An expert system approach to LCI database management
Notten P, Weidema B P (2003)
Presentation to the International Workshop on LCI-Quality, Karlsruhe, 2003.10.20-21
Tools to check for errors and inconsistencies in LCI databases are currently lacking, as are systems to aid in the selection of LCI data. This is because the large and aggregated nature of LCI databases make them difficult to evaluate. However, if viewed as a multivariate problem, where the product and processes are interpreted as the independent variables and the environmental exchanges as the dependent variables, it is possible to take advantage of the many statistical and mathematical techniques developed to gain insight into multivariate systems. In this paper, an expert system combining data retrieval and multivariate data analysis techniques is proposed to enhance database management. Through the use of exploratory pattern recognition techniques, such as correlation analysis and principal component analysis (PCA), the expert system will be able to alert a user to unexpected differences between inventories of the same or similar processes, to identify redundant entries, and to warn of possible errors in the data or mistakes in the data entry process. The expert system can be applied both to a single database and to a network of databases, thus providing a means for integration of diverse data sources.