Investigation of Defects in Models of Functional Requirements Created by Students in Object-oriented Analysis

Abstract

Assessment of the degree of level of IT specialists professional competencies when performing tasks in practice-oriented disciplines is an actual problem. Defects in functional requirements models created by students in object-oriented analysis affect the rest of the information system modeling stages. The aim is to study defects in UCDs. The tasks are study of the distribution of the number of types of defects and their number in UCD with a low level of detail, created by IT-students at the stage of object-oriented analysis; study of the relationship between defects of various types over several academic years in order to detect the influence of pedagogical influence on the density of defects. Statistical analysis methods using relationship scales are applied. As a result of the study of a corpus of 247 models for the period from 2016–2017 to 2021–2022 academic year, 2585 defects were identified. Their classification is used: lexical, syntactic and semantic. Types of defects are identified in each class. The study used indicators: the number of defects types and the number of these defects. As a result, it was revealed that the distributions of UCD defects indicators are not normal distributions. The use of a teaching aid with examples of creating UCD in the educational process, starting from the 2019–2020 academic year, has led to a steady decrease in the number of defects. This is confirmed by a statistically significant 95 % difference in indicators for a pair of academic years 2018–2019 and 2019–2020 according to the Wilcoxon–Mann–Whitney criterion. A positive aver-age relationship was found between the number of defects in the lexical and semantic classes in 2016–2017, 2017–2018 and 2021–2022 academic years. A positive strong relationship was found in the class of lexical defects be-tween types that refer to the attributes of one element of the model. In all other cases, no connection has been iden-tified that can be taken into account. The approach used can be applied to analyze other types of student UML models.