mapping between use events and actions.
3. The computer-implemented method of claim 2 ,
a use event corresponding to one or more HTTP messages,
the mapping being based on a domain name, a payload, and a timestamp of an HTTP message of the one or more HTTP messages.
4. The computer-implemented method of claim 1 ,
the use data of the action including an item being inputted to the computer application or outputted from the computer application,
the item being an instance of a business object of a plurality of business objects under a business model that specifies a set of attributes for the business object.
5. The computer-implemented method of claim 4 , the constructing comprising:
preparing, from the action data, a training dataset of feature vectors and labels,
each feature vector corresponding to a certain item associated with a certain action that identifies a certain activity and having features corresponding to attributes of the business object of which the certain item is an instance,
each label corresponding to a next activity performed next to the certain activity.
6. The computer-implemented method of claim 5 , the constructing further comprising:
building a decision tree from the training dataset;
determining whether any path in the decision tree is associated with a probability that exceeds a threshold;
when a particular path is associated with a particular probability that exceeds the threshold, creating the rule specifying that when a particular item includes attribute values along the particular path, performing a particular activity that corresponds to a destination of the particular path.
7. The computer-implemented method of claim 2 , the detecting comprising:
detecting a current use event from interaction data under an application layer protocol;
creating a current action for the current use event.
8. The computer-implemented method of claim 7 , the automatically applying the rule comprising:
creating a first action, including an identifier of a first computer application to perform the first activity, an input item being inputted into the first computer application based on the specific activity, and a placeholder for each output item being outputted from the first computer application;
determining a first use event from the first action based on the mapping;
pragmatically effecting the first use event.
9. The computer-implemented method of claim 1 , further comprising, when a candidate next activity is associated with a probability that exceeds a threshold, automatically performing the candidate next activity.
10. The computer-implemented method of claim 1 , further comprising:
receiving a specification of a certain next activity of the list of candidate next activities from the device;
automatically performing the certain next activity.
11. A computer-readable, non-transitory storage medium storing computer-executable instructions, which when executed cause one or more processors to perform a method of predicting and automatically performing activities related to computer applications, the method comprising:
constructing, for a specific activity of a plurality of activities, a decision model based on action data of one or more actions,
each action of the one or more actions having an identifier of an activity of the plurality of activities performed with a computer application, use data being operated on by the computer application via the activity, and a timestamp,
the decision model including a rule specifying a first activity as a next activity or a probabilistic classifier that accepts use data being operated on via the specific activity and outputs an identifier of an activity to be performed as a next activity with a probability;
detecting, in real time, performance of the specific activity with specific use data;
when the decision model for the specific activity includes the rule, automatically applying the rule; and
when the decision model for the specific activity includes the probabilistic classifier and no applicable rule:
executing the probabilistic classifier to obtain a list of candidate next activities and an associated list of probabilities; and
sending the list of candidate next activities, indicating a recommended next activity, and the associated list of probabilities to a device.
12. The computer-readable, non-transitory storage medium of claim 11 , the method further comprising:
receiving event data indicating one or more use events related to user interactions with one or more computer applications;
creating the action data from the event data based on a mapping between use events and actions.
13. The computer-readable, non-transitory storage medium of claim 12 ,
a use event corresponding to one or more HTTP messages,
the mapping being based on a domain name, a payload, and a timestamp of an HTTP message of the one or more HTTP messages.
14. The computer-readable, non-transitory storage medium of claim 11 ,
the use data of the action including an item being inputted to the computer application or outputted from the computer application,
the item being an instance of a business object of a plurality of business objects under a business model that specifies a set of attributes for the business object.
15. The computer-readable, non-transitory storage medium of claim 14 , the constructing comprising:
preparing, from the action data, a training dataset of feature vectors and labels,
each feature vector corresponding to a certain item associated with a certain action that identifies a certain activity and having features corresponding to attributes of the business object of which the certain item is an instance,
each label corresponding to a next activity performed next to the certain activity.
16. The computer-readable, non-transitory storage medium of claim 15 , the constructing further comprising:
building a decision tree from the training dataset;
determining whether any path in the decision tree is associated with a probability that exceeds a threshold;
when a particular path is associated with a particular probability that exceeds the threshold, creating the rule specifying that when a particular item includes attribute values along the particular path, performing a particular activity that corresponds to a destination of the particular path.
17. The computer-readable, non-transitory storage medium of claim 12 , the detecting comprising:
detecting a current use event from interaction data under an application layer protocol;
creating a current action for the current use event.
18. The computer-readable, non-transitory storage medium of claim 17 , the automatically applying the rule comprising:
creating a first action, including an identifier of a first computer application to perform the first activity, an input item being inputted into the first computer application based on the specific activity, and a placeholder for each output item being outputted from the first computer application;
determining a first use event from the first action based on the mapping;
pragmatically effecting the first use event.
19. The computer-readable, non-transitory storage medium of claim 11 , the method further comprising, when a candidate next activity is associated with a probability that exceeds a threshold, automatically performing the candidate next activity.
20. The computer-readable, non-transitory storage medium of claim 11 , the method further comprising:
receiving a specification of a certain next activity of the list of candidate next activities from the device;
automatically performing the certain next activity.