ON-LINE DATABASE of SCIENTIFIC PAPERS
10.5593/sgemsocial2017/15/S05.047

EQUATIONLESS QUALITATIVE MODELS OF BUSINESS INTELLIGENCE TASKS BASED ON COMMON SENSE HEURISTICS

J. Kriz, M. Dohnal
Thursday 28 September 2017 by Libadmin2017

References: 4th International Multidisciplinary Scientific Conference on Social Sciences and Arts SGEM 2017, www.sgemsocial.org, SGEM2017 Conference Proceedings, ISBN 978-619-7408-17-1 / ISSN 2367-5659, 24 - 30 August, 2017, Book 1, Vol 5, 371-384 pp, DOI: 10.5593/sgemsocial2017/15/S05.047

ABSTRACT

There is a broad spectrum of BI (Business Intelligence) models. CBIM (Complex Business Intelligence Model) integrate several BI aspects, e.g. artificial intelligence, data bank, psychology, sociology etc. They are unique, partially subjective, inconsistent, vague and multidimensional. CBRMs development suffers from IS (Information Shortage). IS often eliminates straightforward application of traditional statistical methods. It is therefore often prohibitively difficult to analyse them using numerical quantifiers. Oversimplified or highly specific CBRMs are sometimes obtained. Their practical applicability is therefore (very) limited. AI (Artificial Intelligence) has developed a number of tools to solve such problems. Qualitative reasoning is one of them. It is based on the least information intensive quantifiers i.e. trends. There are just three trend / qualitative values used to quantify variables and their derivatives: plus/increasing; zero/constant; negative/decreasing. There are qualitative BI knowledge items in equationless forms such as heuristics. For example – if the investments into BI are increasing then the costs of BI tasks are decreasing. Such verbal knowledge item cannot be incorporated into a traditional numerical model. A qualitative model can be developed under conditions when the relevant quantitative BI model must be heavily simplified. The key information input into CBRMs is expert knowledge. The case study presents a model based on integration of 6 equationless relations using 5 variables e.g. Degrees of data consistence and data completeness. The result is represented by 11 scenarios. The paper is self-contained, no a prior knowledge of qualitative models is required.

Keywords: Business Intelligence, Qualitative Model; Qualitative Reasoning; Scenario; Transitional graph