Detecting deviations between event log and process model

1. A method for updating a process model for treatment of diabetes, the method comprising:

detecting deviations between an event log and the process model by:

converting, with a processing device, the process model into a probability process model for treatment of diabetes, the probability process model comprising multiple nodes in multiple hierarchies and probability distribution associated with the multiple nodes, a leaf node among the multiple nodes corresponding to a diabetes treatment activity in the process model for treatment of diabetes, the diabetes treatment activity being selected from the group consisting of an HbAlc test, a blood sugar test, and a serum creatinine test, a non-leaf node among the multiple nodes corresponding to a phase in the process model, wherein the converting the process model into a probability process model comprises calculating start probabilities of occurrence of the non-leaf nodes, transition probabilities among the non-leaf nodes and emission probabilities from the non-leaf nodes to the leaf nodes, so as to form the probability process model;

detecting differences between at least one event sequence contained in the event log and the probability process model according to a correspondence relationship, wherein the correspondence relationship describes a correspondence relationship between an event in one event sequence of the at least one event sequence and the leaf node in the probability process model, and further wherein the types of the differences comprise violated constraints and determining the differences comprises:

using linear temporal logic to construct an automaton conforming to constraints in the probability process model; and

obtaining the violated constraints from a path sequence in the at least one event sequence that cannot be accepted by the automaton; and

identifying the differences as the deviations in response to the differences exceeding a predefined threshold;

wherein the correspondence relationship describes a correspondence relationship between an event in one event sequence of the at least one event sequence and a leaf node in the probability process model;

updating the process model based on the deviations and

adding new diabetes treatment activities based on the deviations to the process model.

2. The method according to claim 1 , wherein the calculating start probabilities of occurrence of the non-leaf nodes, transition probabilities among the non-leaf nodes and emission probabilities from the non-leaf nodes to the leaf nodes, so as to form the probability process model comprises:

setting the start probabilities, the transition probabilities and the emission probabilities to form the probability process model; and

iteratively training the probability process model.

3. The method according to claim 1 , wherein the detecting differences between at least one event sequence contained in the event log and the probability process model according to the correspondence relationship comprises: with respect to a current event sequence of the at least one event sequence;

aligning each event in the current event sequence with the leaf node in the probability process model based on the probability distribution;

recording a path, of a leaf node with which each event is aligned, in the probability process model so as to convert the current event sequence into a path sequence; and

determining the differences based on the probability process model and the path sequence.

4. The method according to claim 3 , wherein the path comprises a multi-level path corresponding to multiple hierarchies in the probability process model.

5. The method according to claim 3 , wherein types of the differences comprise violated constraints, and the determining the differences based on the probability process model and the path sequence comprises:

using linear temporal logic to construct an automaton conforming to constraints in the probability process model; and

obtaining the violated constraints from a path sequence that cannot be accepted by the automaton.

6. The method according to claim 3 , wherein types of the differences comprise absent activities, and the determining the differences based on the probability process model and the path sequence comprises:

calculating a universal set S Model of paths of the leaf nodes in the probability process model, and a universal set S Log of a path, of a leaf node aligned with an event in an event sequence of the at least one event sequence, in the probability process model, respectively;

calculating a difference set S Absent =S Model −S Log between the universal set S Model and the universal set S Log as absent activities.

7. The method according to claim 3 , wherein types of the differences comprise additional activities, and the determining the differences based on the probability process model and the path sequence comprises:

calculating a universal set S Model of paths of the leaf nodes in the probability process model, and a universal set S Log of a path, of a leaf node aligned with an event in an event sequence of the at least one event sequence, in the probability process model, respectively;

calculating a difference set S Additional =S Log −S Model between the universal set S Log and the universal set S Model as additional activities.

8. The method according to claim 1 , wherein the identifying the differences as the deviations in response to the differences exceeding the predefined threshold comprises:

calculating a ratio of the number of event sequences with differences of the at least one event sequence to the total number of the at least one event sequence; and

in response to the ratio exceeding a predefined ratio, identifying differences of the type as the deviations.