User interaction logic classification
Back end calls triggered by a user interaction with a client user interface may be identified. The user interaction may be correlated with a logic flow, and the logic flow may be associated with the back end calls. A supervised learning model may be trained using a labeled data set comprising the back end calls and their associated logic flow. Rules may be derived from the supervised learning model for classifying other back end calls. The rules may be outputted to a classifier that utilizes the rules to associate the other back end calls with the logic flow.
1. A non-transitory machine-readable storage medium encoded with instructions executable by a processor, the storage medium comprising:
instrumentation instructions within program code of a user interface to track execution of the program code in response to user actions with the user interface, wherein the execution results in a generation of first back end calls, and the tracking provides contextual information for the first back end calls;
correlating instructions to, based on the contextual information, associate the user actions with a logic flow;
associating instructions to associate the logic flow with the first back end calls;
training instructions to train a supervised learning model using a labeled data set comprising the first back end calls and the logic flow;
deriving instructions to derive rules from the supervised learning model for classifying other back end calls, wherein the first back end calls are made to a URL, and the other back end calls are unclassified back end calls that are made to the URL; and
outputting instructions to output the rules to a classifier that utilizes the rules to associate the other back end calls with the logic flow.
2. The non-transitory machine-readable storage medium of claim 1 , wherein the labeled data set comprises HTTP parameters of the first back end calls.
3. The non-transitory machine-readable storage medium of claim 2 , wherein the HTTP parameters comprise at least one of an HTTP header and a POST payload.
4. The non-transitory machine-readable storage medium of claim 1 , further comprising prioritizing instructions to prioritize protocol parameters of the first back end calls according to a probability of association with the logic flow.
5. The non-transitory machine-readable storage medium of claim 1 , wherein the classifier is located on an application server.
6. The non-transitory machine-readable storage medium of claim 1 , wherein the classifier is located on a data processing server that consumes HTTP parameters of the first back end calls.
7. The non-transitory machine-readable storage medium of claim 1 , wherein the logic flow is a purchasing flow or a search flow.
8. The non-transitory machine-readable storage medium of claim 1 , wherein the classifier uses the rules at runtime to associate the other back end calls with the logic flow.
9. The non-transitory machine-readable storage medium of claim 1 , wherein the supervised learning model comprises at least one of a decision tree model, a support vector machine model, a logistic regression model, or classification rules.
10. The non-transitory machine-readable storage medium of claim 1 , wherein the user interface comprises widgets, and the contextual information represents selection of the widgets by the user actions.
11. The non-transitory machine-readable storage medium of claim 1 , wherein the contextual information represents a time for the user interface to respond to a given user action of the user actions.
12. The non-transitory machine-readable storage medium of claim 1 , wherein the instrumentation instructions, when executed by the processor, cause the processor to intercept the program code.
13. A method, comprising:
using instrumentation instructions within program code of a user interface of a client to track execution of the program code in response to user actions with the user interface, wherein the execution results in a generation of first back end calls, and the tracking provides contextual data for the first back end calls;
based on the contextual data, associating the first back end calls to a uniform resource location (URL) to a logic flow;
training a supervised learning model using at least the first back end calls and the logic flow;
prioritizing HTTP parameters of the first back end calls according to a probability of association with the logic flow;
deriving, from the supervised learning model, rules for classifying other back end calls; and
outputting the rules to a classifier application that utilizes the rules to associate the other back end calls with the logic flow.
14. The method of claim 13 , wherein the other back end calls are unclassified back end calls that are made to the URL.
15. The method of claim 13 , wherein the supervised learning model comprises at least one of a decision tree model, a support vector machine model, a logistic regression model, or classification rules.