Initial Draft UML Models Posted for Review

Notify me when a new comment is posted

Comments:

Comments on UML model for Crash Records
By Mohamadreza Banihashemi on Tuesday, March 15, 2005 (EST)

I am a transportation research engineer and I am part of a team (Geometric Design Lab, GDL) that coordinates the development of the IHSDM software (a software in highway safety sponsored by the FHWA). My involvement with XML data has started with defining the relations between the data used in the IHSDM and the LandXML data. The following are my comments regarding the “Crash Records” UML model. I am not familiar with UML much so, please forgive me if some comments of mine were invalid because of my misunderstanding of the model.

1.      It seems un-clear for the cases in which there were more than one vehicle involved in the crash, how the people (Non Motorist, Person, Driver, and Occupant) are related to the appropriate vehicle.

2.      The “numberLanesMajorRoad” and “numberLanesMinorRoad” attributes in “CrashSiteDetail / IntersectionCrashSite” might not be enough for describing the situation. First of all there might be more than two roads intersecting at the intersection. Secondly, the number of lanes on the same road but on different sides of the intersection might not be the same (i.e., if there are turning lanes, on the same road on one side of the intersection there are more lanes than on the other side). And finally, marking the intersecting roads as major and minor might not always work (it might be more practical to define name/numbers for different legs of the intersection and associate the vehicles to these legs).

3.      Most of the attributes of the “VehicleDetail / VehicleCrashSite” are repeated in the “CrashSiteDetail / RoadCrashSite.” First of all this creates un-necessary duplication. Secondly, if there were more than one vehicle and for different vehicles the values for one or more items are different it creates sort of discrepancy of which vehicle’s data be put in the crash’s data element. One solution might be that the attributes in the “CrashSiteDetail / RoadCrashSite” that are vehicle specific appear only in the “VehicleDetail.”

4.      The “Crash / Injury / sourceEMSTransport” attribute has two values as “other.”

5.      The “PeopleDetail / Non-Motorits / safetyEquipmentUsed” attribute has two values as “helmit.”

6.      In “CrashSiteDetail / IntersectionCrashSite / numberLanesMajorRoad” and

“CrashSiteDetail / IntersectionCrashSite / numberLanesMinorRoad” there are values of “four”, “five”, and “six” as well as “fourToSix.” “fourToSix” seems redundant here.

7.      I suggest combining the “PeopleDetail / DriverLicenseJurisdiction” and ““PeopleDetail / DriverLicense” (or actually moving the items of the first to the second and removing the first data element).

8.      I suggest adding “other” and “unknown” values to the list of vehicle makes in the “VehicleDetal / vehicle / vehicleMake.”

9.      I suggest adding something to the “Crash” data to distinguish the “workday” vs. “holiday.”

10. I suggest for multi-lane highways, adding something to the “Crash” data to show the lane on which the crash has occurred.

11. I suggest adding something (either to the “CrashSiteDetail” or to the “VehicleDetail”) to show the origin and destination legs of vehicles involved in the crash.

12. I suggest adding “Suburban” values to the functional classes in “CrashSiteDetail / functionalClass”. Suburban areas are more and more distinguished from urban and rural in the highway safety research.

13. I suggest showing in the “CrashSiteDetail / RoadCrashSite” whether there is a two-way left-turn lane.

14. I suggest showing in the “CrashSiteDetail / IntersectionCrashSite” whether there are left and right turning lanes, and how many of each type.

Reply to this Comment

Add Your Comment