Open AccessUsing multivalued logic for qualitative data analysis
Author(s) -
L. A. Lyutikova
Publication year - 2021
Publication title -
journal of physics. conference series
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.21
H-Index - 85
eISSN - 1742-6596
pISSN - 1742-6588
DOI - 10.1088/1742-6596/2131/3/032046
Subject(s) - computer science , probabilistic logic , consistency (knowledge bases) , theoretical computer science , representation (politics) , logical data model , classifier (uml) , function (biology) , completeness (order theory) , data mining , algorithm , artificial intelligence , mathematics , data modeling , mathematical analysis , evolutionary biology , politics , political science , law , biology , database
The paper considers a logical approach to data analysis to solve the classification problem. The studied data is a set of objects and their features. As a rule, this is scattered heterogeneous information and it is not enough for a reasonable application of probabilistic models. Therefore, logical algorithms are considered, which under certain conditions may be more adequate. For an expressive formal representation of the relationship between objects and their attributes, multivalued logic is used, and the number of values depends on a specific attribute. Therefore, a system of operations is proposed for variables with different domains of definition. As a result, a decisive function is built, which is a classifier of objects that are present in the studied data. The properties and capabilities of this function are analyzed. It is shown that the logical function, which is a conjunction in the space of rules connecting given objects with their characteristic features, uniquely characterizes the initial data, divides the subject area into classes, possesses modifiability properties, meets the requirements of completeness and consistency in a given area. The paper also proposes an algorithm for its implementation.