Training set augmentation in training neural- network language model for ontology population

This paper is a continuation of the research focused on solving the problem of ontology population using training on an automatically generated training set and the subsequent use of a neural-network language model for analyzing texts in order to discover new concepts to add to the ontology. The article is devoted to the text data augmentation - increasing the size of the training set by modification of its samples. Along with this, a solution to the problem of clarifying concepts (i.e. adjusting their boundaries in sentences), which were found during the automatic formation of the training set, is considered. A brief overview of existing approaches to text data augmentation, as well as approaches to extracting so-called nested named entities (nested NER), is presented. A procedure is proposed for clarifying the boundaries of the discovered concepts of the training set and its augmentation for subsequent training a neural-network language model in order to identify new concepts of ontology in the domain texts. The results of the experimental evaluation of the trained model and the main directions of further research are considered.

A HYBRID ALGORITHM FOR FORMING THE SHORTEST TRAJECTORY BASED ON THE APPLICATION OF MULTI-AGENT LEARNING WITH REINFORCEMENT, THE SEARCH ALGORITHM A* AND EXCHANGE OF EXPERIENCE

The article presents a hybrid algorithm for the formation of the shortest trajectory for intelligent agents of a multi-agent system, based on the synthesis of methods of the reinforcement learning paradigm, the heuristic search algorithm A*, which has the functions of exchange of experience, as well as the automatic formation of subgroups of agents based on their visibility areas. The experimental evaluation of the developed algorithm was carried out by simulating the task of finding the target state in the maze in the Microsoft Unity environment. The results of the experiment showed that the use of the developed hybrid algorithm made it possible to reduce the time for solving the problem by an average of 12.7 % in comparison with analogs. The differences between the proposed new “hybrid algorithm for the formation of the shortest trajectory based on the use of multi-agent reinforcement learning, search algorithm A* and exchange of experience” from analogs are as follows: – application of the algorithm for the formation of subgroups of subordinate agents based on the “scope” of the leader agent for the implementation of a multi-level hierarchical system for managing a group of agents; – combining the principles of reinforcement learning and the search algorithm A*.

LOOSENING AND LEVELING DEVICE FOR PREPARING SOIL FOR MELON CROPS

The analysis of soil cultivation technologies for sowing melon crops was carried out. The design of a combined soil tillage tool capable of plowing, pre-sowing treatment and formation of irrigation furrows in one pass was substantiated. The main tillage is recommended to be done by front plow tools for smooth plowing. Plow bodies of two bottom plows should be mounted along the symmetry axis of the implement according to the lister scheme, which allows not to carry out a full rotation of soil layers and provides automatic formation of irrigation furrow. A loosening and leveling device for strip pre-sowing soil tillage in the sowing zone has been developed. The use of a combined soil tillage tool can reduce labor costs up to 25%, energy consumption for soil preparation up to 50%, reduce the duration of work, reduce soil compaction and retain moisture in the soil layer.

Derivation of complex judgments as a conceptual tool thinking intelligent mobile systems

Обоснована целесообразность разработки инструментальных средств автоматического формирования суждений как одного из мыслительных актов понятийного мышления интеллектуальных мобильных систем различного назначения. Применение таких средств вывода суждений позволяет сформировать недостающие для принятия решений знания в процессе планирования целенаправленного поведения в априори неописанных условиях проблемной среды. Для решения данной проблемы в качестве исходных элементов представления знаний использованы условно – зависимые предикаты обеспечивающие возможность вывода как простых, так и сложных суждений определяющих различные закономерности целенаправленного преобразования текущих условий функционирования. Разработаны инструменты позволяющие формировать сложные суждения из простых суждений, полученных путем означивания переменных сорта «объекты» и «отношения» условно зависимых предикатов объектами, находящимися в проблемной среде и оценками отношений, которые наблюдаются в ней между данными объектами. В качестве примера показывающего эффективность использования предложенных инструментальных средств получения новых недостающих для вывода решений знаний, построен алгоритм планирования целенаправленного поведения автономного мобильного подводного робота обеспечивающий ему возможность выполнять достаточно сложные задания, связанные с поиском различных объектов, обладающих определенными свойствами и их перевозкой в заданную точку проблемной среды. The expediency of developing tools for automatic formation of judgments as one of the mental acts of conceptual thinking of intelligent mobile systems for various purposes has been substantiated. The use of such means of inference makes it possible to form the knowledge that is missing for decision-making in the process of planning purposeful behavior in a priori undescribed conditions of a problematic environment. To solve this problem, conditionally dependent predicates were used as the initial elements of knowledge representation, providing the ability to derive both simple and complex judgments that determine various patterns of purposeful transformation of the current conditions of functioning. Tools have been developed that allow to form complex judgments from simple judgments obtained by designating variables of the sort "objects" and "relations" of conditionally dependent predicates by objects located in a problematic environment and estimates of relations that are observed in it between these objects. As an example showing the effectiveness of using the proposed tools for obtaining new knowledge that is missing for the conclusion of solutions, an algorithm for planning the purposeful behavior of an autonomous mobile underwater robot has been built, which provides it with the ability to perform rather complex tasks related to the search for various objects with certain properties and their transportation to a given point of the problem Wednesday.

A gearing forest automaton to automate a gearing system

A gearing system design completely depends on the designer’s skills and experiences because there is no theory or method to standardize the design process. In this paper, a gearing forest automaton is presented to implement automatic formation of design schemes from the original data without any human intervention. Firstly, a gearing tree is defined to describe a gearing system, and a two-shaft connection by M-transmission shafts (M ≥ 0) and an N-shaft connection by a transmission shaft (N ≥ 3), which can iteratively be combined into any gearing system, are denoted with data structure and automatically achieved. Secondly, the connection methods are both the input alphabet and transition functions of the gearing forest automaton. The start state comprises the original data and the accept state outputs the design schemes. Therefore, an entire design process becomes the construction and running of the automaton. The automaton is optimized to increase efficiency. Finally, according to this automaton, a computer program has been developed, and it is validated through three examples in which design quality is significantly improved.

Computer Verification of Descriptive Geometry Task Solutions for Engineering and Graphic Education

In this paper is formulated the relevance of computer tools creation for verification of descriptive geometry task solutions. Are analyzed the shortcomings of available methods and systems for such verification. A new verification method is proposed – the mode of superposition based on overlaying a student’s solution with a template and formal evaluation the overlaying results. To create templates for a wide range of descriptive geometry tasks, it is proposed to use a formal grammar of the correct solution, which is constructed using special nonterminal symbols – “and”, “or”, “transform” and “instance”. As the grammar’s terminal symbols are used geometric figures. Thus, the template consists of a graphic part (a set of figures) and a structural description (grammar). An implementation of this verification method is demonstrated as a software system for verifying of descriptive geometry task solutions in the form of DXF-files. A functional model of the verification system is given. The automatic formation of a template from a graphical model, which is created in a vector graphics editor and does not require a symbolic description, is considered, as well as processing procedure for a student’s solution, during which the verifiable model goes through phases of normalization, filtration, and extracting of higher-level elements. An example of checking for two solutions (the correct one and containing errors) of the task for constructing a perpendicular to a plane of general position is given. The work of a subsystem for verification result visualization is demonstrated too. The created system can be implemented in Internet-libraries of tasks, or in distance learning systems, and can be used for remote support of geometric-graphic courses. Conclusions about feasibility of introducing the proposed method as a tool in CAD-systems are made.

The innovation of accounting and certification of historic architectural monuments using BIM technology

Presented research shows Building Information Modeling (BIM) technology as a highly promising tool for architectural heritage conservation. Analyses of contemporary publications by BIM topic and HBIM (Historic Building Information Modeling), its specific division, are conducted. The usage of modern technology analysis for digitizing monuments along with the outdated ‘manual’ of accounting and passportization methods are executed, and a lack of automation of the above processes is observed. Additionally, the research reveals that during the operational period of historical objects, there was no method of fixing their architectural transformations. The necessity of automation of the abovementioned processes by the capabilities of the BIM technology is emphasized. Automation methods developed utilizing the BIM model properties as a relational database, are described. An example of the method implementation in architectural and construction software applications is given. The case uses the rules of filling in the state accounting card of Ukraine. Nonetheless, the automatic formation of a document like this by the standards of any country is underscored.