Quantitative assessment of the significance of recurrent patterns identified in accident descriptions
The central part of the major accident reporting system (MARS) database is a relational database management system, which allows the analysis of complex data sets consisting of data of several different object classes. Information on the accidents consists, to a large extent, of free texts, containing most of an accident report's relevant and objective information. Therefore, in order to identify significant dependencies among the accident descriptors and thus overall recurrent patterns in the circumstances of accident occurrences (lessons learned), a corresponding mathematical tool has been developed whose main functionalities to assure a high rate of automatic capturing of relevant information are described.
Bibliographic Reference: Paper presented: ESREL 98, Trondheim (NO), June 17-19, 1998
Availability: Available from (1) as Paper EN 41171 ORA
Record Number: 199810937 / Last updated on: 1998-09-01
Original language: en
Available languages: en