Grzegorz Dobrowolski

Agent Populations as Computational Intelligence

The paper deals with agent-based architectures of hybrid soft-computing systems, which should exhibit intelligent behaviour at a population level. Three levels of complexity of such systems are distinguished together with their potential advantages. The considerations are illustrated by prototypical realisations of evolutionary multi-agent systems dedicated to multiobjective optimisation, data classification, and time-series prediction.

Discovery of crises via agent-based simulation of a transportation system

The contribution deals with a class of intelligent decentralized systems that are marked by the possibility of arising critical situations. The work starts from the elaboration of an overall methodology dedicated to the discovery of crises and support of anti-crisis activities. Then the case of transportation enterprise support system is discussed in detail. A simulation study of anti-crisis management in such a system concludes the work.

CLUO: Web-Scale Text Mining System for Open Source Intelligence Purposes

The amount of textual information published on the Internet is considered tobe in billions of web pages, blog posts, comments, social media updates andothers. Analyzing such quantities of data requires high level of distribution –both data and computing. This is especially true in case of complex algorithms,often used in text mining tasks.The paper presents a prototype implementation of CLUO – an Open SourceIntelligence (OSINT) system, which extracts and analyzes significant quantitiesof openly available information.

Rough sets applied to the roughcast system for steel castings

Rough logic and rough sets theory are mainly devoted to analysis of incomplete, uncertain, and inconsistent data. An exploited in the article result of the theory is the so-called rough information system that gives a model for dealing with the situation in which some real objects are represented ambiguously via defined approximation (alternatives of values of the objects attributes). The proposed rough system opens a way to process such information and moreover is equipped with a powerful query language and possibility to classify real objects based on their stored rough characteristics. The implemented RoughCast system and, especially prepared for the case, knowledge base for steel casting defects form such the rough information system. They can be successfully applied in the process of knowledge integration concerning production of steel castings and, in consequence, as a tool in solving technological problems in operating foundries.

Agent-Based Modeling of Supply Chains in Critical Situations

Supply chains are discussed in this article. They are viewed as intelligent decentralized systems that meet the agent paradigm. Thus possibility of arising critical situations in the chains can be analyzed in the agent-based manner. The work is focused on applying an overall methodology dedicated to the discovery of crises and support of anti-crisis activities. The agent-based model is proposed that incorporates majority of features of supply chains so that the modeling of crises can be wide-ranging and easy. As an illustration some simulation results are presented.

Ontology of cooperating agents by means of knowledge components

A special way of structuralisation of knowledge and functionality of multi-agent systems is proposed that allows (self-) organisation and realisation of their global goal. Formally based structuralisation can also aid a process of designing (assembling) and documenting such a system. The approach is illustrated by the case of knowledge structuralisation of the decentralised information decision system oriented to diagnosis of casting defects.

Functional integrity of MAS through the dynamics of the agents' population

In open multi-agent systems, i.e. systems in which agents of different kinds may join or leave the system, the problem of functional integrity seems to be of vast importance. The thesis discussed in this paper is that the number of agents in a multi-agent system can be an important characteristic of its dynamics and, in particular, constitute a base for evaluation of its functional integrity at a population level. Certain mechanisms which allow for controlling the number of agents- preserving their autonomy-have been also proposed.

Agent-Based Integration of Data Acquired from Heterogeneous Sources

Agent-based framework dedicated to acquiring and processing heterogeneous data, collected in various Internet sources is presented. It is built upon a hierarchical, distributed computation system Age that has already been successfully used for various optimization and classification task.

Agent-Based Multi-variant Crisis Handling Strategies for SCADA Systems

Unpredictable changes in the problem parameters, especially when dealing with highly dynamic and complex systems like SCADA, should be predicted and appropriate strategies prepared accordingly. However, if the prediction is hard or impossible, one can consider preparing dedicated plans with dealing with these problems, in order to adapt the strategies ad-hoc when the problems arise. Such planning under uncerntainty has already been studied, and applied e.g. in the case of scheduling problems. The paper discusses SCADA as a problem similar to scheduling and constructs a system dedicated to prepare ready-to-use strategies before certain events arise. The solution is implemented in AgE platform and preliminary results are presented and discussed.

Using shallow semantic analysis and graph modelling for document classification

Using graph-based, shallow semantic analysis-driven approach for modelling text contents allow to extract additional information about meaning of text. This paper discusses using two novel algorithms that are based on this idea. They are compared against ‘legacy’ bag-of-words and Schenker et al. approaches in NN document classification task.