Context is linguistically defined as the background in which the conversation is taking place. In the bot perspective, it can involve retaining the memory of any piece of information that has been either explicitly given by the user, or derived from past user behavior, or any action performed by the IVA.

It can involve information like steps traversed by the IVA in the conversation, entity values collected, ML (Machine Learning) Module which sent the last reply e.g. Disambiguation, Small Talk, etc.

User’s own name, email, phone number, etc. explicitly provided by him, will also form a part of the context. The context needs to be stored before it can be used. Before we understand when the context is used, it’s important to know when it’s deleted.

When is the context information deleted?

The context is deleted by the system in case of any of the following events -

• When a Context Clear step is encountered by the IVA in the conversation

• After 24 hours from the time the conversation was initiated.

When is the context used? If the context has not been deleted as per the above 2 cases, the IVA can use the context within 24 hours, from the time the conversation was initiated by the user.

Illustration - Let’s say there is a conversation that ideally can get 4 levels deep as shown in the image below. It has a START step, a CHILD step, a Grand-Child step, and a Great Grand-child step with the Great Grand Child step being the Context Clear step.

Let’s say the user starts the conversation from a Start Step ABOUT HAPTIK and then chooses an option to reach the child step FUNDING OF HAPTIK. At this point, if a user leaves the conversation, the IVA will retain the context for 24 hours since the no Context Clear Step has not been encountered in the conversation, so far.

So if a user comes back within 24 hours, say e.g. 12 hours, and taps a button or a Quick Reply on the FUNDING OF HAPTIK step, he can reach the Grand-Child step i.e. CAREERS AT HAPTIK Step directly.