Process Discovery Overview
Process Discovery is a technology to automatically construct a representation of an organization's current workflows and their major variations. An organization's workflows are made up of processes, which are composed of tasks, which are composed of events. Processes are workflows which span across users, teams, departments and even organizations, and incorporate all the data from all the available sources over time.
Event data is collected from tracking users and sent to the AI Management Server which connects to the SQL Database and the cloud-hosted GPUs in the end-clients cloud architecture.
The AI Management server (pre and post processing) compiles the data for the GPUs to run the process models and train the AI. The SQL Database stores event data as well as outputs from the process models.
GPUs host the AI software which pulls event data from the database, trains the process models, produces the outputs which are the identification of possible sequences of events, and sends it back to the database.
The AI finds commonality between similar sequences of events and organizes them into a task. The AI “maps” the most common path for that task as well as all the variants. After a task is found (Task Discovery) and mapped (Task Mapping), the AI performs Task Identification and Logic Analysis.
An output of Task Mapping, Task Instant Details, is then used by the AI to perform Process Discovery. Process Discovery constructs a representation of an organization's current workflows and their major variations. After a process is found, the process is mapped (Process Mapping). After the process is mapped, Process Identification is performed on it.
The Dashboard visualizes the outputs of all of the above operations for the user; including tracking, analytics, visual comparisons of the data, task discovery, task mapping, process discovery, process mapping, task identification and process identification.
Task Discovery and Task Mapping:
To perform Task Discovery, there must be sufficient sequential repetitive events collected in the database for comparison. Events must be repeated at least five times, in order for the data to be available for Task Discovery.
To initiate Task Discovery, the Dashboard Administrator for the client must start the AI Training in the Dashboard. This can be set on a schedule or initiated manually. To help conserve resources, AI Training, which uses the GPUs, is not running continually.
When the AI Training is initiated, the AI Manager begins the training process by taking the event data(actions performed by users and accompanying contextual information) stored from tracking users. The AI Manager pre-processes the data to identify possible event sequences or common events.
The AI Manager then starts the GPUs and initiates the deep learning neural network. The GPUs pull the data from the database, and deep learning neural network:
- Trains itself on the data, and then estimates which events belong to a task and groups them together, thereby discovering a task. (Task Discovery)
- Estimates which events within a task follow every other event thereby mapping each tasks’ major pathway and its variations. (Task Mapping)
- After the tasks are mapped, a summary of each task discovered is generated by the AI Manager. The task summary includes a score for the task indicative of a difficulty of automating the task.
Task Mapping and Task Identification:
Also after a task is mapped, the AI Manager performs Task Identification. This finds the task type, gives it a description and scripts out the higher level steps. Task Identification uses an AI to give context to each mapped task. This AI is responsible for creating context for the Task Identification. It provides a paragraph description of the task, gives it a name and writes a short sentence for each higher level step.
Logic Analysis is performed by the AI Manager by finding all the If/Then/Else statements for variations within Task Maps if those variations contain distinct variables and decision points. The AI performs a Logic Analysis which finds the variables and logic in the task steps. The Logic Analysis can change the Task Maps, insert if/then statement.
Process Discovery and Processes Mapping:
Process Discovery is conducted by the AI by finding commonalities and grouping all similar tasks. This is done by analyzing and storing the Task Instant Details as well as any available Process Mining data. Process mining assumes the existence of an event log where each event refers to a case, an activity, and a point in time. An event log can be seen as a collection of cases and a case can be seen as a trace/sequence of events.
Once a process is found it will be mapped by the AI manager by estimating the number of possible processes within an organization via an network algorithm. First, by training a deep learning neural network to estimate which tasks belong to a process. And second, by training a deep learning neural network to estimate which tasks within a process follow every other task.
After a process is completely mapped the AI will perform Process Identification on it. The Process Identification finds the process type, gives it a description, and scripts out the higher level steps from the major pathway.
Process Identification uses an AI called to give context to each mapped process. This AI is responsible for creating context for the Process Identification. It provides a paragraph description of the process, gives it a name and writes a short sentence for each higher level step.
Finally the AI Manager generates a summary of all the mapped processes discovered, including a score for the process indicative of the difficulty of automating the process.
For more information about using OfficeAutomata's Dashboard please contact us or visit the other training guides.
Non-Disclosure Disclaimer: This document and the information contained herein is confidential; This document is provided for information purposes only for the exclusive use of the recipients to whom it is addressed and the contents hereof are subject to change without notice. Whilst the information contained herein has been prepared in good faith, it is not warranted to be error-free, nor subject to any other warranties or conditions, whether expressed orally or implied in law, including implied warranties and conditions of merchantability or fitness for a particular purpose. Office Automata specifically disclaim any liability with respect to this document and no contractual obligations are formed either directly or indirectly by this document. Any reproduction, retransmission, republication, translation, or other use of, all or part of this document is expressly prohibited, unless prior written permission has been granted by Office Automata. Office Automata, the Office Automata logo and other all other Office Automata trademarks, logos and service marks used in this document are the trademarks or service marks of Office Automata and its affiliates. All other marks contained herein are the property of their respective owners. Office Automata has intellectual property rights relating to technology that is described in this document.