ДЕЯКІ СПОСОБИ МОДЕЛЮВАННЯ ВХІДНИХ ДАНИХ ДЛЯ ПОШУКУ ІНФОРМАЦІЇ В БІБЛІОТЕЦІ ЕТАЛОНІВ ПРИ РОЗВ’ЯЗАННІ ЗАДАЧ СЕМАНТИКИ

• Дата: 2020
• Автор: Tymofijeva, N.К.
• Формат: Стаття
• Мова: English
• Опубліковано: V.M. Glushkov Institute of Cybernetics of NAS of Ukraine 2020
• Теми: задачі семантики; комбінаторна оптимізація; кластеризація; покриття заданими ознаками певного об’єкта; пошук еталону в базі даних
• Онлайн доступ: https://jais.net.ua/index.php/files/article/view/500
• Назва журналу: Problems of Control and Informatics

To determine the essence of the subject it is used standard information contained in databases which is a structured set of interconnected data of a specific subject area. To find quickly the information it needs, this database should be structured, input data should also be modeled accordingly. Today there are input data models that have their advantages and disadvantages, and each model has its own scope. The article provides examples of problems of semantics that relate to recognition problems. To determine the essence of a particular object, it requires finding its standard in the database. These are speech recognition, child, female, male voice recognition, the problem of clinical diagnostics, comparison of texts on plagiarism, automatic translation of texts from one language to another, etc. There are two ways of comparing the input information and the standard: by the primary signs that describe the object being sought and by the given object. In the second method previous cover by certain signs of the standard and of the object is not conducted. At modeling input data to search for information by the first way the cover of given objects by certain signs takes place. The signs are divided into those that characterize only the given object, by which it is quite simply to define it in the database. In this case the problem is solvable. If the same signs describe different objects, but using differential analysis you can find the the desired object, then this problem is partially solvable. If the same signs characterize different objects and the desired object cannot be identified, then a situation of uncertainty arises. There are problems that do not require a standard library for their recognition. In some recognition problems, which may be semantics problems, the input data is divided into segments, with subsequent determination of the similarity of the resulting parts. In this case, the input data contains both the object to be recognized and the standard with which it is compared. In some problems it is taken as a standard either an expression that determines the similarity of the input data and standard information, or the conditions are specified by which a given object can be recognized. The standard library is not used to solve these problems.