Browse Prior Art Database

Technique for automatic human task assignment in BPM using machine learning model.

IP.com Disclosure Number: IPCOM000234678D
Publication Date: 2014-Jan-28
Document File: 5 page(s) / 46K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed is a technique for automatic human task reassignment in BPM using machine learning model.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 39% of the total text.

Page 01 of 5

Technique for automatic human task assignment in BPM using machine learning model.

Main Idea

Background and Problem:

Human centric business processes involve multiple users/groups working on different modules/task of the process. In today's BPM system, a human task is assigned to group. The task is then manually assigned to a group member either by the group manager or claimed by a group member for further processing. If the group size is large ( ~10 members or more ), then manager needs to consider several factors for efficient task assignment. The factors can be :


1. Workload of each member.


2. Complexity of the task.


3. History of the task processed by each member.


4. Expert on the task.


5. Availability of each member.

Adhoc/Manual assignment of different tasks in the system to any of the members in the group poses the following problems :


1. Multiple re-assignment of the same task to different members.


2. Uneven work load among group members.

    3. High average time to complete the task due to negligence of prior history in task assignment.


4. Unnecessary involvements of topic experts on some tasks.


5. Reduced productivity of knowledgeable human resources.


6. Increases the chance of errors.

Known solutions:


1. IBM BPM


2. Oracle BPM


3. Sunguard Infinity BPMmes

Drawbacks:

    1. These systems do primitive analysis of task and user history to suggest topic experts which in some cases may not be correct.

    2. These system do not prevent task re-assignment problem but provide information which can be consumed by knowledgeable human for future task assignment.

    3. IBM BPM does task reassignment in a round-robin fashion which does not address any of the problems mentioned above.

    4. Rule based systems do provide flexible rules to schedule assignment of tasks based on workload / leave schedule. However , these rules need to be defined much before a BPM application goes live as we cannot have dynamic rules created on-the-fly

1


Page 02 of 5

if an incoming task comes in and gets assigned to a team member whose leave schedule was not planned before. Also , as the team size grows , the number of rules would grow and the system would slowly become inefficient as maintaining these huge set of rules data would become painful.

In a nutshell, the current state of art BPM system does not have automatic task assignment capability causing above problems.

Prior art:


1. Geo-Location Improves Business Process Management Human Task Assignment

https://www-304.ibm.com/files/form/anonymous/api/library/3cef8521-8f61-4d75-9ade-3d 1ed168a642/document/c170764f-fdac-4170-bc2f-0a3e1b501540/media/aot_atn_v3_n6. pdf

This paper talks about use of Geo-location information in automatic BPM task assignment but doesn't present any system for such automated task assignment.

Solution Description:

Our solution has a personalized and trained Linear regression (LR) based machine learning model for each group member with following features.


1. Task complexity


2. Member Workload


3. M...