Conversation tree - How is it built and traversed

Root of the conversation tree branches for each entry specified in the intents section of the schema. Every time the bot spawns a new microbot, a new child node is created in that path of the tree.

Dialog management and context switching.

The scaffolded bot on receiving an intent which is defined in intent or subIntents and not handled in the current state, searches the current branch of the tree to find if any microbot in the path was expecting this intent. Onces it reaches the root of the conversation tree, it searches if the leaf node of path associated with any triggering intent is expecting the current intent, if nothing it found it says that it could not understand. This is intuitive as to how humans usually have conversations, if something does not make sense in a conversation in the current context, we try to find a conversation we previously had to figure out where it is valid.

results matching ""

    No results matching ""