Method and a system for process discovery

• US 8,631,391 B2
• Filed: 01/24/2005
• Issued: 01/14/2014
• Est. Priority Date: 01/24/2005
• Status: Expired due to Fees

• Current Assignee: Micro Focus LLC (Open Text Corporation)
• Original Assignee: Hewlett-Packard Development Company, L.P. (HP Inc.)
• Inventors: Casati, Fabio, Sayal, Mehmet, Castellanos, Maria Guadalupe, Li, Yifan, Shan, Ming-Chien
• Primary Examiner(s): DENG, ANNA CHEN
• Application Number:
  US11/041,653
• Publication Number:
  US 20060167923A1
• Time in Patent Office: 3,277 Days
• Field of Search: 717126-128, 717/131
• US Class Current: 717/128
• CPC Class Codes: G06Q 10/06 Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling

Abstract

The disclosed embodiments relate to a system and a method for process discovery. Embodiments of the present invention comprise extracting process data from a data source, creating a trace using the extracted process data, wherein the trace comprises a plurality of events, and detecting a process model using the plurality of events.

Claims

1. A computer implemented method comprising:

extracting process data from a data source;

creating a trace using the extracted process data, wherein the trace comprises a plurality of events;

reorganizing the plurality of events within the trace to create a reorganized plurality of events;

deriving a set comprising events that directly follow one another for a task X and a task Y from the reorganized plurality of events;

deriving a set comprising events that indirectly follow one another for the task X and the task Y from the reorganized plurality of events; and

detecting a process model using the set of events that directly follow one another and the set of events that indirectly follow one another;

wherein detecting the process model comprises detecting a process type based on a probabilistic relationship between the task X and the task Y, the probabilistic relationship determined through statistical analysis of the set of events that directly follow one another and the set of events that indirectly follow one another.

2. The method of claim 1, wherein detecting the process type comprises detecting a sequence process type if a number of elements in the set comprising events that directly follow one another is one (1) and a probability of Y^s immediately following X^e is greater than about 0.9.

3. The method of claim 1, wherein detecting the process type comprises detecting an XOR-split process type if one member of the set comprising events that directly follow one another for X comprises at least one of Y₁, Y₂, . . . , Y_n and wherein for any i, j from 1 to n, a probability of co-occurrence of X^e, Y_i^s, and Y_j^s is less than about 0.05.

4. The method of claim 1, wherein detecting the process type comprises detecting an AND-Split process type when Y₁, Y₁, Y₂, . . . , Y_n are in the set comprising events that indirectly follow one another for X and a probability is greater than about 0.95 that there is a co-occurrence of X^e, Y₁^s, Y₂^s, . . . , Y_n^s.

5. The method of claim 1, wherein detecting the process type comprises detecting an OR-Split process type if Y₁, Y₂, . . . , Y_n are in the set comprising events that indirectly follow one another for X, wherein X and Y do not comprise a XOR-Split process type or an AND-Split process type.

6. The method of claim 1, wherein detecting the process type comprises detecting an XOR-Join process type if one member of the set comprising events that directly follow one another for Y comprises at least one of X₁, X₂, . . . , X_n and wherein for any i, j from 1 to n, a probability of co-occurrence of X^e, Y_i^s, and Y_j^s is less than about 0.05.

7. The method of claim 1, wherein detecting the process type comprises detecting an AND-Join process type when X₁, X₂, . . . , X_n are in the set comprising events that indirectly follow one another for Y and a probability is greater than about 0.95 that there is a co-occurrence of Y^e, X₁^s, X₂^s, . . . X_n^s.

8. The method of claim 1, wherein detecting the process type comprises detecting an OR-Join process type if X₁, X₂, . . . , X_n are in the set comprising events that indirectly follow one another for Y, wherein X and Y do not comprise a XOR-Join process type or an AND-Join process type.

9. The method of claim 1, wherein the extracting process data from a data source comprises extracting process data from a log-file.

10. The method of claim 1, wherein creating the trace comprises creating a trace comprising a plurality of events characterized by a start time and a completion time.

11. The method of claim 1, wherein detecting the process model comprises detecting the process model by discovering process structures in the trace.

12. A computer implemented method comprising:

receiving a transmission comprising process data;

creating a trace that comprises a plurality of events using the process data;

reorganizing the plurality of events within the trace to create a reorganized plurality of events;

deriving a set comprising events that directly follow one another for two tasks X and Y from the reorganized plurality of events;

deriving a set comprising events that indirectly follow one another for the two tasks X and Y from the reorganized plurality of events; and

detecting a process structure using the set comprising events that directly follow one another and the set comprising events that indirectly follow one another;

wherein detecting the process structure comprises detecting a process type based on a probabilistic relationship between the task X and the task Y, the probabilistic relationship determined through statistical analysis of the set of events that directly follow one another and the set of events that indirectly follow one another.

13. The method of claim 12, wherein receiving the transmission comprising process data comprises receiving a transmission from an Enterprise Resource Planning system.

14. The method of claim 12, wherein creating the trace comprises creating a trace comprising a plurality of events characterized by a start time and a completion time.

15. The method of claim 12, wherein receiving a transmission comprising process data comprises receiving a transmission from a data warehouse.

16. A computer system for discovering process types, the computer system comprising a processor to execute instruction modules, the instruction modules comprising:

a process data extraction module that extracts data from a log file or database stored in a storage medium;

a trace creation module that creates a trace based on the extracted data;

a trace reorganization module that reorganizes the trace to produce reorganized trace data; and

a model detection module to;

derive a set comprising events that directly follow one another for two tasks X and Y;

derive a set comprising events that indirectly follow one another for the two tasks X and Y; and

detect a process model using the set of events that directly follow one another and the set of events that indirectly follow one another;

wherein detecting the process model comprises detecting a process type based on a probabilistic relationship between the task X and the task Y, the probabilistic relationship determined through statistical analysis of the set of events that directly follow one another and the set of events that indirectly follow one another.

17. The computer system of claim 16, wherein the model detection module is configured to detect a sequence process type when the number of elements in the set comprising events that directly follow one another is one and a probability of Y^s immediately following X^e is greater than about 0.9.

18. The computer system of claim 16, wherein the model detection module is configured to detect an XOR-split process type when one member of the set comprising events that directly follow one another for X comprises at least one of Y₁, Y₂, . . . , Y_n and wherein for any i, j from 1 to n, a probability of co-occurrence of X^e, Y_i^s, and Y_j^s is less than about 0.05.

19. The computer system of claim 16, wherein process data extraction module is configured to extract process data from a plurality of data sources and combining the process data from the plurality of data sources into a single log file.

20. A tangible, non-transitory, computer-readable medium that stores instructions that, when executed, effect process discovery, comprising:

instructions adapted to extract data from a log file or database;

instructions adapted to create a trace based on the extracted data;

instructions adapted to reorganize the trace to produce reorganized trace data; and

instructions adapted to derive a set comprising events that directly follow one another;

instructions adapted to derive a set comprising events that indirectly follow one another;

instructions adapted to detect a process model using the set of events that directly follow one another and the set of events that indirectly follow one another

instructions adapted to detect a process model using the set of events that directly follow one another and the set of events that indirectly follow one another;

wherein detecting the process model comprises detecting a process type based on a probabilistic relationship between the task X and the task Y, the probabilistic relationship determined through statistical analysis of the set of events that directly follow one another and the set of events that indirectly follow one another.