Premium
An Approach to Ordinal Classification Problems
Author(s) -
Larichev O.I.,
Moshkovich H.M.
Publication year - 1994
Publication title -
international transactions in operational research
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.032
H-Index - 52
eISSN - 1475-3995
pISSN - 0969-6016
DOI - 10.1111/1475-3995.d01-48
Subject(s) - consistency (knowledge bases) , ordinal scale , task (project management) , computer science , decision maker , artificial intelligence , ordinal data , basis (linear algebra) , machine learning , ordinal regression , natural language processing , data mining , mathematics , operations research , statistics , geometry , management , economics
An ordinal classification task is defined. An approach to the construction of full ordinal classification on the basis of a decision maker's knowledge is proposed. It allows elicitation of information (or knowledge) in a natural form for the decision maker (through qualitative attribute scales and verbal descriptions of decision classes). It provides verification of the received judgments for consistency and possibilities for corrections and modifications of the elicited classification rules. Problems with obtaining valid judgments from people in ordinal classification tasks are discussed. The decision support system ORCLASS, developed on the of the proposed approach, is described.