Implicit process detection and automation from unstructured activity

Patent metadata

Field	Value
Publication number	US-2016267268-A1
Application number	US-201514657646-A
Country	US
Kind code	A1
Filing date	Mar 13, 2015
Priority date	Mar 13, 2015
Publication date	Sep 15, 2016
Grant date	—

Abstract

An unstructured event is detected and an unstructured event record is generated for the detected event. Case identifier (ID) conflation is performed to estimate a case ID that corresponds to the detected event, and event type identification is performed to estimate a type of the unstructured event. A business process model is applied to the unstructured event record, to identify a process that the unstructured event is related to. A user experience is generated based upon the event type, the case ID, and the corresponding process identified for the detected event.

First claim

What is claimed is:

1 . A computing system, comprising:
an event type identification system that receives an unstructured event record with event information indicative of an unstructured event, the event type identification system accessing an event type identifier model, identifying an event type corresponding to the unstructured event based on the event information and the event type identifier model, and generating an event type label, indicative of the event type, for the unstructured event record; a process model matching system that accesses a process model and a set of logged computing system processes and identifies a process in the computing system that the unstructured event corresponds to, based on the unstructured event record; a user interface component; and a workflow prediction system that predicts a workflow based on the identified process and the identified event type for the unstructured event record and controls the user interface component to surface the predicted workflow for user interaction.

2 . The computing system of claim 1 wherein the process model matching system detects an unstructured activity pattern for a given user.

3 . The computing system of claim 2 wherein the process model matching system detects whether the unstructured event is part of an activity pattern for a given user and wherein it identifies that the activity pattern relates to the identified process.

4 . The computing system of claim 3 and further comprising:
a conformance checking system that compares the unstructured activity record against the activity pattern for the given user and identifies an anomaly in a the activity pattern for the given user, based on the unstructured activity record.

5 . The computing system of claim 4 and further comprising:
a recommendation engine that generates a recommended action based on the anomaly.

6 . The computing system of claim 5 wherein the recommendation engine controls the user interface component to surface a user interface display with a user input mechanism that is actuated to take the recommended action.

7 . The computing system of claim 5 wherein the workflow prediction system identifies a computing system object corresponding to the unstructured event record and opens the computing system object in an application context based on the unstructured event record.

8 . The computing system of claim 2 and further comprising:
a case identifier (ID) conflation system that accesses a case ID conflation model to associate the undetected event record with a case in the computing system.

9 . The computing system of claim 1 and further comprising:
an event detection component that detects the unstructured event; and
an information extraction component that extracts the event information and generates the unstructured event record.

10 . The computing system of claim 1 wherein the event type identifier model comprises:
an event type classifier that classifies the unstructured event record into an event type.

11 . A computing system that associates related, structured events with a case identifier, comprising:
a process model matching system that receives a labeled unstructured event record that includes event information indicative of an unstructured user interface event in the computing system, the unstructured event record being labeled with an event type, the process model latching system accessing a process model to identify an entity in a structured process that the unstructured event record corresponds to; a user interface component; and a workflow prediction system that controls the user interface component to surface the entity based on the unstructured event record.

12 . The computing system of claim 11 and further comprising:
an event type identification system that receives the unstructured event record and accesses an event identification model to identify an event type corresponding to the unstructured event record.

13 . The computing system of claim 12 wherein the event identification model comprises a classifier.

14 . The computing system pf claim 12 and further comprising:
a case identifier (ID) conflation system that receives the unstructured event record and accesses a case ID conflation model and identifies a case ID within the computing system to which the unstructured event record corresponds.

15 . The computing system of claim 14 wherein the process model matching system identifies an unstructured activity pattern and determines whether the unstructured event record is part of the unstructured activity pattern.

16 . The computing system of claim 15 and further comprising:
a conformance checking system that determines whether the unstructured event record represents an anomaly in the unstructured activity pattern.

17 . The computing system of claim 16 and further comprising:
a recommendation engine that controls the user interface component to surface a recommendation in response to the conformance checking system determining that the unstructured event record represents an anomaly.

18 . A computer implemented method, comprising:
detecting an unstructured event in a computing system; extracting information indicative of the unstructured event to generate an unstructured event record; classifying the unstructured event record to identify a corresponding event type; identifying a part of a structured workflow in the computing system that the unstructured event record corresponds to; and controlling a user interface component to surface the part of the structured workflow for user interaction.

19 . The computer implemented method of claim 18 and further comprising:
identifying a set of activity patterns in the computing system corresponding to a given user; and
determining whether the unstructured activity record represents an anomaly in a given activity pattern; and
if so, controlling the user interface component to generate a recommendation to address the anomaly.

20 . The computer implemented method of claim 19 and further comprising:
accessing a case identifier (ID) conflation model to assign the unstructured event record to a case ID within the computer system.

Assignees

• Microsoft Technology Licensing Llc

Inventors

• Sheldon Graham Andrew Michael
• Gokalp Sedat

Classifications

• G06F21/565

  by checking file integrity · CPC title

• G06F21/554 Primary

  involving event detection and direct action · CPC title

• G06Q10/0635

  Risk analysis of enterprise or organisation activities · CPC title

• G06Q10/0633

  Workflow analysis · CPC title