Класифікаційна система з підбору персоналу
Loading...
Date
2019
Authors
Жежерун, Олександр
Рєпкін, Максим
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
У статті представлено класифікаційну систему, побудовану у вигляді бази знань, яка допомагає розв’язати задачу класифікації під час підбору персоналу. Ця система порівнюється із системами на основі нейронних мереж. Проведено огляд засобів побудови онтологій, доречних для застосування у подібних системах. Описано функціонал та реалізацію класифікаційної системи на основі фреймворку Protégé, що співпрацює з Telegram-ботом і дає змогу ефективно відбирати кандидатів, які найбільше відповідають вимогам вакансії.
The main goal of classification is to split a plurality of elements into categories so that these elements have a sufficient number of identical features, which allows to ignore their individual differences. There are classification systems based on neural networks, which use the method of "supervised learning". Such systems are effective and popular, and they are usually used on big amounts of data. The work of such a network is more effective than the knowledge base, because a well-trained network calculates results faster. Nevertheless, neural networks have a problem of necessity to retrain the network on the structure change. Ontology-based classification systems solve this problem and allow to work with it right away due to semantic reasoning. Such systems are also used when it is impossible to get the data-set for the neural-network training. Due to these reasons the ontology-based classification system is more suitable for resolving the described problem. Protégé is the most popular ontology editor which is able to export ontologies in various formats (RDF/ XML, OWL/XML, JSON-LD) and provide a number of semantic reasoners (Pellet, HermiT, FaCT). Telegram is one of the most secure and popular messengers. Moreover, Telegram bots have a wide functionality. Taking into consideration these facts, it was decided to implement the system which helps company managers to select the best candidates for their vacancies, according to the required skills and the skills of a candidate. The system was implemented by using the ontology exported from the Protégé in OWL/XML format to define the hierarchy, semantic reasoner HermiT to classify data, and Python programming language to implement the bot’s functionality. A description of the structure, functionality, test-case, and the main problems of the system is given in the article.
The main goal of classification is to split a plurality of elements into categories so that these elements have a sufficient number of identical features, which allows to ignore their individual differences. There are classification systems based on neural networks, which use the method of "supervised learning". Such systems are effective and popular, and they are usually used on big amounts of data. The work of such a network is more effective than the knowledge base, because a well-trained network calculates results faster. Nevertheless, neural networks have a problem of necessity to retrain the network on the structure change. Ontology-based classification systems solve this problem and allow to work with it right away due to semantic reasoning. Such systems are also used when it is impossible to get the data-set for the neural-network training. Due to these reasons the ontology-based classification system is more suitable for resolving the described problem. Protégé is the most popular ontology editor which is able to export ontologies in various formats (RDF/ XML, OWL/XML, JSON-LD) and provide a number of semantic reasoners (Pellet, HermiT, FaCT). Telegram is one of the most secure and popular messengers. Moreover, Telegram bots have a wide functionality. Taking into consideration these facts, it was decided to implement the system which helps company managers to select the best candidates for their vacancies, according to the required skills and the skills of a candidate. The system was implemented by using the ontology exported from the Protégé in OWL/XML format to define the hierarchy, semantic reasoner HermiT to classify data, and Python programming language to implement the bot’s functionality. A description of the structure, functionality, test-case, and the main problems of the system is given in the article.
Description
Keywords
класифікаційна система, база знань, онтологія, Protégé, Telegram-бот, стаття, classification system, knowledge base, ontology, Protégé, Telegram-bot, article
Citation
Жежерун О. П. Класифікаційна система з підбору персоналу / Жежерун О. П., Рєпкін М. С. // Наукові записки НаУКМА. Комп'ютерні науки. - 2019. - Т. 2. - С. 38-42.