Intelligent Automation at work (Part Two)
Intelligent Automation (IA) at work (Part Two)
For the second part of our Intelligent Automation blog, we talk about e-mail classification with Microsoft Cognitive Services combined with RPA.
The strength of Microsoft Cognitive Services in Azure lies within a simple and low-code drag and drop interface. With some clicks and easy data inputs/manipulations, you can build your own machine learning model that gives you an edge in today's Machine Learning (ML)/Natural Language Processing (NLP) technologies.
AZURE MACHINE LEARNING STUDIO 💻
With the available standard Microsoft Azure Machine Learning Studio interface, we could easily integrate with the market leading RPA tools (UiPath, Blue Prism or Automation Anywhere). The trained model on Azure can be published as a Webservice that allows bidirectional communication with RPA tools through an API call. UiPath and Blue Prism provide build-in activities for this HTTP request. Communicating with out-of-the-box ML has never been so easy ! 👌
Before we start with the use case itself, we need to train the model with the proper data. You first need to decide what kind of learning alghorithm fits the needs of your model. The training of your model is all about data. We can assume that 200 training records are already sufficient to get a reasonable accuracy but you know, the more data, the better.
"The goal is to turn data into information, and information into insight."
"Imagine receiving hundreds of emails every day from customers with different types of questions and requests. It takes a lot of effort to first understand and categorize them, before they can be actually processed. Once the category is known, the RPA-bot can direct them into the desired authority that will answer the question/request."
First, the software robot catches and reads all incoming mail messages. First, the e-mails need to be cleaned like removing white spaces, commas, … This is required for the Machine Learning Model to understand the incoming data and make the right predictions. After this 'cleaning' process, the robot writes the messages to a .csv document and puts the items into the queue.
The robot takes the next queued item and executes the Azure prediction algorithm. Basically, the robot sends an HTTP request with the content to the model in Azure. After that, you get the obtained result back in a JSON format. The robot extracts the category and forwards the e-mail to the appropriate instance.
The whole process above takes less than 1 second. This proves the strength of this powerful technology and the endless possibilities of integration into your various robotic tools and business relevant use cases.
ROBOT AT WORK
RPA technology can be fluently combined with Machine Learning tools in Azure (or others). Intelligent automation is not an experimental adventure anymore, it became a tangible journey with real life cases. Let's get this train moving and use the available tools and intelligence they provide us. 👊
Reach out to us if you have an exciting use case that you might want to automate with these technologies! We are ready for you!