Platform architecture

Introduction

This guide describes the architecture of the UiPath Process Mining. In general, the UiPath Process Mining is used to load data and present results to end-users in web browsers.

System architecture

UiPath Process Mining consists of several components for developing process improvement applications.

See the illustration below for an overview of the system architecture of UiPath Process Mining.

Below is a description of the elements of UiPath Process Mining.

Apps / Discovery Accelerators

Process Mining provides ready-to-use apps and discovery accelerators for gaining insights on processes, carrying out root-cause analysis, and for continuous monitoring.

Using AppOne, the Purchase-to-Pay Discovery Accelerator and the Order-to-Cash Discovery Accelerator users can start analyzing processes immediately, without having to create a new app from scratch.

AppOne is a default dashboard template for generic processes. New process mining apps and discovery accelerators in UiPath Process Mining are made using AppOne, which is then configured to specific needs.

The functionality of apps and discovery accelerators can be extended with functions specific to your organization under its own brand identity. In case a different app is necessary altogether, the full functionality of UiPath Process Mining can be used to create a completely new app.

Git

Git is used for storing dashboards and collaborative development on the UiPath Process Mining platform. For single-server deployments, the built-in Git server can be used which does not require any additional setup. For multiple-server deployments, an (existing) Git server within your organization or a cloud-based Git server (e.g. from UiPath Process Mining or GitHub) can be used.

TRACY

TRACY is a technique that defines the layout of process graphs. TRACY lets process graphs look more like how you would draw a process yourself. When drawing a process, you normally begin with the start activity and finish with the end activity of the process. In between, you try to position all other activities in their executed order. TRACY takes the overall flow of the process into account and displays this as the main flow in your process graph. With TRACY all the activities of the process are positioned and ordered in a way that makes sense. This helps users to more easily understand their processes.

When changing data, TRACY minimizes changes to the process graph. When users add process filters to display happy paths or to filter out data, TRACY keeps the layout of the process graph as stable as possible. When analyzing a process, the process graph now always looks about the same, no matter which dashboard is used, or which filters apply. This makes analyzing the process easier.

TRACY smoothly animates the transitions between filter states. This helps users to understand what happens when filtering.

In-memory database

The in-memory database stores all data for fast access by the process mining algorithms. In this way, the data can be accessed very quickly without using the input databases.

Datasources

UiPath Process Mining supports text files such as .txt, .csv, .tsv files for which the delimiter and quotes can be selected. ASCII text files are supported in Latin-1 (ISO-8859-1) encoding, and UTF-8 files are supported with and without BOM.

Also, Excel files (.XLSX and .XLS) can be imported. The sheet or range within the file can be specified. For sheets, an autodetect is performed for the actual data range, of the auto-detect fails the range needs to be specified in Excel and then used. Unicode within Excel is fully supported.

All databases that can be accessed via a 64-bit ODBC driver can be used as a data source. Below is a list of commonly used databases.

• MSSQL
• Oracle
• MySQL
• MariaDB
• PostgreSQL
• Firebird
• Access

Script datasources (R/Python)

UiPath Process Mining contains functionality that supports processing data with external tools, such as R script and Python, that can be used as a datasource. Application developers can define which attributes need to be exported to the external process.

See: Use generic script datasources.