Defining and using road network data in an accident database

This thesis proposes improvements to the design of road
accident databases typically used by local authorities in
England. The present design tends to lead to inconsistencies in the information relating to the road network contained in the database.
A methodology for the redesign of the database is proposed which will lead to greater data integrity and provide additional and more detailed information. The advantages of the system are demonstrated by producing accident predictive relationships for sharp bends and minor junctions.
The design has been carried out in the context of a relational database system incorporating data from a geographical information system. The advantages of an object-oriented system are also considered and proposed as
a direction for further research.