Node Types
A dialogue model consists of nodes representing different steps in the conversation.
All the node types are the following
Entering & Leaving the Conversation
The principle is simple - when you want to talk you must enter and exit each conversation.
Entering the Conversation
In our conversational platform entering the conversation is always through the Enter node.
Leaving the conversation - Exit, End, Sleep
In a specific part of the conversation you might want to allow user to come back and continue. Use Sleep node in such situation. The conversation will be sleeping/waiting for an hour by default or you can specify how many minutes you want to left the conversation in sleep mode before the final termination.
Use the end node for the final end of the conversation.
When the application contains just one (main) dialogue then you can't recognize the difference between the exit and end nodes. But when using the subdialogues the situation becomes complex.
End the Conversation
The End node terminates the session and that means the conversation with the user at any point.
Exit the Dialogue Model
Exiting the dialogue model means that the model represented by the graph and its logic behind is terminated but when the dialogue is used as a subdialogue in other dialogue than the conversation (session) continues. To be more clear let's have a look at the example.
Let's have an application linked to the main dialogue including several dialogues used as subdialogues representing the specific topics. Each could be used independently. When it ends then the conversation is over. But when used as a sub-dialogue we can have two situations
user talking to the application wants to end the whole conversation
user went through the whole subdialogue and we want to allow him to jump back to the main dialogue
The following example shows how the main dialogue for Lucid Lindia references the dialogue related to the music (and many more topics). After the discussion about music and reaching the exit node the conversation continues in the main dialogue.
Entering the Subdialogue node Music activates the dialogue music/main we jump into this independent dialogue
A conversation goes from Enter to Exit
We jump back to the main dialogue and the conversation continues by Speech6.
Entering the Subdialogue node Music activates the dialogue music/main we jump into this independent dialogue
A conversation goes from Enter to Exit
We jump back to the main dialogue and the conversation continues by Speech6.
Subdialogue
The independent dialogue model but used in the bigger application typically referenced in the main dialogue linked to the application.
Hierarchy
Subdialogue can include another subdialogue.
Speech
The most used node type. What you type in the application will say aloud.
The text in the node will be sent to the Text To Speech converting the text to voice. A node can be set as repeatable. If true, the text in this node will be included when the bots repeat a response. The typical situation is when the intent is not recognized and the application repeats the question asking for the input again.
Intent
Intent nodes serve to represent the user’s utterances. These are similar to the intent node but they do not start a branch, they continue one. Usually, several nodes are connected to an input node. Several Contextual Intent nodes are provided with pre-filled common utterances. These nodes have a “Negate“ function (available above the text editor). It creates a new node with utterances which are negations of the original node utterances (e.g. from node with utterance “I like that“ will be created a node containing an utterance “I don’t like that“).
NOTE: The new nodes are created on top of the originals.
Global Intent
This node serves for starting a branch of a dialogue. One node labeled ‘Start’ is required for each dialogue, others are optional. In the headline, fill in the intent’s name (must be unique in context of the current dialogue). The content of this node are examples of user utterances which represent this intent (e.g. fill in “What time it is“ or “Do you know the time“ for intent which prompts the bot to tell its user the time). The exception is the start node which is empty.
Action (local)
We get the input into the model but not said by the user. The most typical action is silence. There is no intent because a muted microphone or a user really stayed silent. See the global actions below for more examples.
Global Action
This node serves for starting a branch of a dialogue.
What actions to cover on the global level?
silence
outofdomain
error
toomanywords
Save your time and use the prepared snippet Global actions required.
Go Back
A Go Back node follows typically after the (global) action + speech. See the example for the global actions.
Sound
Sound node is a special version of the Response node. The designer can write either an audio item from File Assets or a full URL pointing to audio resource file.
NOTE: Option “External” must be chosen in the dropdown menu above the text editor for the latter to work.
Function
In case the dialogue needs to be directed programmatically, use this node. One link can go in and multiple out. Headline is the name of the function and should be unique.
The JS function should return the number of the next node. When the dialogue is loaded, the functions is included in the code on the “Functions“ tab.
User Input
This node can also be filled with code similarly to Function node.
Image & Sound
Drag & drop an image or a sound from pane into your graph.
Upload files to your Space
Open the multimedia tab (assets) in the dialogue designer
Insert a file into a dialogue (drag&drop)
For the step-by-step guide got to Add Sounds and Images.
Command
Use a command node when you want to invoke a special action on the client (device). The example can be that a robot will say what comes in as speech and raise a hand when receiving a specific command.
Last updated
