Інтелектуальний метод формування людських ресурсів на короткостроковий проект

Abstract

Штучний інтелект, хмарні обчислення, підключені пристрої, автономна мобільність та інші цифрові технології вже змінюють всі галузі суспільства. Це вимагає нового типу управління проектами, який спирається на гнучкі команди між підрозділами та потребує консультантів з навичками, які перевершують просте управління процесами. У статті проведено ґрунтовний аналіз наукових досліджень у сфері використання методів машинного навчання та управління людськими ресурсами. Розроблено інтелектуальний метод формування короткострокового HR-проекту на основі машинного навчання, що дозволяє скоротити витрати часу на HR і, відповідно, знизить витрати рекрутерів. Представлено розумний чат-бот, який проводить співбесіду з потенційним членом команди. Чат-бот виконує роль HR, який після співбесіди передає дані в базу даних по кожному претенденту. На основі отриманих даних чат-бот виводить найоптимальніший результат щодо розподілу ролей у команді серед найкращих претендентів.

Statement of the problem. With the amount of information that is available in the recruitment process, machine learning can recognize many more patterns than a recruiter and apply effective methods for identifying candidates. When combining machine learning with a chatbot, it will reduce routine work: resume selection, calls to applicants, answers to typical questions, interview invitations, etc. In this regard, it can be assumed that the study of ways to form an HR short-term project based on machine learning is one of the promising areas in the management of short-term projects. Analysis of recent research and publications. It should be noted that mostly the authors analyze the work of chatbots or the selection of team members based on machine learning methods. However, the combination of an intelligent chatbot with the generation of a response based on a detailed analysis of applicants and, accordingly, making goal-oriented management decisions in the formation of a team for a short-term project is an important point. The analysis of candidates can be expanded with the use of resume parsing from job search sites, which will allow to maximize the base of applicants. The purpose of the article is to develop an intelligent method of forming a short-term HR project based on machine learning, which allows to reduce the time spent on HR and, accordingly, reduce the cost of recruiters. Presenting main material. In the work, the HR management strategy of a short-term IT project, which highlights the main stages of team formation and work with it, has been formed. To select applicants for a short-term IT project, a method for forming an HR of a short-term project based on the combination of an intelligent chatbot, parsing of job search sites and analysis of applicants using machine learning has been developed. This makes it possible, unlike analogues, to form a closed virtual environment for selecting applicants for a short-term project team without an HR manager. The advantages of this are the reduction of time spent on HR management and, accordingly, the reduction of recruiters' labor costs. Results. Calculations have shown that the parameter of the average quadratic error mse = 9 at a value of 7.5 on the 7th sample, which corresponds to ID – 7. The accuracy of the prediction is 94%. From the results of the decision tree, it can be seen that the applicant with ID =7 has the best characteristics for the position of developer. This applicant has given answers closest to the ideal ones in the training sample. Then this result is sent to the manager and he makes the appropriate decision, which is sent through the chatbot to all applicants. Conclusion. The results of the experimental studies have confirmed the effectiveness of using a decision tree to select applicants for the position of developer, which is of practical importance in the formation of HR of a short-term project. The use of additional resume parser from job search sites has allowed to increase the sample of applicants, which in turn increased the accuracy of forecasting from 94% to 98%.