Automated interpretation of key performance indicator values and its application in education

Abstract

Business intelligence applications are supposed to provide accurate, on-time reports to business analysts and executives. However, when the latter are concerned, it seems that reports they receive just contain tabular and/or visualized data (charts, dials and other visualizations) and little or no information. No matter how well presented, data has no meaning unless it is interpreted - only then it becomes information and only then can it provide actionable insight (knowledge) into how business is doing. Since this interpretation is performed manually by users, numerous issues can affect the quality of derived information: data volume and presentation, user's concentration, previous knowledge, experience, bias, etc. A solution for automated interpretation of (business) key performance indicator values is presented in this paper. It uses domain knowledge and expert system technology to transform business data into information. Information is then presented as natural-language-like sentences and incorporated into reports. An implementation of this solution and its application as an educational tool are presented and evaluated.