TerminologyΒΆ
- action
- A verb that corresponds to a single task that LILI can complete. For example, the action follow, as in the command “Follow me to the kitchen”, tells LILI to follow someone.
- action set
- A group of actions that have the same meaning to the interpreter. For example, the actions turn, twist, and rotate comprise an action set. These three actions can be used interchangeably in the same command. The commands “Turn right”, “Twist right”, and “Rotate right” will all be interpreted as the same command, all yielding the same output from LILI.
- action set index
- A positive (or zero) integer value that uniquely identifies an action set.
- object
- A key word that must be interpreted in order for LILI to perform the correct task. In the command “Follow me”, follow is the action and me is an object because LILI must know who to follow in order to do that task as the user intends.
- object dictionary
- A dictionary, a set of key-value pairs, that maps a set of objects to semantic labels that correspond to the context and meaning of each object. In the command “Follow me”, me is stored as a value in an object dictionary whose key is person.
- object extractor function
- A function that uses predetermined rules to extract objects from commands, tags the objects with semantic labels, and generates an object dictionary.