Technical Articles
Recast.ai for IT Ticket Classification is an efficient and accurate approach.
Hi Everyone,
Thanks for visiting my blog. This blog is actually an update to one of my previous blogs on IT Ticket classification using Machine Learning. Here, I will tell you how Recast.ai can be a better substitute to Python Machine Learning based approach or the Text Mining features of SAP HANA.
My blog on Python approach: https://blogs.sap.com/2017/08/09/automatic-it-ticket-classifier-powered-by-machine-learning/
To start with, Recast.ai is a Conversational Artificial Intelligence solution from SAP which could perform Natural Language Processing on Text/Voice media.
The output of our PoC is to predict the team/queue to which an IT Ticket could be cascaded depending on the text in the ticket. After working with Recast.ai on a different Chat bot project, we understood how this wonderful tool can be used to automate or replace the Machine Learning based Python script we were using previously.
Part 1 – Recast.ai
Step 1 – Create a new bot once you login into Recast.ai (SAP CAI)
Step 2: Create intents in the Recast.ai, for example I am creating my first Intent (Consider intents as Teams/queues in this requriement).
I have created 3 entities here.
Step 3: Adding expressions or text into the Intents (Tickets will be classified based on the expressions we mention here, agree its perfectly pattern matching. But, this goes with my requirement.I am not considering the grammar or true intention of user in the IT tickets!)
Similarly, I’m doing this for the other two Intents we have.
Step 4: Creating skills, I am creating one for each of the Intents I have. We do this in the ‘Build’ Part of Recast.ai
Fallback is a default Skill.
Step 5: Linking the intents to the skills. In Intents i.e., ‘Train’ section in Recast.ai we can do this in ‘Triggers’ part.
Similarly we do the same for the other two skills.
Step 6: Let us test if the classification is working using the chat interface provided in Recast.ai itself.
Let us consider ‘No internet’ is an IT ticket, here we see it got classified into network_dept which will be the team.
Step 7: You can follow the steps mentioned in this blog https://blogs.sap.com/2018/08/13/chatbot-integrationrecast.ai-with-sap-ui5 to call the dialog API to which we can send our IT tickets and get back the Intents which are queues in our case.
Step 8: Build a SAPUI5 tool as I have shown in my blog.
Tada ! Done.
Please contact me if you need more help on this. I will keep posting more about this.
Best,
Giridhar