The synonymic relevance degree estimation of the text answer in modern information systems

Abstract

Introduction of progressive forms of training and development of modern information technologies creates the necessity of automated assessment of the student’s knowledge. Great value for automated educational systems have models of assessing answers not in the form of chosen variants, but in the form of a free text of any length with synonym concept estimation. A special urgency gets the problem of development of answers’ analysis model on the task of the open type, demanding to enter from the keyboard the certain formulation of this or that term of a subject domain. There is an objective necessity of transition to computer testing of students’ knowledge. Thus, on the foreground rises the problem of automatic assessment of students’ knowledge. This problem is simple enough, if the student is offered to choose one or more right answers from a set of variants, but it becomes considerably difficult, if the procedure of testing provides a developed answer in any form, that is with his or her own words in natural language. In the latter case it is possible to appreciate the student’s knowledge only by comparative text analysis of the answer with the set standard reference text and to assess their relevance. Thus, all word-forms, terms of the subject domain and grammatical structures of the statement should be considered and assessed with the use of all possible synonyms. At the automated knowledge control of the terminology of a subject domain the try to solve a task to compare two definitions of one term: the definition given by the teacher (reference definition), and the definition given by a trainee. The result of the comparison should be conformity with these definitions. The development of this method of standard reference definition and answer analysis is the purpose of the given artice.