A use case is a collection of scenario's.
A scenario is a time-sequenced set of interactions between actor and system.
Alternate postulate:
A use case is a set of interactions between actor and system that generate a (at least one) value outcome to the actor(s) (usually the primary actor but not necessarily).
A scenario is a statement of the (significant) possible outcomes of a use case.
The factors that usually influence a "batch interaction" (IMO) are the states of input information, both at a macro and detail level, for example:
a) input batch file is full/empty, exists/is missing etc
b) input data/record/object is in state x/y/z.
There are several ways to model these things. Which you use depends on "what the current problem is".
At the conceptual stage, I would be looking at the macro influences at the top level and investigating the details in lower level models such as activity models at the "per record" level. At the macro level, there are still (at least) two different techniques that come to mind.
First, the primary actor is the person, event or thing that invokes the job and the input data is, conceptually, an artifact whose composition is immaterial. Is this style we can investigate the normal, empty file, missing file etc etc scenarios.
Secondly, you can view the input data as the primary actor and thus design use cases that reflect the system outcomes depending on what the input data wants to happen to it. In other words, dont consider an equivalence between the batch job and the use case (design before analysis?).
An example may help. Consider a batch input file for a bank clearing process. An input transaction could "debit customer account" or "credit customer account" or even (if they are significant possibilities) "create customer account" or "close customer account". Wham, 4 use cases and a primary actor ("input file transaction").
Which technique is betterer? Depends on what the problem is.
hth
bruce
