An algorithm for improving the consistency of structural interoperability

Aim. The consistency of a collectively generated body of knowledge is one of the most difficult problems of artificial intelligence. Additionally, there is a very important aspect of the knowledge acquisition process, i.e., the availability of knowledge support tools. Within the overall scheme of operation of knowledge system maintenance procedures, those associated with ensuring the consistency of interrelated components hold a special place. Out of the lines of knowledge representation research, the study of the operating procedures of corporate knowledge systems stood out. One of its most unconquered challenges is finding an optimal structure for the interoperability of knowledge system components. Let us consider a set of n elements that represent agents, information systems or knowledge components, then we can set the goal of assessing the level of interoperability and present the structure of interoperability based on an analysis of preferred relationships between elements. Methods. Thus, the effect of the structure of interacting elements on the motivation for interoperability is to be taken into account based on certain features or characteristics in the structure of elements that contribute or hinder the achievement of interoperability. It was noted that the potential for establishing interoperability based on a structure of interrelated elements can be defined as structural interoperability. Accordingly, for the purpose of studying the tendency of information systems or elements to be interoperable depending on the correlation of their own features or characteristics, a structural correspondence methodology was proposed that allows evaluating groups of potentially close elements using the structural consistency apparatus. Modelling structural interoperability based on the analysis of the connections structure using the selected consistency criterion allows finding the consistent preimage that is the closest to the original set. The subsets of the found preimage indicate the preferred grouping of elements that enables interoperability with the least inconsistency. Results. As a result, the set of potentially interacting elements is divided into sets of elements motivated to interact. This paper proposes and justifies an algorithm for finding – for a random structure of elements – the closest consistent structure that allows concluding on the choice of the interoperability structure. Conclusions. Out of the analysis of the proposed algorithm and its alternative follows that, under defined conditions for the existence of structural interoperability, there are several options for an acceptable interoperability structure. However, finding the optimal option will require searching through all acceptable options, or using reasonable heuristics that take into account the specifics of the connectivity matrix of the original signed graph. Comparing the presented algorithms, it must be noted that the label propagation algorithm is evaluative in its nature. Meanwhile, despite the complexity of combinatorial estimates and transformations, the inconsistency reduction algorithm based on the vertex-to-vertex difference vector is a tool for permanent analysis and managing the consistency of sets of elements motivated for interoperability.

1. Rozenberg I.N., Dulin S.K., Dulina N.G. Modeling the structure of interoperability by means of structural consistency. Informatics and Applications 2023;17(1):57-65. DOI 10.14357/19922264230108. (in Russ.)

2. GOST R 55062-2012. Information technologies. Industrial automation systems and integration. Interoperability. Basic principles. Moscow: Standartinform; 2014. (in Russ.)

3. Ullberg J., Johnson P., Buschle M. A Modeling language for interoperability assessments. Lecture Notes in Business Information Processing 2011;76:61-74. DOI: 10.1007/978-3-642-19680-5-7.

4. Makarenko S.I., Solovieva O.S. [Key provisions of the concept of semantic interoperability of networkcentric systems]. Journal of radio electronics 2021;4. DOI:10.30898/1684-1719.2021.4.10. (in Russ.)

5. Heider F. The psychology of interpersonal relation. New York; 1982. DOI: 10.30898/1684-1719.2021.6.3.

6. Dulin S.K., Dulina N.G., Kozhunova O.S. Synthesis of geodata in spatial infrastructures based on related data. Informatics and Applications 2019;13(1):82-90. DOI: 10.14357/19922264190112. (in Russ.)

7. Xiaojin Z., Ghahramani Z. Learning from labeled and unlabeled data with label propagation. Technical Report CMU-CALD-02–107, Carnegie Mellon University; 2002. (accessed 01.04.2024). Available at: https://mlg.eng.cam.ac.uk/zoubin/papers/CMU-CALD-02-107.pdf.