RobbySpider history (1)
RobbySpider has been handling important compliance tasks, such as onboarding and risk verification in the KYC Toolbox since 2019. This is a good opportunity to have a look at the most notable chatbot predecessors to RobbySpider. First, we have the pioneer system, GUS [1], which was developed in 1977 on the legendary Xerox PARC [2]. GUS is based on what is known as a frame-driven dialog system. This principle of “frame-based slot-filling” is still relevant today and is used in Apple’s Siri, for example [3].
Resultate
GUS is based on an agenda that iteratively determines the next task to be performed in this multi-process environment. When one task is completed, new tasks emerge that are then also added to the agenda. Frames are the central data structures in GUS. A frame contains what are referred to as slots, which do not simply consist of key value pairs. Instead of a value, they also use procedures to define how a slot value is determined. There are two types of procedures. The demon procedures are executed automatically if the frame is modified. The demon procedure will then attempt to perform an additional derivation due to the modification. The servant procedures are the second type. A servant procedure calculates a value only if it is needed elsewhere.
GUS supports dialogs that are not exclusively controlled by the machine. A person can also play an active role to some extent (mixed initiative), considerably expanding the range of potential statements made by the human participant in the dialog and therefore making interpretation more difficult. For this reason, developers of GUS-based systems have ensured that the machines can easily take control of the dialog again and that the dialog is restricted to a certain domain.
Sources and Links:
[1] Genial Understander System (GUS) bedeutet auf Deutsch etwa ein liebeswürdes, leutseliges System, das zuhört und versteht.
[2] Xerox Palo Alto Research Center (Xerox PARC)
[3] Conversational Agents, Lecture by Dan Jurafsky at Stanford University, CS 124/LINGUIST 180 From Languages to Information
