Dylogos
The Dylogos software is a multi-agent decision-making system in which the agents are arranged hierarchically in two levels.
The first level is formed by agents of the meta category, which in Dylogos is called the macro system. The second level is represented by two categories. The categories are called micro system and The information system. Both systems are in turn divided into sub-categories, the objects. The mapping of the categories within and between the levels, as well as the mapping of the categories within a level, is done by means of functors.
In addition to the common Boolean operators, the class of functors includes extensions that go back to General Dynamic Logic and are part of Dylogos. The "exchange of information" between two objects is a morphism. Information can be mapped from one object to another, or it can be self-referential.
The functioning of the agents and their interaction illustrate this.
The information system has the function of a self-learning and self-optimizing semantic network. It is comparable to a 'semantic network processing system' according to Shapiro. By integrating such an information system as an intelligent agent, it is possible to consider textual values in a model formulation in addition to numerical values. Since Dylogos is designed as a decision-making system in its entirety, decisions are not only based on numerical values, such as measured values, but texts such as guidelines or laws can also be analyzed and taken into account.
Microsystems are intelligent agents for processing numerical values. These values can exist in two different states. The best known and simplest form is that of a real, measurable numerical value. The second state is often found in the description of natural systems, which are described by estimates in addition to a numerical description. In Dylogos, such fuzzy values are transformed by means of fuzzy logic into membership values, thus numerical values, and can therewith be made computable. This functionality applies to each object of the micro system. The objects are the visualization of the numerical parameters of the system or model to be described.
The interaction between the categories is done via the functors. In addition to the standard functors, calculi are used as functors as central components of Dylogos. Each calculus is tuned to its function of exchanging information between the respective categories. The calculi are self-optimizing and self-learning. The construction of the calculi takes place by means of causal chains, which are formed by linking the respective parameters by means of junctors. The parameters have a connotation through adjectives, which are represented by means of fuzzy sets.
The junction list is extended by functions of time, weights as well as experience and thus enables the formulation of dynamic systems.
The description of Dylogos is to be concluded with the third category and with the meta-category. This agent is called a macrosystem. This system can contain any number of micro systems as well as information systems. Each macro system delivers an overall result, which is the sum of the results of the individual micro and information systems included. The structure is fractal due to the possibility of including a macro system in addition to the two system types.
The Dylogos software was developed by the Private Institute for Dynamic Logic (PIFDL), Bernhard Stoinski. It is based on the General Dynamic Logic, which was also developed by the PIFDL. The Dylogos software is distributed under license by Dylogos GmbH.