One of the major concerns of AI researchers and implementers is how to ensure that the systems stay aligned with the aspirations of the humans they interact with. This problem becomes even more complex for systems that develop their own operational rules and where multiple agents are involved. The paper addresses some of the implications of using genetic/epigenetic design techniques where the control structure is developed without direct human involvement. This presents particular difficulties in ensuring that the control protocols stay aligned with the desires of the instigators and do not cause unpredicted harm. It also explores how this problem is further complicated when the AI system has many agents. Modern control systems are often decentralized which provides a more robust solution than using a central controller. A specific example of this approach is Self-Organising Swarms where the agents act independently of the central control. From an alignment point of view, it generates particular problems. Not only must the individual agents act in the best human interest but the swarm as a collective must do it as well. This is difficult for a homogeneous swarm and no proposal for a heterogeneous one has yet been made. There have been and continue to be considerable research and discussions on how to create and what form a global AI ethics might take, but any progress has been slow. This is partly because even the Universal Declaration of Human Rights has difficulties. All the nations that have signed up to the UN Human Rights Declaration believe they are at least trying to implement it. The problem is in the interpretation where many signatories believe others are in breach. The same would apply to any universal AI ethics agreement. This paper proposes a solution where the AI systems’ basic ethics are individual but have to comply where they interface with either other AI entities or humans. Figs.: 5. Refs.: 27 titles.
This paper relates to the developed Spatial Grasp (SG) Paradigm for solving complex problems in a holistic and fully distributed way. It presents the results of its further development in two interlinked directions: philosophical-conceptual and technological-implementational. In the first direction, there are discussed the details of how SG develops in distributed spaces as waves or even viruses and how grasps at the same time solutions of spatial problems, also how it fundamentally differs from traditional representations of systems and their solutions as parts exchanging messages. The SG philosophy also resembles higher concepts like perception, awareness, consciousness, and even soul. In the other direction, the resultant Spatial Grasp Technology details are briefed where its Spatial Grasp Language (SGL) interpreters can be networked as spatial computers covering any terrestrial and celestial environments. Distributed interpretation mechanisms of basic SGL constructs are discussed in detail, allowing for the implementation of spatial functionality without centralized resources. The paper also provides examples of fully distributed SGL solutions for observing and evaluating very large phenomena, such as hurricanes, forest fires, even galaxies, as well as discovering images in distributed networks, which can be perceived as a whole in line with the conceptual orientation of SG. The developed paradigm allows direct expression of top semantics and holistic methods for solving complex problems, dynamic creation and composition of the needed implementation environment, thus providing the strictest way from problem definition to a practical solution. Figs.: 17. Refs.: 47 titles.
The complexity of the tasks of making decisions in organizational management is stipulated by the extremely large dimension of the management space, the weak structure of the tasks being solved due to their uniqueness, the uncertainty of conditions, and the changeability of the goals (aspects) of management. The ontological approach provides a range of advantages in solving the problem of organizational management automatization. The use of the activity ontology makes it possible to significantly expand the scope of formalized methods application to organizational multi-aspect management. Due to the ontology, the detailing of managerial alternatives is increased, and the factors of the performers’ activity and the multi-aspect nature of management are taken into account. The article considers the interpretation of the process of forming managerial alternatives as a process of the threefold grouping of elements of the activity ontology and discrete moments of time. The method of distributed formation of an ontological network with the grouping of its elements using expert methods, implemented in the practice of designing corporate ACS, is described. The topical scientific tasks of the development of the scientific and methodological apparatus of organizational management based on ontology are formulated. These tasks include the formation of decision-making spaces that are invariant to aspects of management, the concretization of characteristics-invariants and reconfiguration of procedural regulations concerning the actual aspect of management, synthesis of alternatives based on the ontology elements. Some approaches to solving the formulated tasks are shown. The formulation of the problem of synthesizing a system of procedures with given input and output parameters is presented as a problem of tensor transformation of an initial multi-coil electrical network (library of procedures) into a network with the given parameters. A tensor equation is proposed. Its solution allows determining the values of voltages on the intermediate coils of the original electrical network which forms the structural connections for the synthesized procedural regulations. Figs.: 11. Refs.: 40 titles.
The intensive development of information technologies predetermined the increased interest of society in the problems of information and information security and led to a sharp increase in the industry-wide scientific activity in this area. Numerous scientific works in the subject area have become the basis for the development of theoretical foundations for the regulation of public relations in the information sphere. At the same time, the listed works are of an exclusively civil law nature. The issues of regulating relations regarding commercial secrets are either not considered at all or are considered fragmentarily. A comprehensive study on the multi-criteria assessment of the levels of information confidentiality to be included in the document has not previously been undertaken. The article discusses an approach to assessing the level of information confidentiality to be included in a future document. This approach is based on the application of expert judgments regarding the degree of influence of confidentiality features on the final level of confidentiality. The fuzzy inference mechanism, which constitutes the analytical essence of the approach, solves the task, relying on a direct cause-effect relation between the signs of the information confidentiality and the levels of the confidentiality of the documents. Earlier, because of the application of the fuzzy inference system, it was possible to form a reasonable scale of gradation of the levels of the confidentiality of the documents and it is quite easy to obtain final estimates of the levels of the information confidentiality to be included in a future document. In the current article, a similar problem is solved, but with the use of another method of fuzzification of confidentiality features as qualitative assessment criteria. Tabl.: 2. Fig.: 1. Refs.: 8 titles.
To monitor and maintain the proper functioning of distributed systems, there are used log events to further analyze and identify the circumstances and causes of deviations from normal operation. Tools to achieve this goal should primarily provide: 1) rapid search and processing of log files in the process of analysis and identification of failure causes; 2) generation of offers for solving the identified problem. This article presents the results of the analysis of the existing tools for logging and log file processing and substantiates the feasibility of the development of the proposed diagnostic system. The general structure of the developed diagnostic system and its functional model are described. In the paper, there are considered the functions of the main software modules: a parser that processes a log file and collects parts of messages into a single log object; a diagnostic module that processes information about an error in a log file object; and a module for accumulating and using knowledge about the known node malfunctions, ways to fix them and problems that were not previously identified. The choice of MongoDB NoSQL for saving the log file data and knowledge base storageDBMS is substantiated. A unified structure for log file objects has been developed for the MongoDB environment. Diagnostics system control – interoperability of modules and their interaction with the database and knowledge base in the MongoDB environment – is provided by the main program LogHelper which allows the user to perform different queries on log objects, monitor their results, inspect known and not previously identified errors, work with unknown malfunctions, manually set their statuses and find a solution from an existing set of solutions. The results of the conducted research show that the developed diagnostic system is highly effective as a maintenance solution for remote data processing nodes and it supports rapid identification of the malfunction and generation of proposed solutions.Figs.: 6. Refs.: 8 titles.
The article examines the features of algorithmization of processes in recommendation systems for personnel training and decision support. Such systems are a subclass of information filtering systems that allow building a certain rating based on requests or preferences. The development of recommendation systems is aimed precisely at finding the best solution within a limited period of time. In-depth analysis of complex problems is necessary to develop several alternatives that do differ significantly, including the possibility of inactivity. This is why each alternative should be evaluated. The paper reveals the features of recommendation systems in various fields of application. The conducted research allows us to conclude that there are some significant differences in the nature of providing recommendations to the end user when using distance learning or decision support systems. Recommendations in such systems are shifted from the user's preferences directly to the person and the problem. When structuring the processes of the recommendation system to support decision-making, there is made a focus on the scientific method of substantiating questions. Therefore, the construction of algorithms for such systems is complicated from the first stage – observation, where obstacles or opportunities are identified. Such a problem can be solved through systems of equations that describe processes in an organization with appropriate constraints. Presented in the article algorithm for evaluating and selecting alternatives is one of the options for building a system of recommendations if it is necessary to justify a management decision to expand activities of the company. It is shown that although they lie in the same plane, the solution to remove the problem and the search for an optimal solution are not equivalent. In addition, finding the best solution requires a larger database of information to create a recommendation system. Figs.: 2. Refs.: 12 titles.
Modern automated control systems can not do without the special means of organizing communication with humans. The main peculiarity of human-machine interaction is the combination of formalized information processes and structures with non-formalized (as a rule, creative) information processing. The latter helps not only find ways to solve complex problems that can not be found in formal models (algorithms) but also organize the process of problem solving as a holistic system. Human-machine interaction can not be fully algorithmic (in the formal sense), but it is possible to develop technology for mutual activities of humans and technical systems for problem solving. The role of people in technical control systems is constantly changing. Previously, a human was the link that was mostly responsible for process management. With the development of automated systems that was supported by the ever-increasing use of computers, people are increasingly taking over the functions of the control unit, at the same time keeping the role of an observer of the process. However, in recent years the role of humans in control systems has changed. In modern technical systems, they again play a leading role, which fully reveals their abilities and uses its boundary capabilities to optimally distribute the functions between human and machine, human and process. Human-machine interaction is a form of mutual use of reserves of humans and automated systems that is ultimately focused on humans. The article considers the stages of human-machine automation systems development, the composition and distribution of functions in human-machine systems, the possibility of their implementation, emergencies in automated technological facilities, and some examples of human-machine control in ACS TP thick-sheet rolling mills. Таbl.: 2. Figs.: 2. Refs.: 16 titles.
In the current work, there has been developed for the first time a geo-information web-system for presenting the results of the inventory of air pollution emissions in Kyiv from industrial enterprises on the basis of the processing of data from the State Statistics Service of Ukraine, as well as estimates of geo-distributed emission fields from vehicles. The system contains information on emissions of 21 pollutants, including those that are regularly monitored in Kyiv by governmental agencies. The system is available via the link http://env.kiev.ua:8080/pollutionsystem/ pollutionsystem.html. Its main features – viewing on the map the locations of industrial emission sources for the selected substance, and the corresponding amount of emissions – can be used even by unregistered users. The developed system can be used in combination with the existing public and state systems for monitoring of air pollution in Kyiv for preliminary (screening) analysis of possible causes of high levels of air pollution during their observing with the monitoring systems. The paper presents a relevant example of the system application to explain the high levels of hydrogen sulfide concentrations at one of the stations in Kyiv. Another purpose of the system may be to detect shortcomings in the existing inventory of industrial emissions, namely the lack of information on certain sources, as well as verification and adjustment of data on emissions from the registered sources. To fully use the system potential, it is necessary to expand its functionality in the direction of mathematical modeling of the spread of pollution due to emissions from industrial enterprises and vehicles. For this purpose, it is also necessary to integrate hydrometeorological parameters and levels of air pollution in Kyiv into the data monitoring system.Figs.: 6. Refs.: 10 titles.
QUALITY, RELIABILITY AND CERTIFICATION OF COMPUTER TECHNIQUE AND SOFTWARE
The article is dedicated to the development of an attributive model of the dependable software, defining the metrics for dependable software, and the development of an object-independent universal approach to attributes assessment and dependable software metrics. The attributive model of the dependable software (AMDS) is designed by analogy with the attributive model for dependable computer systems (CS) and expanded by special attributes and metrics of CS software for critical purposes. Different areas of the national economy have developed their own industry standards for the creation of safe software which can be expanded and refined with the proposed AMDS and guided when defining attributes, metrics, and criteria for software assessment. In the AMDS, there are defined the following basic attributes: functionality, dependability, usability, efficiency, maintainability, portability, redundancy, self-control. An objectively-independent universal approach to the assessment of the software dependability by certain attributes, metrics, and criteria has been developed. A comprehensive assessment of the software dependability level is carried out according to AMDS in the form of linear functionality which contains assessments for all attributes of the model. There has also been developed a procedure for comparative assessment of the software dependability level using the analysis of the offered AMDS which is presented in the form of a generalized software dependability level assessment by which a comparative analysis of the dependability level of software various kinds is carried out. The attributive model of the dependable software, comprehensive and comparative assessment of the dependability level can be used by reliability specialists in various fields of application.Таbl.: 8. Refs.: 10 titles.
