On 3 June 2014, Opole University hosted Prof. dr hab. Barbara Dunin-Kęplicz of the Institute of Computer Science of Warsaw University, who delivered her lecture under the title Making Complex Beliefs Tractable. The meeting with Prof. B. Dunin-Kęplicz, which was co-organized by the Institute of Mathematics and Informatics of Opole University, was held in Room no. 15 of Collegium Civitas of Opole University at 89 Katowicka Street.
Professor Urszula Wybraniec-Skardowska opened the meeting and welcomed the lecturer of the day. Professor Janusz Czelakowski followed with an introduction into the problem area raised in the lecture. Then, Professor Lidia Tendera presented the scientific profile of the guest - Professor Barbara Dunin-Kęplicz.
In her lecture, Prof. B. Dunin-Kęplicz presented the classic understanding of collective knowledge and its virtues, as well as the understanding of collective knowledge in multi-agent systems, pointing - at the same time - to a certain evolution with relation to the classical approach. The lecturer discussed modeling of complex belief structuress, distinguishing the so-called epistemic profiles which serve to transform beliefs. She made reference to using inference founded on para-consistent knowledge bases and 4QL language that constitutes the foundations while creating this type of knowledge bases.
The classic framework of collective (common) knowledge underlines the significance of a consensus among group members. Reaching understanding facilitates the process of reasoning which goes on within the group - its members draw conclusions from generally known and accepted premises. Although the above approach is useful in modeling various aspects of group behaviour and inference, e.g., while working out different norms and conventions, the cost of reaching a consensus can be too high. In contemporary works, the role of consensus is not accentuated. On the other hand, the following epistemic aspects of group cooperation are highlighted:
Instead of the principle of "what everybody knows", collective knowledge is directed towards supplying synthetic information abstracted from data provided by individual group members. This synthetic information concerns selected aspects (or fragments) of collective knowledge and is then made use of within various forms of group activity.
Collective conclusions are not treated as "private" conclusions of individual group members who, however, adjust in their actions (including epistemic ones) to the conclusions which bind the whole group, accepting then the meta-principle of precedence of collective knowledge over individual knowledge.
Indicating non-dogmatic procedures of forming collective knowledge on the basis of knowledge of individual members.
Tolerance of contradictions appearing at different stages of the process of collective knowledge formation. As a result, the para-consistent bases of knowledge are accepted.
The existence of information gaps, where certain aspects of knowledge are not available in a variety of applications. As a result, various non-monotonic forms of inference are accepted.
The approach proposed by Prof. Prof. B. Dunin-Kęplicz and Szałas leads to a new framework of complex belief structures. The epistemic profiles mentioned above are thought to be a tool which is supposed to fill in the void between various concepts of multi-agent epistemic logic and practical challenges, whose task is to transform initial data into useful final conclusions. Małuszyński and Szałas defined 4QL, a four-valued language of queries. It secures practical calculability for para-consistent bases of knowledge, as well as implementation of the epistemic profile and belief structures
In the absorbing discussion that followed the lecture, different interesting themes connected with the discussed problem area were raised.